fix: resolve missing AgentTeam import and RouterAgent kwargs forwarding

- Fix missing AgentTeam import in self-reflection-optimization-patterns.py - Forward kwargs properly in RouterAgent._build_system_prompt to super() - Addresses critical issues identified by Gemini code reviewer Co-authored-by: Mervin Praison <[email protected]>
fix: resolve critical RouterAgent **kwargs bug and add missing if __name__ guards
2026-11-08 01:08:39 -10:00 · 2026-06-12 23:25:32 +00:00 · 2026-06-12 23:25:32 +00:00 · 2026-06-12 23:25:32 +00:00 · 2026-06-12 23:25:12 +00:00 · 2026-06-12 23:25:12 +00:00
645 changed files with 13441 additions and 83880 deletions
@@ -227,10 +227,9 @@ runs:
          2. Wrapper (praisonai/): CLI, heavy implementations, optional dependencies.
          3. Tools (PraisonAI-Tools): Specific integrations, external tools (e.g. SurrealDB, Slack), and community extensions MUST go to the `MervinPraison/PraisonAI-Tools` repository. If the feature is a tool, clone `https://github.com/MervinPraison/PraisonAI-Tools`, implement it there, and create the PR using `gh pr create -R MervinPraison/PraisonAI-Tools`.
        - Does it duplicate existing functionality? Check if Agent already supports this via existing params (reflection, planning, tools, hooks, memory).
-        - Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.
        - Does it inherit from Agent properly? New agent types MUST inherit Agent, not wrap it with composition.
        - Does it add new dependencies? Only optional deps allowed, must be lazy-imported.
-        - Will agent.py grow larger? If the change adds >50 lines to agent.py, find a way to extract instead. PRs with >100 lines added to agent.py require manual review (merge gate will not auto-merge).
+        - Will agent.py grow larger? If the change adds >50 lines to agent.py, find a way to extract instead.
        - Is there a name collision with existing exports in __init__.py?
        If ANY of these conceptual checks fail (excluding routing), add a comment to the issue explaining why and close it. Do NOT create a PR.

@@ -261,24 +260,16 @@ runs:
        ANTHROPIC_API_KEY: ${{ inputs.anthropic_api_key }}
        CLAUDE_CODE_OAUTH_TOKEN: ${{ inputs.claude_code_oauth_token }}
        GITHUB_TOKEN: ${{ inputs.github_token }}
-        INPUT_MODEL: ${{ inputs.model }}
        INPUT_MAX_TURNS: ${{ inputs.max_turns }}
        INPUT_ALLOWED_TOOLS: ${{ inputs.allowed_tools }}
        INPUT_DISALLOWED_TOOLS: ${{ inputs.disallowed_tools }}
        INPUT_TIMEOUT_MINUTES: ${{ inputs.timeout_minutes }}
-        CLAUDE_CODE_SUBAGENT_MODEL: inherit
-        ANTHROPIC_DEFAULT_SONNET_MODEL: claude-sonnet-4-6
      run: |
        echo "🚀 Running Claude Code with GitHub context..."
        
        # Build command arguments
        CMD_ARGS=("-p" "--verbose" "--output-format" "stream-json")
        
-        if [[ -n "$INPUT_MODEL" ]]; then
-          CMD_ARGS+=("--model" "$INPUT_MODEL")
-          export ANTHROPIC_MODEL="$INPUT_MODEL"
-        fi
-        
        # Add max turns if specified
        if [[ -n "$INPUT_MAX_TURNS" ]]; then
          CMD_ARGS+=("--max-turns" "$INPUT_MAX_TURNS")
@@ -289,36 +280,11 @@ runs:
        if [[ -n "$INPUT_ALLOWED_TOOLS" ]]; then
          TOOLS="$TOOLS,$INPUT_ALLOWED_TOOLS"
        fi
-        
-        # Remove any explicitly disallowed tools from the effective allow-list so
-        # the read-only contract is enforced deterministically (no reliance on CLI
-        # precedence between --allowedTools and --disallowedTools).
-        if [[ -n "$INPUT_DISALLOWED_TOOLS" ]]; then
-          DISALLOWED_CSV=$(echo "$INPUT_DISALLOWED_TOOLS" | tr ',\n' '  ' | xargs)
-          FILTERED=""
-          IFS=',' read -ra TOOL_ARR <<< "$TOOLS"
-          for tool in "${TOOL_ARR[@]}"; do
-            tool=$(echo "$tool" | xargs)
-            [[ -z "$tool" ]] && continue
-            skip="false"
-            for bad in $DISALLOWED_CSV; do
-              if [[ "$tool" == "$bad" ]]; then
-                skip="true"
-                break
-              fi
-            done
-            if [[ "$skip" == "false" ]]; then
-              FILTERED="${FILTERED:+$FILTERED,}$tool"
-            fi
-          done
-          TOOLS="$FILTERED"
-        fi
        CMD_ARGS+=("--allowedTools" "$TOOLS")
        
        # Add disallowed tools
        if [[ -n "$INPUT_DISALLOWED_TOOLS" ]]; then
-          DISALLOWED_FLAG=$(echo "$INPUT_DISALLOWED_TOOLS" | tr '\n' ',' | sed 's/,*$//')
-          CMD_ARGS+=("--disallowedTools" "$DISALLOWED_FLAG")
+          CMD_ARGS+=("--disallowedTools" "$INPUT_DISALLOWED_TOOLS")
        fi
        
        echo "📝 Executing Claude Code with prompt from file..."
@@ -1,34 +0,0 @@
-#!/usr/bin/env node
-const chain = require('./bot-pr-review-chain.js');
-
-let failed = 0;
-function assert(name, cond) {
-  if (!cond) {
-    console.error('FAIL:', name);
-    failed++;
-  } else {
-    console.log('ok:', name);
-  }
-}
-
-assert('chainKickPosted requires both markers', chain.chainKickPosted([
-  { user: { login: 'MervinPraison' }, body: '@coderabbitai review' },
-  { user: { login: 'MervinPraison' }, body: '/review' },
-]));
-assert('chainKickPosted false with only coderabbit', !chain.chainKickPosted([
-  { user: { login: 'MervinPraison' }, body: '@coderabbitai review' },
-]));
-assert('coderabbitKickPosted detects coderabbit', chain.coderabbitKickPosted([
-  { user: { login: 'MervinPraison' }, body: '@coderabbitai review' },
-]));
-assert('qodoKickPosted detects qodo', chain.qodoKickPosted([
-  { user: { login: 'github-actions[bot]' }, body: '/review' },
-]));
-assert('ignores coderabbit bot comment', !chain.coderabbitKickPosted([
-  { user: { login: 'coderabbitai[bot]' }, body: '@coderabbitai review' },
-]));
-assert('bot PR author match', chain.isBotOpenedPr({ user: { login: 'praisonai-triage-agent[bot]' } }));
-assert('generic bot type match', chain.isBotOpenedPr({ user: { login: 'my-bot', type: 'Bot' } }));
-assert('human author excluded', !chain.isBotOpenedPr({ user: { login: 'MervinPraison', type: 'User' } }));
-
-process.exit(failed ? 1 : 0);
@@ -1,153 +0,0 @@
-/**
- * Idempotent CodeRabbit/Qodo kick for bot-opened PRs.
- * @see .github/workflows/auto-pr-comment.yml, claude.yml, bot-pr-recovery.yml
- */
-
-const KICK_AUTHORS = new Set(['MervinPraison', 'github-actions[bot]']);
-const BOT_PR_AUTHORS = new Set(['praisonai-triage-agent[bot]', 'github-actions[bot]']);
-
-async function listAllComments(github, owner, repo, issueNumber) {
-  if (typeof github.paginate === 'function') {
-    return github.paginate(github.rest.issues.listComments, {
-      owner,
-      repo,
-      issue_number: issueNumber,
-      per_page: 100,
-    });
-  }
-  const { data } = await github.rest.issues.listComments({
-    owner,
-    repo,
-    issue_number: issueNumber,
-    per_page: 100,
-  });
-  return data;
-}
-
-function kickAuthored(comment, marker) {
-  return (
-    KICK_AUTHORS.has(comment.user?.login) &&
-    (comment.body || '').includes(marker)
-  );
-}
-
-function coderabbitKickPosted(comments) {
-  return comments.some((c) => kickAuthored(c, '@coderabbitai review'));
-}
-
-function qodoKickPosted(comments) {
-  return comments.some((c) => kickAuthored(c, '/review'));
-}
-
-function chainKickPosted(comments) {
-  return coderabbitKickPosted(comments) && qodoKickPosted(comments);
-}
-
-function isBotOpenedPr(pr) {
-  if (BOT_PR_AUTHORS.has(pr.user?.login)) return true;
-  return pr.user?.type === 'Bot';
-}
-
-async function kickReviewChain(github, owner, repo, prNumber, core, preFetchedComments = null) {
-  const comments = preFetchedComments || await listAllComments(github, owner, repo, prNumber);
-  const needCoderabbit = !coderabbitKickPosted(comments);
-  const needQodo = !qodoKickPosted(comments);
-
-  if (!needCoderabbit && !needQodo) {
-    core?.info?.(`Review chain already kicked on PR #${prNumber}`);
-    return { kicked: false, reason: 'already_kicked' };
-  }
-
-  if (needCoderabbit) {
-    await github.rest.issues.createComment({
-      owner,
-      repo,
-      issue_number: prNumber,
-      body: '@coderabbitai review',
-    });
-  }
-  if (needQodo) {
-    await github.rest.issues.createComment({
-      owner,
-      repo,
-      issue_number: prNumber,
-      body: '/review',
-    });
-  }
-  core?.info?.(`Kicked review chain for PR #${prNumber} (coderabbit=${needCoderabbit}, qodo=${needQodo})`);
-  return { kicked: true, coderabbit: needCoderabbit, qodo: needQodo };
-}
-
-async function findOpenPrForIssue(github, owner, repo, issueNumber) {
-  const prefix = `claude/issue-${issueNumber}-`;
-  const { data: prs } = await github.rest.pulls.list({
-    owner,
-    repo,
-    state: 'open',
-    sort: 'created',
-    direction: 'desc',
-    per_page: 30,
-  });
-  return (
-    prs.find(
-      (p) => (p.head?.ref || '').startsWith(prefix) && isBotOpenedPr(p)
-    ) || null
-  );
-}
-
-async function kickReviewChainForIssue(github, owner, repo, issueNumber, core) {
-  const pr = await findOpenPrForIssue(github, owner, repo, issueNumber);
-  if (!pr) {
-    core?.info?.(`No open PR for issue #${issueNumber}, skipping review kick`);
-    return { kicked: false, reason: 'no_pr' };
-  }
-  const result = await kickReviewChain(github, owner, repo, pr.number, core);
-  return { ...result, prNumber: pr.number };
-}
-
-async function recoverStalledBotPrs(github, owner, repo, options, core) {
-  const { prNumber = null, minAgeMs = 10 * 60 * 1000, maxRecover = 10 } = options || {};
-  let prs;
-  if (prNumber) {
-    const { data } = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
-    prs = [data];
-  } else if (typeof github.paginate === 'function') {
-    prs = await github.paginate(github.rest.pulls.list, {
-      owner,
-      repo,
-      state: 'open',
-      per_page: 100,
-    });
-  } else {
-    const { data } = await github.rest.pulls.list({ owner, repo, state: 'open', per_page: 100 });
-    prs = data;
-  }
-
-  const cutoff = Date.now() - minAgeMs;
-  let recovered = 0;
-  for (const pr of prs) {
-    if (recovered >= maxRecover) break;
-    if (!isBotOpenedPr(pr)) continue;
-    if (!prNumber && new Date(pr.created_at).getTime() > cutoff) continue;
-    const comments = await listAllComments(github, owner, repo, pr.number);
-    if (chainKickPosted(comments)) continue;
-    await kickReviewChain(github, owner, repo, pr.number, core, comments);
-    recovered += 1;
-  }
-  core?.info?.(`Recovery complete (${recovered} PR(s) kicked)`);
-  return recovered;
-}
-
-module.exports = {
-  KICK_AUTHORS,
-  BOT_PR_AUTHORS,
-  listAllComments,
-  chainKickPosted,
-  coderabbitKickPosted,
-  qodoKickPosted,
-  isBotOpenedPr,
-  kickReviewChain,
-  findOpenPrForIssue,
-  kickReviewChainForIssue,
-  recoverStalledBotPrs,
-};
@@ -1,126 +0,0 @@
-#!/usr/bin/env node
-/**
- * Local self-test for merge-gate.js heuristics (no GitHub API).
- * Run: node .github/scripts/merge-gate-selftest.js
- */
-const mg = require('./merge-gate.js');
-
-let failed = 0;
-function assert(name, cond) {
-  if (!cond) {
-    console.error('FAIL:', name);
-    failed++;
-  } else {
-    console.log('ok:', name);
-  }
-}
-
-// Stale FINAL: push after FINAL, no @claude since head
-const finals = [
-  { user: { login: 'github-actions[bot]' }, body: '@claude FINAL architecture reviewer', created_at: '2026-06-12T08:00:00Z' },
-];
-assert('stale when head after final', mg.isStaleFinalAfterPush(finals, '2026-06-12T09:00:00Z'));
-
-const withRecovery = [
-  ...finals,
-  { user: { login: 'github-actions[bot]' }, body: '@claude FINAL architecture reviewer', created_at: '2026-06-12T09:30:00Z' },
-];
-assert('not stale when @claude after head', !mg.isStaleFinalAfterPush(withRecovery, '2026-06-12T09:00:00Z'));
-
-// Bot CHANGES_REQUESTED then APPROVE
-const reviews = [
-  { user: { login: 'coderabbit[bot]', type: 'Bot' }, state: 'CHANGES_REQUESTED', submitted_at: '2026-06-12T08:00:00Z' },
-  { user: { login: 'coderabbit[bot]', type: 'Bot' }, state: 'APPROVED', submitted_at: '2026-06-12T09:00:00Z' },
-];
-assert('bot approve clears CR', !mg.hasAnyChangesRequested(reviews));
-assert('human CR blocks', mg.hasAnyChangesRequested([
-  { user: { login: 'MervinPraison', type: 'User' }, state: 'CHANGES_REQUESTED', submitted_at: '2026-06-12T09:00:00Z' },
-]));
-
-// Verdict after HEAD
-const verdictComments = [
-  { body: 'MERGE_GATE_VERDICT: APPROVE', created_at: '2026-06-12T08:00:00Z' },
-];
-assert('verdict before head rejected', mg.findMergeGateVerdict(verdictComments, null, '2026-06-12T09:00:00Z') === null);
-assert('verdict after head accepted', mg.findMergeGateVerdict(
-  [{ body: 'MERGE_GATE_VERDICT: APPROVE', created_at: '2026-06-12T10:00:00Z' }],
-  null,
-  '2026-06-12T09:00:00Z'
-) === 'APPROVE');
-
-const noise = [{ user: { login: 'MervinPraison' }, body: '**Merge gate scan** — wait for `@claude`', created_at: new Date().toISOString() }];
-assert('diagnostic comment not a trigger', !mg.hasRecentClaudeTrigger(noise, 35));
-
-// Sensitive + secrets
-assert('workflow path sensitive', mg.sensitivePathReasons([{ filename: '.github/workflows/foo.yml' }]).length === 1);
-assert('ci-only label exempts workflows', mg.sensitivePathReasons(
-  [{ filename: '.github/workflows/foo.yml' }],
-  [mg.WORKFLOW_ONLY_LABEL]
-).length === 0);
-assert('ci-only label not exempt mixed changes', mg.sensitivePathReasons(
-  [{ filename: '.github/workflows/foo.yml' }, { filename: 'src/foo.py' }],
-  [mg.WORKFLOW_ONLY_LABEL]
-).length === 1);
-assert('secret in patch', mg.secretScanReasons([{ filename: 'x.py', patch: '+key = "ghp_abcdefghijklmnopqrstuvwxyz1234567890"' }]).length === 1);
-
-// Claude run scoping — other PR branches must not block
-assert('other branch claude does not block', !mg.hasBlockingClaudeRunForPr(
-  [{ event: 'issue_comment', head_branch: 'other-branch' }],
-  'my-branch'
-));
-assert('same branch claude blocks', mg.hasBlockingClaudeRunForPr(
-  [{ event: 'issue_comment', head_branch: 'my-branch' }],
-  'my-branch'
-));
-assert('issues event never blocks', !mg.hasBlockingClaudeRunForPr(
-  [{ event: 'issues', head_branch: 'my-branch' }],
-  'my-branch'
-));
-
-// Conflict rebase clears after bot completion + FINAL on HEAD
-const conflictTrigger = {
-  user: { login: 'MervinPraison' },
-  body: '@claude this PR has merge conflicts with main. Please rebase',
-  created_at: '2026-06-26T10:30:38Z',
-};
-const rebaseDone = {
-  user: { login: 'praisonai-triage-agent[bot]' },
-  body: 'Rebase complete — PR #2308 onto latest main',
-  created_at: '2026-06-26T10:30:55Z',
-};
-const finalAfterRebase = {
-  user: { login: 'MervinPraison' },
-  body: '@claude You are the FINAL architecture reviewer.',
-  created_at: '2026-06-26T10:39:43Z',
-};
-const headAfterRebase = '2026-06-26T10:39:25Z';
-const rebaseComments = [conflictTrigger, rebaseDone, finalAfterRebase];
-assert('conflict blocks before rebase done', mg.hasRecentConflictComment([conflictTrigger], headAfterRebase));
-assert('conflict clears after rebase + FINAL on HEAD', !mg.hasRecentConflictComment(rebaseComments, headAfterRebase));
-assert('conflict still blocks without FINAL on HEAD', mg.hasRecentConflictComment(
-  [conflictTrigger, rebaseDone],
-  headAfterRebase
-));
-
-// Tests heuristic
-assert('sdk without tests', mg.missingTestsReason([{ filename: 'src/praisonai-agents/a/b.py', additions: 3 }]) !== null);
-assert('sdk with tests ok', mg.missingTestsReason([
-  { filename: 'src/praisonai-agents/a/b.py', additions: 3 },
-  { filename: 'src/praisonai-agents/tests/test_x.py', additions: 10 },
-]) === null);
-
-// PR size
-assert('large PR blocked', mg.prSizeReasons([{ additions: 900 }]).length > 0);
-
-assert('internal PR link accepted', mg.isInternalPullRequestLink(
-  { base: { repo: { full_name: 'MervinPraison/PraisonAI' } } },
-  'MervinPraison',
-  'PraisonAI'
-));
-assert('fork sync PR link rejected', !mg.isInternalPullRequestLink(
-  { number: 21, base: { repo: { full_name: 'Milkmange/PraisonAI' } } },
-  'MervinPraison',
-  'PraisonAI'
-));
-
-process.exit(failed ? 1 : 0);
@@ -1,747 +0,0 @@
-/**
- * Shared merge-gate helpers for claude-merge-gate.yml
- * @see .github/workflows/claude-merge-gate.yml
- */
-
-const CLAUDE_TRIGGER_LOGINS = ['MervinPraison', 'github-actions[bot]'];
-const AUTO_ACTORS = CLAUDE_TRIGGER_LOGINS;
-const CONFLICT_COOLDOWN_MS = 12 * 60 * 60 * 1000;
-const CLAUDE_ACTIVE_MS = 35 * 60 * 1000;
-const POST_PUSH_BUFFER_MS = 5 * 60 * 1000;
-const ALLOWED_MERGE_STATES = new Set(['CLEAN', 'UNSTABLE']);
-const AGENT_PY_MAX_AUTO_LINES = 100;
-const PR_MAX_AUTO_ADDITIONS = 800;
-const PR_MAX_AUTO_FILES = 30;
-const MANUAL_ONLY_LABELS = new Set(['security', 'breaking-change', 'needs-manual-review', 'release']);
-const WORKFLOW_ONLY_LABEL = 'merge-gate-ci-only';
-const CI_ONLY_PATH_PREFIXES = ['.github/workflows/', '.github/actions/', '.github/scripts/merge-gate'];
-const SDK_PATH_PREFIXES = ['src/praisonai-agents/', 'src/praisonai/'];
-const SENSITIVE_PATH_PATTERNS = [
-  /^\.github\/workflows\//,
-  /praisonaiagents\/(auth|approval|policy|sandbox)\//,
-  /^src\/praisonai-agents\/pyproject\.toml$/,
-  /^src\/praisonai\/pyproject\.toml$/,
-  /\.env(\.|$)/,
-  /credentials\.json$/i,
-];
-const REQUIRED_SDK_CHECK_PATTERNS = [/test/i, /smoke/i, /core/i, /python package/i, /comprehensive/i, /optimized/i];
-const SECRET_PATTERNS = [
-  /sk-[a-zA-Z0-9]{20,}/,
-  /AKIA[0-9A-Z]{16}/,
-  /ghp_[a-zA-Z0-9]{36,}/,
-  /github_pat_/,
-  /Bearer eyJ[a-zA-Z0-9_-]+\.[a-zA-Z0-9_-]+/,
-  /-----BEGIN (RSA |EC |OPENSSH )?PRIVATE KEY-----/,
-];
-const BLOCK_LABELS = new Set([
-  'claude-conflict-pending',
-  'claude-merge-gate-active',
-  'no-auto-merge',
-  'auto-merged-by-gate',
-]);
-const BOT_REVIEWER_PATTERNS = [
-  'coderabbit',
-  'qodo',
-  'gemini',
-  'copilot',
-  'greptile',
-];
-
-const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
-
-function isFinalClaudeTriggerComment(c) {
-  const body = (c.body || '').toLowerCase();
-  if (!AUTO_ACTORS.includes(c.user.login)) return false;
-  if (!body.includes('@claude')) return false;
-  if (body.includes('merge conflict')) return false;
-  return body.includes('final architecture reviewer') || body.includes('lead engineer');
-}
-
-function isClaudeTriggerNoise(c) {
-  const body = c.body || '';
-  if (body.includes('Merge gate scan')) return true;
-  if (body.includes('MERGE_GATE_VERDICT')) return true;
-  if (body.includes('Merged by **Claude PR merge gate**')) return true;
-  return false;
-}
-
-function hasRecentClaudeTrigger(comments, minutes = 35) {
-  const cutoff = Date.now() - minutes * 60 * 1000;
-  return comments.some((c) => {
-    if (!CLAUDE_TRIGGER_LOGINS.includes(c.user.login)) return false;
-    if (isClaudeTriggerNoise(c)) return false;
-    if (!(c.body || '').includes('@claude')) return false;
-    return new Date(c.created_at).getTime() > cutoff;
-  });
-}
-
-function isConflictRebaseTriggerComment(c) {
-  if (!AUTO_ACTORS.includes(c.user.login)) return false;
-  const body = (c.body || '').toLowerCase();
-  return body.includes('@claude') && body.includes('merge conflict');
-}
-
-function isConflictRebaseCompletionComment(c) {
-  const login = (c.user?.login || '').toLowerCase();
-  if (!login.includes('praisonai-triage') && !login.includes('github-actions')) return false;
-  const body = (c.body || '').toLowerCase();
-  return (
-    body.includes('rebase complete') ||
-    body.includes('rebase onto') ||
-    (body.includes('conflict') && body.includes('resolved'))
-  );
-}
-
-function conflictRebaseQuiescent(comments, headPushedAt) {
-  const conflictTriggers = comments.filter(isConflictRebaseTriggerComment);
-  if (conflictTriggers.length === 0) return true;
-
-  const latestConflict = conflictTriggers.reduce((a, b) =>
-    new Date(a.created_at) > new Date(b.created_at) ? a : b
-  );
-  const conflictTime = new Date(latestConflict.created_at).getTime();
-
-  const rebaseDone = comments.some(
-    (c) =>
-      new Date(c.created_at).getTime() > conflictTime && isConflictRebaseCompletionComment(c)
-  );
-  if (!rebaseDone) return false;
-
-  return finalClaudeCompletedOnSha(comments, headPushedAt);
-}
-
-function hasRecentConflictComment(comments, headPushedAt = null) {
-  const cutoff = Date.now() - CONFLICT_COOLDOWN_MS;
-  const hasRecentTrigger = comments.some((c) => {
-    if (!isConflictRebaseTriggerComment(c)) return false;
-    return new Date(c.created_at).getTime() > cutoff;
-  });
-  if (!hasRecentTrigger) return false;
-
-  if (headPushedAt && conflictRebaseQuiescent(comments, headPushedAt)) {
-    return false;
-  }
-  return true;
-}
-
-function isBotReviewer(login, userType) {
-  const lower = (login || '').toLowerCase();
-  if (userType === 'Bot') return true;
-  if (lower.endsWith('[bot]')) return true;
-  return ['coderabbit', 'qodo', 'gemini', 'copilot', 'greptile'].some((p) => lower.includes(p));
-}
-
-function latestReviewsByUser(reviews) {
-  const latestByUser = new Map();
-  for (const r of reviews) {
-    const login = r.user?.login;
-    if (!login) continue;
-    const prev = latestByUser.get(login);
-    if (!prev || new Date(r.submitted_at) > new Date(prev.submitted_at)) {
-      latestByUser.set(login, r);
-    }
-  }
-  return latestByUser;
-}
-
-function hasHumanChangesRequested(reviews) {
-  for (const [login, review] of latestReviewsByUser(reviews)) {
-    if (review.state !== 'CHANGES_REQUESTED') continue;
-    if (!isBotReviewer(login, review.user?.type)) return true;
-  }
-  return false;
-}
-
-function hasAnyChangesRequested(reviews) {
-  for (const [, review] of latestReviewsByUser(reviews)) {
-    if (review.state === 'CHANGES_REQUESTED') return true;
-  }
-  return false;
-}
-
-function hasFinalClaudeReviewTrigger(comments) {
-  return comments.some(isFinalClaudeTriggerComment);
-}
-
-function isStaleFinalAfterPush(comments, headPushedAt) {
-  if (!headPushedAt) return false;
-  const headTime = new Date(headPushedAt).getTime();
-  const finals = comments.filter(isFinalClaudeTriggerComment);
-  if (finals.length === 0) return true;
-  const latestFinal = finals.reduce((a, b) =>
-    new Date(a.created_at) > new Date(b.created_at) ? a : b
-  );
-  const finalTime = new Date(latestFinal.created_at).getTime();
-  if (headTime <= finalTime + 60000) return false;
-  const claudeSinceHead = comments.some((c) => {
-    if (!CLAUDE_TRIGGER_LOGINS.includes(c.user.login)) return false;
-    if (isClaudeTriggerNoise(c)) return false;
-    if (!(c.body || '').includes('@claude')) return false;
-    return new Date(c.created_at).getTime() >= headTime - 60000;
-  });
-  return !claudeSinceHead;
-}
-
-function needsStaleFinalRecovery(comments, headPushedAt) {
-  return (
-    hasFinalClaudeReviewTrigger(comments) &&
-    isStaleFinalAfterPush(comments, headPushedAt)
-  );
-}
-
-function shouldSkipFinalRecovery(comments, headPushedAt) {
-  const isStale = isStaleFinalAfterPush(comments, headPushedAt);
-  if (isStale) return false;
-  return hasRecentClaudeTrigger(comments, 35);
-}
-
-function finalClaudeCompletedOnSha(comments, headPushedAt) {
-  if (!hasFinalClaudeReviewTrigger(comments)) return false;
-  if (isStaleFinalAfterPush(comments, headPushedAt)) return false;
-  return true;
-}
-
-const FINAL_CLAUDE_REVIEW_BODY =
-  '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL comments above from Gemini, Qodo, CodeRabbit, and Copilot carefully before responding.\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n4. SDK value: review in depth whether the change genuinely adds value to the SDK — never add features for the sake of adding them. It must strengthen the SDK (simpler, more user-friendly, robust, world-class, secure). If it does not clearly add value, request changes or recommend rejecting/closing rather than merging scope creep\n5. Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.\n\n**Phase 2: FIX Valid Issues**\n5. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n6. Push all code fixes directly to THIS branch (do NOT create a new PR)\n7. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n8. If all issues are resolved, approve the PR / close the Issue\n9. If blocking issues remain, request changes / leave clear action items';
-
-async function getMergeState(github, owner, repo, prNumber) {
-  const query = `
-    query($owner: String!, $repo: String!, $number: Int!) {
-      repository(owner: $owner, name: $repo) {
-        pullRequest(number: $number) {
-          mergeStateStatus
-          maintainerCanModify
-          isDraft
-          headRefOid
-          headRef { repository { nameWithOwner } }
-        }
-      }
-    }
-  `;
-  for (let attempt = 0; attempt < 3; attempt++) {
-    const result = await github.graphql(query, { owner, repo, number: prNumber });
-    const prGql = result.repository.pullRequest;
-    const status = (prGql?.mergeStateStatus || '').toUpperCase();
-    if (status && status !== 'UNKNOWN') {
-      return {
-        status,
-        isDraft: prGql.isDraft,
-        headRepo: prGql.headRef?.repository?.nameWithOwner,
-        headSha: prGql.headRefOid,
-        maintainerCanModify: prGql.maintainerCanModify === true,
-      };
-    }
-    if (attempt < 2) await sleep(10000);
-  }
-  const { data: pr } = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
-  return {
-    status: (pr.mergeable_state || '').toUpperCase(),
-    isDraft: pr.draft,
-    headRepo: pr.head.repo?.full_name,
-    headSha: pr.head.sha,
-    maintainerCanModify: pr.maintainer_can_modify === true,
-  };
-}
-
-async function allChecksGreenOnSha(github, owner, repo, sha, core) {
-  const { data } = await github.rest.checks.listForRef({
-    owner,
-    repo,
-    ref: sha,
-    per_page: 100,
-  });
-  const runs = (data.check_runs || []).filter((r) => r.head_sha === sha);
-  if (runs.length === 0) {
-    core?.info?.(`No check runs on ${sha.slice(0, 7)} — allowing (e.g. docs-only PR)`);
-    return true;
-  }
-  for (const run of runs) {
-    if (run.status !== 'completed') {
-      core?.info?.(`Check pending: ${run.name} (${run.status})`);
-      return false;
-    }
-    const ok = ['success', 'neutral', 'skipped'].includes(run.conclusion);
-    if (!ok) {
-      core?.info?.(`Check failed: ${run.name} (${run.conclusion})`);
-      return false;
-    }
-  }
-  return true;
-}
-
-function claudeRunBlocksPr(run, headRef) {
-  if (!run || run.event === 'issues') return false;
-  if (!headRef) return true;
-  return run.head_branch === headRef;
-}
-
-function hasBlockingClaudeRunForPr(runs, headRef) {
-  return (runs || []).some((r) => claudeRunBlocksPr(r, headRef));
-}
-
-async function hasInProgressClaudeAssistant(github, owner, repo, prNumber = null) {
-  try {
-    const { data } = await github.rest.actions.listWorkflowRuns({
-      owner,
-      repo,
-      workflow_id: 'claude.yml',
-      status: 'in_progress',
-      per_page: 20,
-    });
-    const runs = (data.workflow_runs || []).filter((r) => r.event !== 'issues');
-    if (runs.length === 0) return false;
-    if (prNumber == null) return runs.length > 0;
-
-    const { data: pr } = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
-    const headRef = pr.head.ref;
-    return hasBlockingClaudeRunForPr(runs, headRef);
-  } catch {
-    return false;
-  }
-}
-
-function isCiOnlyChange(files) {
-  if (!files.length) return false;
-  return files.every((f) => CI_ONLY_PATH_PREFIXES.some((p) => f.filename.startsWith(p)));
-}
-
-function isInternalPullRequestLink(link, owner, repo) {
-  if (link?.base?.repo?.full_name === `${owner}/${repo}`) return true;
-  const baseUrl = link?.base?.repo?.url || '';
-  return baseUrl.endsWith(`/repos/${owner}/${repo}`);
-}
-
-async function resolvePrNumberFromLinkedPullRequests(github, owner, repo, linked) {
-  const repoFull = `${owner}/${repo}`;
-  const internal = (linked || []).filter((l) => isInternalPullRequestLink(l, owner, repo));
-  for (const link of internal) {
-    if (!link.number) continue;
-    try {
-      const { data } = await github.rest.pulls.get({
-        owner,
-        repo,
-        pull_number: link.number,
-      });
-      if (data.state === 'open' && data.base?.repo?.full_name === repoFull) {
-        return data.number;
-      }
-    } catch (err) {
-      if (err.status !== 404) throw err;
-    }
-  }
-  return null;
-}
-
-async function resolvePrNumberFromHeadBranch(github, owner, repo, branch) {
-  if (!branch || branch === 'main' || branch === 'master') return null;
-  const { data } = await github.rest.pulls.list({
-    owner,
-    repo,
-    state: 'open',
-    head: `${owner}:${branch}`,
-    per_page: 1,
-  });
-  return data[0]?.number || null;
-}
-
-async function resolvePrNumberFromWorkflowRun(github, owner, repo, workflowRun) {
-  const fromLinked = await resolvePrNumberFromLinkedPullRequests(
-    github, owner, repo, workflowRun.pull_requests
-  );
-  if (fromLinked) return fromLinked;
-
-  const fromBranch = await resolvePrNumberFromHeadBranch(
-    github, owner, repo, workflowRun.head_branch
-  );
-  if (fromBranch) return fromBranch;
-
-  return null;
-}
-
-async function listPullFiles(github, owner, repo, prNumber) {
-  if (typeof github.paginate === 'function') {
-    return github.paginate(github.rest.pulls.listFiles, {
-      owner,
-      repo,
-      pull_number: prNumber,
-      per_page: 100,
-    });
-  }
-  const { data } = await github.rest.pulls.listFiles({
-    owner,
-    repo,
-    pull_number: prNumber,
-    per_page: 100,
-  });
-  return data;
-}
-
-function countNewAgentParams(patch) {
-  if (!patch) return 0;
-  return patch
-    .split('\n')
-    .filter((l) => l.startsWith('+') && !l.startsWith('+++'))
-    .filter((l) => {
-      const t = l.slice(1).trim();
-      if (!t || t.startsWith('#') || t.startsWith('"""') || t.startsWith("'''")) return false;
-      if (/^(def|class|@|return\b|if\b|elif\b|else\b|for\b|while\b)/.test(t)) return false;
-      if (/^\w+\s*[:=]/.test(t) && !t.startsWith('self.')) return true;
-      return false;
-    }).length;
-}
-
-function getAgentPyChangeFromFiles(files) {
-  const agentFile = files.find((f) => f.filename.endsWith('praisonaiagents/agent/agent.py'));
-  if (!agentFile) {
-    return { touched: false, additions: 0, newParams: 0 };
-  }
-  return {
-    touched: true,
-    additions: agentFile.additions || 0,
-    newParams: countNewAgentParams(agentFile.patch || ''),
-  };
-}
-
-async function getAgentPyChange(github, owner, repo, prNumber) {
-  const files = await listPullFiles(github, owner, repo, prNumber);
-  return getAgentPyChangeFromFiles(files);
-}
-
-function touchesSdk(files) {
-  return files.some((f) => SDK_PATH_PREFIXES.some((p) => f.filename.startsWith(p)));
-}
-
-function hasManualOnlyLabel(labels) {
-  return labels.some((l) => MANUAL_ONLY_LABELS.has(l));
-}
-
-function sensitivePathReasons(files, labels = []) {
-  if (labels.includes(WORKFLOW_ONLY_LABEL) && isCiOnlyChange(files)) {
-    return [];
-  }
-  const reasons = [];
-  for (const f of files) {
-    if (SENSITIVE_PATH_PATTERNS.some((p) => p.test(f.filename))) {
-      reasons.push(`sensitive path: ${f.filename}`);
-      break;
-    }
-  }
-  return reasons;
-}
-
-function prSizeReasons(files) {
-  const reasons = [];
-  const totalAdditions = files.reduce((sum, f) => sum + (f.additions || 0), 0);
-  if (totalAdditions > PR_MAX_AUTO_ADDITIONS) {
-    reasons.push(`PR +${totalAdditions} lines (>${PR_MAX_AUTO_ADDITIONS}) requires manual review`);
-  }
-  if (files.length > PR_MAX_AUTO_FILES) {
-    reasons.push(`${files.length} files changed (>${PR_MAX_AUTO_FILES}) requires manual review`);
-  }
-  return reasons;
-}
-
-function missingTestsReason(files) {
-  const sdkAdds = files.filter(
-    (f) =>
-      f.filename.startsWith('src/praisonai-agents/') &&
-      f.filename.endsWith('.py') &&
-      !f.filename.endsWith('__init__.py') &&
-      (f.additions || 0) > 0
-  );
-  if (sdkAdds.length === 0) return null;
-  const hasTestChange = files.some(
-    (f) => /\/tests?\//.test(f.filename) || /test_.*\.py$/.test(f.filename) || /_test\.py$/.test(f.filename)
-  );
-  if (!hasTestChange) return 'SDK code added without test file changes — requires manual review';
-  return null;
-}
-
-function secretScanReasons(files) {
-  for (const f of files) {
-    if (/\/tests?\//.test(f.filename) || /test_.*\.py$/.test(f.filename)) continue;
-    const patch = f.patch || '';
-    if (!patch) continue;
-    for (const pattern of SECRET_PATTERNS) {
-      if (pattern.test(patch)) {
-        return [`possible secret in diff (${f.filename}) — requires manual review`];
-      }
-    }
-  }
-  return [];
-}
-
-async function sdkTestChecksReason(github, owner, repo, sha, files, core) {
-  if (!touchesSdk(files)) return null;
-  const { data } = await github.rest.checks.listForRef({
-    owner,
-    repo,
-    ref: sha,
-    per_page: 100,
-  });
-  const runs = (data.check_runs || []).filter((r) => r.head_sha === sha);
-  if (runs.length === 0) {
-    return 'SDK code changed but no CI checks on HEAD';
-  }
-  const testRuns = runs.filter((r) => REQUIRED_SDK_CHECK_PATTERNS.some((p) => p.test(r.name || '')));
-  if (testRuns.length === 0) {
-    return 'SDK code changed but no test check runs on HEAD';
-  }
-  core?.info?.(`SDK test checks on HEAD: ${testRuns.map((r) => r.name).join(', ')}`);
-  return null;
-}
-
-function manualReviewReasonForAgentPy(agentChange) {
-  if (!agentChange.touched) return null;
-  if (agentChange.additions > AGENT_PY_MAX_AUTO_LINES) {
-    return `agent.py +${agentChange.additions} lines (>${AGENT_PY_MAX_AUTO_LINES}) requires manual review`;
-  }
-  if (agentChange.newParams > 0) {
-    return `agent.py adds ${agentChange.newParams} new Agent param(s) — requires manual review`;
-  }
-  return null;
-}
-
-async function listAllComments(github, owner, repo, issueNumber) {
-  if (typeof github.paginate === 'function') {
-    return github.paginate(github.rest.issues.listComments, {
-      owner,
-      repo,
-      issue_number: issueNumber,
-      per_page: 100,
-    });
-  }
-  const { data } = await github.rest.issues.listComments({
-    owner,
-    repo,
-    issue_number: issueNumber,
-    per_page: 100,
-  });
-  return data;
-}
-
-async function getHeadCommitDate(github, owner, repo, prNumber) {
-  try {
-    let commits;
-    if (typeof github.paginate === 'function') {
-      commits = await github.paginate(github.rest.pulls.listCommits, {
-        owner,
-        repo,
-        pull_number: prNumber,
-        per_page: 100,
-      });
-    } else {
-      const { data } = await github.rest.pulls.listCommits({
-        owner,
-        repo,
-        pull_number: prNumber,
-        per_page: 100,
-      });
-      commits = data;
-    }
-    const last = commits[commits.length - 1];
-    return last?.commit?.committer?.date || last?.commit?.author?.date || null;
-  } catch {
-    return null;
-  }
-}
-
-async function loadPrContext(github, owner, repo, prNumber) {
-  const { data: pr } = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
-  const { data: issue } = await github.rest.issues.get({ owner, repo, issue_number: prNumber });
-  const comments = await listAllComments(github, owner, repo, prNumber);
-  const { data: reviews } = await github.rest.pulls.listReviews({
-    owner,
-    repo,
-    pull_number: prNumber,
-    per_page: 100,
-  });
-  const mergeState = await getMergeState(github, owner, repo, prNumber);
-  const headSha = mergeState.headSha || pr.head.sha;
-  const headCommitDate = await getHeadCommitDate(github, owner, repo, prNumber);
-  const headPushedAt = headCommitDate || pr.updated_at;
-
-  return {
-    pr,
-    issue,
-    comments,
-    reviews,
-    mergeState,
-    headSha,
-    headPushedAt,
-    labels: issue.labels.map((l) => l.name),
-    baseRepo: `${owner}/${repo}`,
-  };
-}
-
-async function evaluatePipelineQuiescent(github, owner, repo, prNumber, core, options = {}) {
-  const {
-    forMergeStep = false,
-    skipGlobalClaudeRunCheck = false,
-    skipRecentClaudeCooldown = false,
-  } = options;
-  const ctx = await loadPrContext(github, owner, repo, prNumber);
-  const reasons = [];
-
-  if (ctx.pr.draft) reasons.push('draft');
-  if (ctx.pr.state !== 'open') reasons.push('not open');
-  if (ctx.labels.includes('auto-merged-by-gate')) reasons.push('already merged by gate');
-  if (ctx.labels.includes('no-auto-merge')) reasons.push('no-auto-merge label');
-  if (ctx.labels.includes('claude-conflict-pending')) reasons.push('claude-conflict-pending');
-  if (!forMergeStep && ctx.labels.includes('claude-merge-gate-active')) {
-    reasons.push('claude-merge-gate-active');
-  }
-
-  const { status, headRepo, maintainerCanModify } = ctx.mergeState;
-  if (!ALLOWED_MERGE_STATES.has(status)) reasons.push(`mergeState=${status}`);
-
-  if (headRepo && headRepo !== ctx.baseRepo) {
-    reasons.push('fork PR');
-  }
-
-  if (hasRecentConflictComment(ctx.comments, ctx.headPushedAt)) {
-    reasons.push('recent merge-conflict @claude');
-  }
-  if (!skipRecentClaudeCooldown && hasRecentClaudeTrigger(ctx.comments, 35)) {
-    reasons.push('recent @claude within 35min');
-  }
-
-  if (!skipGlobalClaudeRunCheck && (await hasInProgressClaudeAssistant(github, owner, repo, prNumber))) {
-    reasons.push('claude.yml in progress');
-  }
-
-  const headTime = new Date(ctx.headPushedAt).getTime();
-  if (Date.now() - headTime < POST_PUSH_BUFFER_MS) reasons.push('post-push buffer (<5min)');
-
-  if (!finalClaudeCompletedOnSha(ctx.comments, ctx.headPushedAt)) {
-    if (!hasFinalClaudeReviewTrigger(ctx.comments)) {
-      reasons.push('no FINAL Claude review trigger');
-    } else if (isStaleFinalAfterPush(ctx.comments, ctx.headPushedAt)) {
-      reasons.push('stale FINAL after new commits (needs @claude re-review)');
-    } else {
-      reasons.push('FINAL Claude not complete on HEAD');
-    }
-  }
-
-  if (hasAnyChangesRequested(ctx.reviews)) reasons.push('CHANGES_REQUESTED review');
-
-  if (hasManualOnlyLabel(ctx.labels)) {
-    const manualLabel = ctx.labels.find((l) => MANUAL_ONLY_LABELS.has(l));
-    reasons.push(`manual-only label: ${manualLabel}`);
-  }
-
-  const pullFiles = await listPullFiles(github, owner, repo, prNumber);
-
-  const agentChange = getAgentPyChangeFromFiles(pullFiles);
-  const agentManual = manualReviewReasonForAgentPy(agentChange);
-  if (agentManual) reasons.push(agentManual);
-
-  reasons.push(...sensitivePathReasons(pullFiles, ctx.labels));
-  reasons.push(...prSizeReasons(pullFiles));
-  reasons.push(...secretScanReasons(pullFiles));
-
-  const testsReason = missingTestsReason(pullFiles);
-  if (testsReason) reasons.push(testsReason);
-
-  const sdkCheckReason = await sdkTestChecksReason(github, owner, repo, ctx.headSha, pullFiles, core);
-  if (sdkCheckReason) reasons.push(sdkCheckReason);
-
-  const checksOk = await allChecksGreenOnSha(github, owner, repo, ctx.headSha, core);
-  if (!checksOk) reasons.push('CI not green on HEAD');
-
-  if (ctx.pr.mergeable === false) reasons.push('not mergeable');
-
-  const ready = reasons.length === 0;
-  return {
-    ready,
-    reasons,
-    headSha: ctx.headSha,
-    prNumber,
-  };
-}
-
-async function selectMergeGateCandidates(github, owner, repo, prNumbers, maxCandidates, core) {
-  const readyList = [];
-  const skipped = [];
-  for (const num of prNumbers) {
-    const result = await evaluatePipelineQuiescent(github, owner, repo, num, core);
-    if (result.ready) {
-      readyList.push({ pr_number: num, head_sha: result.headSha });
-    } else {
-      skipped.push({ pr: num, reasons: result.reasons });
-    }
-  }
-  readyList.sort((a, b) => a.pr_number - b.pr_number);
-  return {
-    candidates: readyList.slice(0, maxCandidates),
-    skipped,
-  };
-}
-
-function findMergeGateVerdict(comments, minCreatedAt = null, headPushedAt = null) {
-  const minTime = minCreatedAt ? new Date(minCreatedAt).getTime() : 0;
-  const headTime = headPushedAt ? new Date(headPushedAt).getTime() - 60000 : 0;
-  const gateComments = comments
-    .filter((c) => {
-      if (!(c.body || '').includes('MERGE_GATE_VERDICT:')) return false;
-      const created = new Date(c.created_at).getTime();
-      if (minTime && created < minTime) return false;
-      if (headTime && created < headTime) return false;
-      return true;
-    })
-    .sort((a, b) => new Date(b.created_at) - new Date(a.created_at));
-  if (gateComments.length === 0) return null;
-  const body = gateComments[0].body || '';
-  if (body.includes('MERGE_GATE_VERDICT: APPROVE')) return 'APPROVE';
-  if (body.includes('MERGE_GATE_VERDICT: BLOCK')) return 'BLOCK';
-  return null;
-}
-
-module.exports = {
-  CLAUDE_TRIGGER_LOGINS,
-  AUTO_ACTORS,
-  isFinalClaudeTriggerComment,
-  isClaudeTriggerNoise,
-  hasRecentClaudeTrigger,
-  hasRecentConflictComment,
-  isConflictRebaseTriggerComment,
-  isConflictRebaseCompletionComment,
-  conflictRebaseQuiescent,
-  hasHumanChangesRequested,
-  hasAnyChangesRequested,
-  hasFinalClaudeReviewTrigger,
-  finalClaudeCompletedOnSha,
-  getMergeState,
-  allChecksGreenOnSha,
-  hasInProgressClaudeAssistant,
-  claudeRunBlocksPr,
-  hasBlockingClaudeRunForPr,
-  isCiOnlyChange,
-  isInternalPullRequestLink,
-  resolvePrNumberFromLinkedPullRequests,
-  resolvePrNumberFromHeadBranch,
-  resolvePrNumberFromWorkflowRun,
-  WORKFLOW_ONLY_LABEL,
-  getAgentPyChange,
-  getAgentPyChangeFromFiles,
-  manualReviewReasonForAgentPy,
-  countNewAgentParams,
-  listPullFiles,
-  touchesSdk,
-  hasManualOnlyLabel,
-  sensitivePathReasons,
-  prSizeReasons,
-  missingTestsReason,
-  secretScanReasons,
-  sdkTestChecksReason,
-  listAllComments,
-  getHeadCommitDate,
-  isStaleFinalAfterPush,
-  needsStaleFinalRecovery,
-  shouldSkipFinalRecovery,
-  FINAL_CLAUDE_REVIEW_BODY,
-  loadPrContext,
-  evaluatePipelineQuiescent,
-  selectMergeGateCandidates,
-  findMergeGateVerdict,
-};
@@ -1,76 +0,0 @@
-#!/usr/bin/env node
-/**
- * Run: node .github/scripts/pipeline-status-selftest.js
- */
-const ps = require('./pipeline-status.js');
-const mg = require('./merge-gate.js');
-
-let failed = 0;
-function assert(name, cond) {
-  if (!cond) {
-    console.error('FAIL:', name);
-    failed++;
-  } else {
-    console.log('ok:', name);
-  }
-}
-
-const kickComments = [
-  { user: { login: 'MervinPraison' }, body: '@coderabbitai review' },
-  { user: { login: 'MervinPraison' }, body: '/review' },
-];
-
-const finalComment = {
-  user: { login: 'MervinPraison' },
-  body: '@claude You are the FINAL architecture reviewer.',
-  created_at: '2026-06-26T10:00:00Z',
-};
-
-assert(
-  'reviews pending when not kicked',
-  ps.deriveStage([], { ready: false, reasons: ['no FINAL Claude review trigger'] })
-    === 'pipeline/reviews-pending'
-);
-assert(
-  'final pending after kick',
-  ps.deriveStage(kickComments, { ready: false, reasons: ['no FINAL Claude review trigger'] })
-    === 'pipeline/final-claude-pending'
-);
-assert(
-  'awaiting merge gate after final',
-  ps.deriveStage(
-    [...kickComments, finalComment],
-    { ready: false, reasons: ['recent @claude within 35min'] }
-  ) === 'pipeline/awaiting-merge-gate'
-);
-assert(
-  'merge ready',
-  ps.deriveStage([...kickComments, finalComment], { ready: true, reasons: [] })
-    === 'pipeline/merge-ready'
-);
-
-const blockers = ps.deriveBlockerLabels({
-  ready: false,
-  reasons: ['CI not green on HEAD', 'SDK code added without test file changes — requires manual review'],
-});
-assert('maps ci blocker', blockers.includes('pipeline/blocked:ci'));
-assert('maps manual blocker', blockers.includes('pipeline/blocked:manual-review'));
-
-assert(
-  'internal link matches upstream base',
-  mg.isInternalPullRequestLink(
-    { base: { repo: { full_name: 'MervinPraison/PraisonAI' } } },
-    'MervinPraison',
-    'PraisonAI'
-  )
-);
-assert(
-  'fork sync link rejected',
-  !mg.isInternalPullRequestLink(
-    { number: 21, base: { repo: { full_name: 'Milkmange/PraisonAI' } } },
-    'MervinPraison',
-    'PraisonAI'
-  )
-);
-
-process.exit(failed ? 1 : 0);
@@ -1,202 +0,0 @@
-/**
- * Sync pipeline stage/blocker labels on open PRs.
- * @see .github/workflows/pipeline-status-sync.yml
- */
-
-const mergeGate = require('./merge-gate.js');
-const chain = require('./bot-pr-review-chain.js');
-
-const PIPELINE_PREFIX = 'pipeline/';
-const STAGE_LABELS = [
-  'pipeline/reviews-pending',
-  'pipeline/final-claude-pending',
-  'pipeline/awaiting-merge-gate',
-  'pipeline/merge-ready',
-  'pipeline/merged',
-];
-const BLOCKER_LABELS = [
-  'pipeline/blocked:ci',
-  'pipeline/blocked:conflict',
-  'pipeline/blocked:manual-review',
-  'pipeline/blocked:cooldown',
-  'pipeline/blocked:stale-final',
-  'pipeline/blocked:no-final',
-];
-const ALL_PIPELINE_LABELS = [...STAGE_LABELS, ...BLOCKER_LABELS];
-
-const LABEL_SPECS = [
-  { name: 'pipeline/reviews-pending', color: 'fbca04', description: 'Waiting for CodeRabbit/Qodo/Copilot reviews' },
-  { name: 'pipeline/final-claude-pending', color: 'd4c5f9', description: 'Reviews done; waiting for FINAL @claude' },
-  { name: 'pipeline/awaiting-merge-gate', color: 'c5def5', description: 'FINAL done; waiting for merge gate / CI' },
-  { name: 'pipeline/merge-ready', color: '0e8a16', description: 'Eligible for merge gate auto-merge' },
-  { name: 'pipeline/merged', color: '6f42c1', description: 'Merged via pipeline' },
-  { name: 'pipeline/blocked:ci', color: 'd93f0b', description: 'Blocked: CI not green on HEAD' },
-  { name: 'pipeline/blocked:conflict', color: 'b60205', description: 'Blocked: merge conflict or rebase pending' },
-  { name: 'pipeline/blocked:manual-review', color: 'e99695', description: 'Blocked: requires manual review' },
-  { name: 'pipeline/blocked:cooldown', color: 'fef2c0', description: 'Blocked: post-push or @claude cooldown' },
-  { name: 'pipeline/blocked:stale-final', color: 'f9d0c4', description: 'Blocked: FINAL stale after new commits' },
-  { name: 'pipeline/blocked:no-final', color: 'ededed', description: 'Blocked: no FINAL @claude trigger yet' },
-];
-
-function deriveStage(comments, evalResult) {
-  if (evalResult.ready) return 'pipeline/merge-ready';
-  if (!chain.chainKickPosted(comments)) return 'pipeline/reviews-pending';
-  if (!mergeGate.hasFinalClaudeReviewTrigger(comments)) return 'pipeline/final-claude-pending';
-  return 'pipeline/awaiting-merge-gate';
-}
-
-function reasonToBlockerLabel(reason) {
-  const r = (reason || '').toLowerCase();
-  if (r.includes('ci not green') || r.includes('sdk code changed but no ci')) return 'pipeline/blocked:ci';
-  if (
-    r.includes('conflict') ||
-    r.includes('mergestate=dirty') ||
-    r.includes('not mergeable')
-  ) return 'pipeline/blocked:conflict';
-  if (
-    r.includes('manual') ||
-    r.includes('requires manual') ||
-    r.includes('sensitive path') ||
-    r.includes('no-auto-merge') ||
-    r.includes('manual-only label') ||
-    r.includes('without test') ||
-    r.includes('files changed') ||
-    r.includes('possible secret') ||
-    r.includes('agent.py')
-  ) return 'pipeline/blocked:manual-review';
-  if (r.includes('recent @claude') || r.includes('post-push buffer')) return 'pipeline/blocked:cooldown';
-  if (r.includes('stale final')) return 'pipeline/blocked:stale-final';
-  if (r.includes('no final claude')) return 'pipeline/blocked:no-final';
-  if (r.includes('final claude not complete')) return 'pipeline/blocked:stale-final';
-  if (r.includes('claude.yml in progress') || r.includes('claude-merge-gate-active')) {
-    return null;
-  }
-  return null;
-}
-
-function deriveBlockerLabels(evalResult) {
-  if (evalResult.ready) return [];
-  const labels = new Set();
-  for (const reason of evalResult.reasons || []) {
-    const mapped = reasonToBlockerLabel(reason);
-    if (mapped && mapped.startsWith('pipeline/blocked:')) labels.add(mapped);
-  }
-  return [...labels];
-}
-
-function computePipelineLabels(comments, evalResult) {
-  const stage = deriveStage(comments, evalResult);
-  const blockers = stage === 'pipeline/merge-ready' ? [] : deriveBlockerLabels(evalResult);
-  return { stage, blockers, all: [stage, ...blockers] };
-}
-
-async function ensurePipelineLabels(github, owner, repo, core) {
-  let existing;
-  if (typeof github.paginate === 'function') {
-    existing = await github.paginate(github.rest.issues.listLabelsForRepo, {
-      owner,
-      repo,
-      per_page: 100,
-    });
-  } else {
-    ({ data: existing } = await github.rest.issues.listLabelsForRepo({
-      owner,
-      repo,
-      per_page: 100,
-    }));
-  }
-  const names = new Set(existing.map((l) => l.name));
-  for (const spec of LABEL_SPECS) {
-    if (names.has(spec.name)) continue;
-    try {
-      await github.rest.issues.createLabel({
-        owner,
-        repo,
-        name: spec.name,
-        color: spec.color,
-        description: spec.description,
-      });
-      core?.info?.(`Created label ${spec.name}`);
-    } catch (err) {
-      if (err.status !== 422) throw err;
-      core?.info?.(`Label ${spec.name} already exists`);
-    }
-  }
-}
-
-async function syncPipelineLabels(github, owner, repo, prNumber, core) {
-  const ctx = await mergeGate.loadPrContext(github, owner, repo, prNumber);
-  if (ctx.pr.state !== 'open') return { synced: false, reason: 'not_open' };
-
-  const evalResult = await mergeGate.evaluatePipelineQuiescent(
-    github, owner, repo, prNumber, core
-  );
-  const { stage, blockers, all } = computePipelineLabels(ctx.comments, evalResult);
-
-  const current = ctx.labels.filter((l) => l.startsWith(PIPELINE_PREFIX));
-  const desired = new Set(all);
-  const toRemove = current.filter((l) => !desired.has(l) && l !== 'pipeline/merged');
-  const toAdd = all.filter((l) => !current.includes(l));
-
-  for (const name of toRemove) {
-    try {
-      await github.rest.issues.removeLabel({ owner, repo, issue_number: prNumber, name });
-    } catch (err) {
-      if (err.status !== 404) throw err;
-    }
-  }
-  if (toAdd.length) {
-    await github.rest.issues.addLabels({
-      owner,
-      repo,
-      issue_number: prNumber,
-      labels: toAdd,
-    });
-  }
-
-  core?.info?.(`PR #${prNumber}: stage=${stage} blockers=[${blockers.join(', ')}]`);
-  return { synced: true, stage, blockers, ready: evalResult.ready, reasons: evalResult.reasons };
-}
-
-async function syncOpenPullRequests(github, owner, repo, options, core) {
-  const { maxPrs = 20 } = options || {};
-  await ensurePipelineLabels(github, owner, repo, core);
-  let prs;
-  if (maxPrs <= 100) {
-    ({ data: prs } = await github.rest.pulls.list({
-      owner,
-      repo,
-      state: 'open',
-      per_page: 100,
-    }));
-  } else {
-    prs = await github.paginate(github.rest.pulls.list, {
-      owner,
-      repo,
-      state: 'open',
-      per_page: 100,
-    });
-  }
-  let synced = 0;
-  for (const pr of prs) {
-    if (synced >= maxPrs) break;
-    if (pr.draft) continue;
-    if (pr.head?.repo?.full_name && pr.head.repo.full_name !== `${owner}/${repo}`) continue;
-    await syncPipelineLabels(github, owner, repo, pr.number, core);
-    synced += 1;
-  }
-  core?.info?.(`Pipeline label sync complete (${synced} PR(s))`);
-  return synced;
-}
-
-module.exports = {
-  STAGE_LABELS,
-  BLOCKER_LABELS,
-  ALL_PIPELINE_LABELS,
-  deriveStage,
-  deriveBlockerLabels,
-  computePipelineLabels,
-  ensurePipelineLabels,
-  syncPipelineLabels,
-  syncOpenPullRequests,
-};
@@ -15,10 +15,7 @@ on:
  pull_request_review:
    types: [submitted]
  pull_request:
-    types: [opened, synchronize]
-  schedule:
-    - cron: '30 3,9,15,21 * * *'
-  workflow_dispatch:
+    types: [opened]

 env:
  MAX_POLL_ATTEMPTS: 20
@@ -146,12 +143,9 @@ jobs:
              owner: context.repo.owner,
              repo: context.repo.repo
            });
-            const alreadyPosted = comments.data.some(c => {
-              const body = (c.body || '').toLowerCase();
-              if (!body.includes('@claude') || !triggerLogins.includes(c.user.login)) return false;
-              if (body.includes('merge conflict')) return false;
-              return body.includes('final architecture reviewer') || body.includes('lead engineer');
-            });
+            const alreadyPosted = comments.data.some(c =>
+              c.body.includes('@claude') && triggerLogins.includes(c.user.login)
+            );
            core.setOutput('already_triggered', alreadyPosted ? 'true' : 'false');
            core.setOutput('comments_count', comments.data.length.toString());
            
@@ -192,8 +186,7 @@ jobs:
            const claudeBody = '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. ' + reviewSummary + '.\n\n' +
              'Review Context: ' + comments.data.length + ' total comments, ' + 
              (reviews.coderabbit.length + reviews.qodo.length + reviews.gemini.length + reviews.copilot.length + reviews.greptile.length) + ' bot reviews.\n\n' +
-              'Phase 1: Review per AGENTS.md principles (protocol-driven, backward compatible, no perf impact, SDK value gate — no feature bloat).\n' +
-              'Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.\n' +
+              'Phase 1: Review per AGENTS.md principles (protocol-driven, backward compatible, no perf impact).\n' +
              'Phase 2: FIX valid issues found by prior reviewers.\n' +
              'Phase 3: Provide final verdict (approve or request changes).';
            await github.rest.issues.createComment({issue_number: context.issue.number, owner: context.repo.owner, repo: context.repo.repo, body: claudeBody});
@@ -216,10 +209,6 @@ jobs:
      issues: write
      pull-requests: write
    steps:
-      - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
-
      - name: Wait for Copilot to respond (poll)
        id: poll
        uses: actions/github-script@v7
@@ -306,12 +295,9 @@ jobs:
              owner: context.repo.owner,
              repo: context.repo.repo
            });
-            const alreadyPosted = comments.data.some(c => {
-              const body = (c.body || '').toLowerCase();
-              if (!body.includes('@claude') || !triggerLogins.includes(c.user.login)) return false;
-              if (body.includes('merge conflict')) return false;
-              return body.includes('final architecture reviewer') || body.includes('lead engineer');
-            });
+            const alreadyPosted = comments.data.some(c =>
+              c.body.includes('@claude') && triggerLogins.includes(c.user.login)
+            );
            core.setOutput('already_triggered', alreadyPosted ? 'true' : 'false');

      - name: Trigger Claude final review
@@ -327,23 +313,10 @@ jobs:
              issue_number: pr,
              owner: context.repo.owner,
              repo: context.repo.repo,
-              body: '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL comments above from Gemini, Qodo, CodeRabbit, and Copilot carefully before responding.\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n4. SDK value: review in depth whether the change genuinely adds value to the SDK — never add features for the sake of adding them. It must strengthen the SDK (simpler, more user-friendly, robust, world-class, secure). If it does not clearly add value, request changes or recommend rejecting/closing rather than merging scope creep\n5. Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.\n\n**Phase 2: FIX Valid Issues**\n5. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n6. Push all code fixes directly to THIS branch (do NOT create a new PR)\n7. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n8. If all issues are resolved, approve the PR / close the Issue\n9. If blocking issues remain, request changes / leave clear action items'
+              body: '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL comments above from Gemini, Qodo, CodeRabbit, and Copilot carefully before responding.\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n\n**Phase 2: FIX Valid Issues**\n4. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n5. Push all code fixes directly to THIS branch (do NOT create a new PR)\n6. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n7. If all issues are resolved, approve the PR / close the Issue\n8. If blocking issues remain, request changes / leave clear action items'
            });
            console.log(`Claude triggered on PR #${pr}`);

-      - name: Sync pipeline status labels
-        uses: actions/github-script@v7
-        env:
-          PR_NUMBER: ${{ needs.copilot-after-coderabbit.outputs.pr_number }}
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const pr = parseInt(process.env.PR_NUMBER, 10);
-            await ps.ensurePipelineLabels(github, context.repo.owner, context.repo.repo, core);
-            await ps.syncPipelineLabels(github, context.repo.owner, context.repo.repo, pr, core);
-
  # ================================================================
  # Claude after Qodo fallback path (same polling approach)
  # ================================================================
@@ -359,10 +332,6 @@ jobs:
      issues: write
      pull-requests: write
    steps:
-      - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
-
      - name: Wait for Copilot to respond (poll)
        id: poll
        uses: actions/github-script@v7
@@ -430,12 +399,9 @@ jobs:
              owner: context.repo.owner,
              repo: context.repo.repo
            });
-            const alreadyPosted = comments.data.some(c => {
-              const body = (c.body || '').toLowerCase();
-              if (!body.includes('@claude') || !triggerLogins.includes(c.user.login)) return false;
-              if (body.includes('merge conflict')) return false;
-              return body.includes('final architecture reviewer') || body.includes('lead engineer');
-            });
+            const alreadyPosted = comments.data.some(c =>
+              c.body.includes('@claude') && triggerLogins.includes(c.user.login)
+            );
            core.setOutput('already_triggered', alreadyPosted ? 'true' : 'false');

      - name: Trigger Claude final review
@@ -451,23 +417,10 @@ jobs:
              issue_number: pr,
              owner: context.repo.owner,
              repo: context.repo.repo,
-              body: '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL comments above from Gemini, Qodo, CodeRabbit, and Copilot carefully before responding.\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n4. SDK value: review in depth whether the change genuinely adds value to the SDK — never add features for the sake of adding them. It must strengthen the SDK (simpler, more user-friendly, robust, world-class, secure). If it does not clearly add value, request changes or recommend rejecting/closing rather than merging scope creep\n5. Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.\n\n**Phase 2: FIX Valid Issues**\n5. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n6. Push all code fixes directly to THIS branch (do NOT create a new PR)\n7. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n8. If all issues are resolved, approve the PR / close the Issue\n9. If blocking issues remain, request changes / leave clear action items'
+              body: '@claude You are the FINAL architecture reviewer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL comments above from Gemini, Qodo, CodeRabbit, and Copilot carefully before responding.\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n\n**Phase 2: FIX Valid Issues**\n4. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n5. Push all code fixes directly to THIS branch (do NOT create a new PR)\n6. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n7. If all issues are resolved, approve the PR / close the Issue\n8. If blocking issues remain, request changes / leave clear action items'
            });
            console.log(`Claude triggered on PR #${pr}`);

-      - name: Sync pipeline status labels
-        uses: actions/github-script@v7
-        env:
-          PR_NUMBER: ${{ needs.copilot-after-qodo.outputs.pr_number }}
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const pr = parseInt(process.env.PR_NUMBER, 10);
-            await ps.ensurePipelineLabels(github, context.repo.owner, context.repo.repo, core);
-            await ps.syncPipelineLabels(github, context.repo.owner, context.repo.repo, pr, core);
-
  # ================================================================
  # Claude after Gemini — restores the trigger lost from the deleted
  # chain-claude-after-copilot.yml workflow.
@@ -500,12 +453,9 @@ jobs:
              owner: context.repo.owner,
              repo: context.repo.repo
            });
-            const alreadyPosted = comments.data.some(c => {
-              const body = (c.body || '').toLowerCase();
-              if (!body.includes('@claude') || !triggerLogins.includes(c.user.login)) return false;
-              if (body.includes('merge conflict')) return false;
-              return body.includes('final architecture reviewer') || body.includes('lead engineer');
-            });
+            const alreadyPosted = comments.data.some(c =>
+              c.body.includes('@claude') && triggerLogins.includes(c.user.login)
+            );
            core.setOutput('already_triggered', alreadyPosted ? 'true' : 'false');

      - name: Post Claude analysis request (Lead Engineer)
@@ -518,127 +468,9 @@ jobs:
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
-              body: '@claude You are the Lead Engineer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL analysis and reviews above carefully (Gemini, CodeRabbit, Qodo, Copilot, etc).\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n4. SDK value: review in depth whether the change genuinely adds value to the SDK — never add features for the sake of adding them. It must strengthen the SDK (simpler, more user-friendly, robust, world-class, secure). If it does not clearly add value, request changes or recommend rejecting/closing rather than merging scope creep\n5. Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.\n\n**Phase 2: FIX Valid Issues**\n5. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n6. Push all code fixes directly to THIS branch (do NOT create a new PR)\n7. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n8. If all issues are resolved, approve the PR / close the Issue\n9. If blocking issues remain, request changes / leave clear action items'
+              body: '@claude You are the Lead Engineer. If the branch is under MervinPraison/PraisonAI (not a fork), you are able to make modifications to this branch and push directly. SCOPE: Focus ONLY on Python packages (praisonaiagents, praisonai). Do NOT modify praisonai-rust or praisonai-ts. Read ALL analysis and reviews above carefully (Gemini, CodeRabbit, Qodo, Copilot, etc).\n\n**Phase 1: Review per AGENTS.md**\n1. Protocol-driven: check heavy implementations vs core SDK\n2. Backward compatible: ensure zero feature regressions\n3. Performance: no hot-path regressions\n\n**Phase 2: FIX Valid Issues**\n4. For any VALID bugs or architectural flaws found by Gemini, CodeRabbit, Qodo, Copilot, or any other reviewer: implement the fix\n5. Push all code fixes directly to THIS branch (do NOT create a new PR)\n6. Comment a summary of exact files modified and what you skipped\n\n**Phase 3: Final Verdict**\n7. If all issues are resolved, approve the PR / close the Issue\n8. If blocking issues remain, request changes / leave clear action items'
            });

-  # ================================================================
-  # Recovery: post FINAL @claude if merge-conflict comment blocked the chain
-  # (e.g. conflict fired before claude-after-copilot could post review prompt)
-  # ================================================================
-  claude-review-recovery:
-    if: |
-      github.event_name == 'pull_request' &&
-      github.event.action == 'synchronize'
-    runs-on: ubuntu-latest
-    permissions:
-      issues: write
-      pull-requests: read
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Re-post FINAL Claude review when stale or missing
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const pr = context.payload.pull_request.number;
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-
-            const comments = await mergeGate.listAllComments(github, owner, repo, pr);
-            const headPushedAt = (await mergeGate.getHeadCommitDate(github, owner, repo, pr))
-              || context.payload.pull_request.updated_at;
-
-            const hasFinal = mergeGate.hasFinalClaudeReviewTrigger(comments);
-            const isStale = mergeGate.isStaleFinalAfterPush(comments, headPushedAt);
-
-            if (mergeGate.shouldSkipFinalRecovery(comments, headPushedAt)) {
-              core.info('Recent @claude within 35min and FINAL not stale — skip recovery');
-              return;
-            }
-
-            const finalBody = mergeGate.FINAL_CLAUDE_REVIEW_BODY;
-
-            if (hasFinal && isStale) {
-              await github.rest.issues.createComment({ issue_number: pr, owner, repo, body: finalBody });
-              core.info(`Posted stale-FINAL recovery @claude on PR #${pr}`);
-              return;
-            }
-
-            if (hasFinal) {
-              core.info('FINAL Claude review current on HEAD — skip recovery');
-              return;
-            }
-
-            const hasReviewerActivity = comments.some(c => {
-              const login = c.user.login.toLowerCase();
-              return login.includes('coderabbit') || login.includes('copilot') ||
-                login.includes('greptile') || login.includes('qodo') || login.includes('gemini');
-            });
-            if (!hasReviewerActivity) {
-              core.info('No bot reviewer activity yet, skipping recovery');
-              return;
-            }
-
-            await github.rest.issues.createComment({ issue_number: pr, owner, repo, body: finalBody });
-            core.info(`Posted missing FINAL Claude review on PR #${pr}`);
-
-      - name: Sync pipeline status labels
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const pr = context.payload.pull_request.number;
-            await ps.ensurePipelineLabels(github, context.repo.owner, context.repo.repo, core);
-            await ps.syncPipelineLabels(github, context.repo.owner, context.repo.repo, pr, core);
-
-  # Scheduled: re-post @claude when FINAL is stale (Claude pushed after last FINAL trigger)
-  stale-final-recovery-scheduled:
-    if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
-    runs-on: ubuntu-latest
-    permissions:
-      issues: write
-      pull-requests: read
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Re-post stale FINAL @claude on open internal PRs
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const prs = await github.paginate(github.rest.pulls.list, {
-              owner, repo, state: 'open', per_page: 100,
-            });
-            let posted = 0;
-            for (const pr of prs) {
-              if (pr.draft) continue;
-              if (pr.head?.repo?.full_name && pr.head.repo.full_name !== `${owner}/${repo}`) continue;
-              const comments = await mergeGate.listAllComments(github, owner, repo, pr.number);
-              const headPushedAt = (await mergeGate.getHeadCommitDate(github, owner, repo, pr.number)) || pr.updated_at;
-              if (!mergeGate.needsStaleFinalRecovery(comments, headPushedAt)) continue;
-              if (mergeGate.hasRecentClaudeTrigger(comments, 10)) {
-                core.info(`PR #${pr.number}: stale FINAL but @claude within 10min — skip`);
-                continue;
-              }
-              await github.rest.issues.createComment({
-                issue_number: pr.number, owner, repo,
-                body: mergeGate.FINAL_CLAUDE_REVIEW_BODY,
-              });
-              posted++;
-              core.info(`Posted scheduled stale-FINAL recovery on PR #${pr.number}`);
-              if (posted >= 5) break;
-            }
-            core.info(`Scheduled stale-FINAL recovery complete (${posted} posted)`);
-
  # ================================================================
  # BOT PRs: CodeRabbit/Qodo skip bot-authored PRs by default.
  # Trigger them explicitly. Do NOT trigger Copilot here — it will
@@ -656,29 +488,38 @@ jobs:
    runs-on: ubuntu-latest
    permissions:
      pull-requests: write
-      issues: write
    steps:
-      - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
-
      - name: Trigger CodeRabbit and Qodo reviews
        uses: actions/github-script@v7
        with:
          github-token: ${{ secrets.GH_TOKEN }}
          script: |
-            const path = require('path');
-            const chain = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/bot-pr-review-chain.js'));
            const pr = context.payload.pull_request.number;
-            await chain.kickReviewChain(github, context.repo.owner, context.repo.repo, pr, core);
+            const owner = context.repo.owner;
+            const repo = context.repo.repo;

-      - name: Sync pipeline status labels
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const pr = context.payload.pull_request.number;
-            await ps.ensurePipelineLabels(github, context.repo.owner, context.repo.repo, core);
-            await ps.syncPipelineLabels(github, context.repo.owner, context.repo.repo, pr, core);
+            // 1. Trigger CodeRabbit review
+            await github.rest.issues.createComment({
+              issue_number: pr, owner, repo,
+              body: '@coderabbitai review'
+            });
+            console.log(`Triggered CodeRabbit for bot PR #${pr}`);
+
+            // 2. Trigger Qodo review
+            await github.rest.issues.createComment({
+              issue_number: pr, owner, repo,
+              body: '/review'
+            });
+            console.log(`Triggered Qodo for bot PR #${pr}`);
+
+            // 3. Trigger Gemini review (commented out)
+            // await github.rest.issues.createComment({
+            //   issue_number: pr, owner, repo,
+            //   body: '@gemini review this PR'
+            // });
+            // console.log(`Triggered Gemini for bot PR #${pr}`);
+
+            // NOTE: @copilot is NOT triggered here.
+            // It will be triggered by copilot-after-coderabbit or copilot-after-qodo
+            // once those bots finish, ensuring Copilot always reviews LAST.
+            console.log('Copilot will be triggered after CodeRabbit/Qodo complete.');
@@ -1,43 +0,0 @@
-name: Bot PR Recovery
-
-# Belt-and-braces: kick CodeRabbit/Qodo when pull_request:opened never fired for bot PRs.
-
-on:
-  schedule:
-    - cron: '*/15 * * * *'
-  workflow_dispatch:
-    inputs:
-      pr_number:
-        description: 'Specific PR number to kick (optional)'
-        required: false
-        type: string
-
-permissions:
-  pull-requests: write
-  issues: write
-
-jobs:
-  recover:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
-
-      - name: Find and kick stalled bot PRs
-        uses: actions/github-script@v7
-        env:
-          PR_NUMBER_INPUT: ${{ github.event.inputs.pr_number }}
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const chain = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/bot-pr-review-chain.js'));
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const manual = (process.env.PR_NUMBER_INPUT || '').trim();
-            const prNumber = manual ? parseInt(manual, 10) : null;
-            if (manual && Number.isNaN(prNumber)) {
-              throw new Error(`Invalid pr_number: ${manual}`);
-            }
-            await chain.recoverStalledBotPrs(github, owner, repo, { prNumber }, core);
@@ -1,474 +0,0 @@
-name: Claude PR Merge Gate
-
-# Read-only Claude assesses merge readiness; GH_TOKEN merges (no @claude, no gh pr review).
-# Does not overlap with merge-conflict-claude.yml or auto-pr-comment @claude chains.
-
-on:
-  schedule:
-    - cron: '0 2,8,14,20 * * *'
-  workflow_dispatch:
-    inputs:
-      pr_number:
-        description: 'Optional PR number (scan all open internal PRs if empty)'
-        required: false
-        type: string
-      merge_method:
-        description: 'Merge method'
-        required: false
-        type: choice
-        default: merge
-        options:
-          - merge
-          - squash
-          - rebase
-  workflow_run:
-    workflows:
-      - Claude Assistant
-      - Core Tests
-    types:
-      - completed
-  repository_dispatch:
-    types:
-      - claude-merge-gate
-
-concurrency:
-  group: claude-merge-gate-${{ github.repository }}-${{ github.event.inputs.pr_number || github.event.client_payload.pr_number || github.event.workflow_run.id || 'scan' }}
-  cancel-in-progress: true
-
-env:
-  GH_TOKEN: ${{ secrets.GH_TOKEN || github.token }}
-  MAX_CANDIDATES: 5
-  DISPATCH_PR_NUMBER: ${{ github.event.inputs.pr_number || github.event.client_payload.pr_number || '' }}
-
-permissions:
-  contents: read
-  pull-requests: read
-  issues: read
-  actions: read
-
-jobs:
-  auto-dispatch:
-    if: github.event_name == 'workflow_run'
-    runs-on: ubuntu-latest
-    permissions:
-      contents: read
-      pull-requests: read
-      issues: read
-      actions: write
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Dispatch merge gate when PR pipeline is ready
-        uses: actions/github-script@v7
-        env:
-          PAT: ${{ secrets.GH_TOKEN }}
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const wr = context.payload.workflow_run;
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-
-            if (!process.env.PAT) {
-              core.setFailed('GH_TOKEN secret is required for merge gate auto-dispatch (needs repo + workflow scope).');
-              return;
-            }
-
-            if (wr.conclusion !== 'success') {
-              core.info(`Skip dispatch: ${wr.name} conclusion=${wr.conclusion}`);
-              return;
-            }
-            const allowed = ['Claude Assistant', 'Core Tests'];
-            if (!allowed.includes(wr.name)) {
-              core.info(`Skip dispatch: workflow ${wr.name}`);
-              return;
-            }
-
-            const prNumber = await mergeGate.resolvePrNumberFromWorkflowRun(github, owner, repo, wr);
-            if (!prNumber) {
-              core.info('Skip dispatch: could not resolve PR from workflow run');
-              return;
-            }
-
-            const fromClaude = wr.name === 'Claude Assistant';
-            const check = await mergeGate.evaluatePipelineQuiescent(
-              github, owner, repo, prNumber, core, {
-                skipGlobalClaudeRunCheck: fromClaude,
-                skipRecentClaudeCooldown: fromClaude,
-              }
-            );
-            if (!check.ready) {
-              core.info(`PR #${prNumber} not ready: ${check.reasons.join(', ')}`);
-              return;
-            }
-
-            await github.rest.repos.createDispatchEvent({
-              owner,
-              repo,
-              event_type: 'claude-merge-gate',
-              client_payload: { pr_number: prNumber },
-            });
-            core.info(`Dispatched merge gate for PR #${prNumber} after ${wr.name}`);
-
-  scan-candidates:
-    if: github.event_name != 'workflow_run'
-    runs-on: ubuntu-latest
-    permissions:
-      contents: read
-      pull-requests: read
-      issues: write
-      actions: read
-    outputs:
-      matrix: ${{ steps.scan.outputs.matrix }}
-      has_candidates: ${{ steps.scan.outputs.has_candidates }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Scan open PRs for merge gate eligibility
-        id: scan
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const maxCandidates = parseInt(process.env.MAX_CANDIDATES || '5', 10);
-            const inputPr = String(process.env.DISPATCH_PR_NUMBER || '').trim();
-
-            const candidates = [];
-            const skipped = [];
-
-            if (inputPr) {
-              const n = parseInt(inputPr, 10);
-              if (Number.isNaN(n)) {
-                throw new Error(`Invalid pr_number input: ${inputPr}`);
-              }
-              const result = await mergeGate.evaluatePipelineQuiescent(github, owner, repo, n, core);
-              if (result.ready) {
-                candidates.push({ pr_number: n, head_sha: result.headSha });
-              } else {
-                skipped.push({ pr: n, reasons: result.reasons });
-              }
-            } else {
-              const prs = await github.paginate(github.rest.pulls.list, {
-                owner, repo, state: 'open', per_page: 100, sort: 'created', direction: 'asc',
-              });
-              const prNumbers = prs.map((p) => p.number);
-              const selected = await mergeGate.selectMergeGateCandidates(
-                github, owner, repo, prNumbers, maxCandidates, core
-              );
-              candidates.push(...selected.candidates);
-              skipped.push(...selected.skipped);
-            }
-
-            core.info('Skipped: ' + JSON.stringify(skipped));
-            core.setOutput('matrix', JSON.stringify({ include: candidates }));
-            core.setOutput('has_candidates', candidates.length > 0 ? 'true' : 'false');
-
-            core.summary.addRaw('## Merge gate scan\n');
-            core.summary.addRaw(`Candidates: ${candidates.length}\n`);
-            core.summary.addRaw('```json\n' + JSON.stringify({ candidates, skipped }, null, 2) + '\n```');
-            await core.summary.write();
-
-            if (inputPr && skipped.length === 1 && skipped[0].pr === parseInt(inputPr, 10)) {
-              const reasons = skipped[0].reasons;
-              const body = [
-                '**Merge gate scan** — not eligible for auto-merge.',
-                '',
-                ...reasons.map((r) => `- \`${r}\``),
-                '',
-                '**Actions:** wait for CI and the Claude review chain, or add label `needs-manual-review` and merge manually.',
-                '**Opt out:** label `no-auto-merge`.',
-              ].join('\n');
-              await github.rest.issues.createComment({
-                owner, repo, issue_number: skipped[0].pr, body,
-              });
-            } else if (inputPr && candidates.length === 1) {
-              await github.rest.issues.createComment({
-                owner, repo, issue_number: candidates[0].pr_number,
-                body: '**Merge gate scan** — eligible for assessment. Claude merge gate will assess and may auto-merge if `MERGE_GATE_VERDICT: APPROVE`.',
-              });
-            }
-
-            await ps.syncOpenPullRequests(github, owner, repo, { maxPrs: 20 }, core);
-
-  claude-assess:
-    needs: scan-candidates
-    if: needs.scan-candidates.outputs.has_candidates == 'true'
-    runs-on: ubuntu-latest
-    timeout-minutes: 20
-    strategy:
-      fail-fast: false
-      matrix: ${{ fromJson(needs.scan-candidates.outputs.matrix) }}
-    permissions:
-      contents: read
-      pull-requests: read
-      issues: write
-      actions: read
-      id-token: write
-    steps:
-      - name: Generate GitHub App Token
-        id: app-token
-        uses: actions/create-github-app-token@v1
-        with:
-          app-id: ${{ secrets.CLAUDE_APP_ID }}
-          private-key: ${{ secrets.CLAUDE_APP_PRIVATE_KEY }}
-          owner: ${{ github.repository_owner }}
-
-      - uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          token: ${{ steps.app-token.outputs.token }}
-
-      - name: Mark merge gate active
-        id: mark_active
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            core.setOutput('started_at', new Date().toISOString());
-            await github.rest.issues.addLabels({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              issue_number: ${{ matrix.pr_number }},
-              labels: ['claude-merge-gate-active'],
-            });
-
-      - name: Stash merge gate helpers
-        run: cp .github/scripts/merge-gate.js /tmp/merge-gate.js
-
-      - name: Fetch PR branch
-        env:
-          PR_NUMBER: ${{ matrix.pr_number }}
-        run: |
-          git fetch origin "pull/${PR_NUMBER}/head:pr-${PR_NUMBER}"
-          git checkout "pr-${PR_NUMBER}"
-
-      - name: Claude merge gate assessment (read-only)
-        id: assess
-        continue-on-error: true
-        uses: ./.github/actions/claude-code-action
-        env:
-          GH_TOKEN: ${{ steps.app-token.outputs.token }}
-          CLAUDE_CODE_SUBAGENT_MODEL: inherit
-          ANTHROPIC_DEFAULT_SONNET_MODEL: claude-sonnet-4-6
-        with:
-          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
-          github_token: ${{ steps.app-token.outputs.token }}
-          model: claude-opus-4-8
-          trigger_phrase: '@claude-merge-gate-internal'
-          direct_prompt: |
-            You are the MERGE GATE reviewer for PraisonAI PR #${{ matrix.pr_number }}.
-            READ-ONLY: do not modify code, commit, push, rebase, or create branches.
-
-            CRITICAL OUTPUT RULE:
-            Post exactly one PR comment via GitHub MCP containing either:
-            MERGE_GATE_VERDICT: APPROVE
-            or
-            MERGE_GATE_VERDICT: BLOCK
-            followed by a short rationale on the next lines.
-
-            NEVER include @claude in any comment. NEVER mention merge conflict resolution.
-
-            Review checklist:
-            1. Read AGENTS.md and the PR diff on the current branch.
-            2. Read all PR comments and reviews (FINAL architecture reviewer / Lead Engineer must have run).
-            3. Confirm SDK value gate: change strengthens praisonaiagents (no scope creep).
-            4. Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.
-            5. If agent.py adds >100 lines or new Agent __init__ params, respond MERGE_GATE_VERDICT: BLOCK (manual review required).
-            6. BLOCK if labels security/breaking-change/needs-manual-review/release, sensitive paths (.github/workflows, auth, pyproject.toml), secrets in diff, PR >800 lines or >30 files, or SDK changes without tests. Label `merge-gate-ci-only` exempts CI-only changes under `.github/workflows/`, `.github/actions/`, or merge-gate scripts.
-            7. Confirm no CHANGES_REQUESTED reviews and CI green on HEAD ${{ matrix.head_sha }}.
-            8. If ready to merge: MERGE_GATE_VERDICT: APPROVE. Otherwise: MERGE_GATE_VERDICT: BLOCK with reasons.
-
-            Do not approve or merge via GitHub UI/API — only post the verdict comment.
-          disallowed_tools: |
-            Bash
-            Edit
-            Write
-            Replace
-          allowed_tools: |
-            View
-            GlobTool
-            GrepTool
-            Read
-          timeout_minutes: 15
-
-      - name: Post fallback verdict if Claude did not comment
-        id: fallback_verdict
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const mergeGate = require('/tmp/merge-gate.js');
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const prNumber = ${{ matrix.pr_number }};
-
-            const ctx = await mergeGate.loadPrContext(github, owner, repo, prNumber);
-            const minCreatedAt = new Date(Date.now() - 25 * 60 * 1000).toISOString();
-            const existing = mergeGate.findMergeGateVerdict(ctx.comments, minCreatedAt, ctx.headPushedAt);
-            if (existing) {
-              core.info(`Verdict already posted: ${existing}`);
-              core.setOutput('verdict', existing);
-              return;
-            }
-
-            const check = await mergeGate.evaluatePipelineQuiescent(
-              github, owner, repo, prNumber, core, { forMergeStep: true }
-            );
-            const verdict = check.ready ? 'APPROVE' : 'BLOCK';
-            const lines = [
-              `MERGE_GATE_VERDICT: ${verdict}`,
-              '',
-              'Automated fallback — Claude assess did not post a verdict comment (e.g. GitHub MCP unavailable).',
-            ];
-            if (check.ready) {
-              lines.push('All deterministic merge gates passed.');
-            } else {
-              lines.push('Blockers:', ...check.reasons.map((r) => `- ${r}`));
-            }
-            await github.rest.issues.createComment({
-              owner, repo, issue_number: prNumber,
-              body: lines.join('\n'),
-            });
-            core.info(`Posted fallback verdict: ${verdict}`);
-            core.setOutput('verdict', verdict);
-
-      - name: Verify verdict posted
-        id: verdict
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const mergeGate = require('/tmp/merge-gate.js');
-            const comments = await mergeGate.listAllComments(
-              github, context.repo.owner, context.repo.repo, ${{ matrix.pr_number }}
-            );
-            const ctx = await mergeGate.loadPrContext(
-              github, context.repo.owner, context.repo.repo, ${{ matrix.pr_number }}
-            );
-            const minCreatedAt = new Date(Date.now() - 25 * 60 * 1000).toISOString();
-            const verdict = mergeGate.findMergeGateVerdict(comments, minCreatedAt, ctx.headPushedAt);
-            core.setOutput('verdict', verdict || 'NONE');
-            if (!verdict) core.warning('No recent MERGE_GATE_VERDICT comment found after assessment');
-
-      - name: Remove merge gate active label
-        if: always()
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const prNumber = ${{ matrix.pr_number }};
-            try {
-              await github.rest.issues.removeLabel({
-                owner: context.repo.owner,
-                repo: context.repo.repo,
-                issue_number: prNumber,
-                name: 'claude-merge-gate-active',
-              });
-              core.info(`Removed claude-merge-gate-active from PR #${prNumber}`);
-            } catch (e) {
-              core.info(`Label claude-merge-gate-active already removed or missing on PR #${prNumber}`);
-            }
-
-      - name: Fail job if assessment failed
-        if: steps.assess.outcome == 'failure'
-        run: exit 1
-
-  merge-only:
-    needs: [scan-candidates, claude-assess]
-    if: |
-      always() &&
-      needs.scan-candidates.outputs.has_candidates == 'true' &&
-      needs.claude-assess.result == 'success'
-    runs-on: ubuntu-latest
-    strategy:
-      fail-fast: false
-      matrix: ${{ fromJson(needs.scan-candidates.outputs.matrix) }}
-    permissions:
-      contents: write
-      pull-requests: write
-      issues: write
-      actions: read
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Re-check gates and merge
-        uses: actions/github-script@v7
-        env:
-          MERGE_METHOD: ${{ github.event.inputs.merge_method || 'merge' }}
-          SCAN_HEAD_SHA: ${{ matrix.head_sha }}
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const prNumber = ${{ matrix.pr_number }};
-            const scanHeadSha = process.env.SCAN_HEAD_SHA;
-
-            const removeLabels = async () => {
-              try {
-                await github.rest.issues.removeLabel({
-                  owner, repo, issue_number: prNumber, name: 'claude-merge-gate-active',
-                });
-              } catch (e) {
-                core.info('Label claude-merge-gate-active already removed or missing');
-              }
-            };
-
-            try {
-              const ctx = await mergeGate.loadPrContext(github, owner, repo, prNumber);
-              const comments = ctx.comments;
-              const verdictMin = new Date(Date.now() - 25 * 60 * 1000).toISOString();
-              const verdict = mergeGate.findMergeGateVerdict(comments, verdictMin, ctx.headPushedAt);
-              if (verdict !== 'APPROVE') {
-                core.info(`Skip merge: verdict=${verdict || 'NONE'}`);
-                await removeLabels();
-                return;
-              }
-
-              const check = await mergeGate.evaluatePipelineQuiescent(
-                github, owner, repo, prNumber, core, { forMergeStep: true }
-              );
-              if (!check.ready) {
-                core.info(`Skip merge: ${check.reasons.join(', ')}`);
-                await removeLabels();
-                return;
-              }
-
-              if (check.headSha !== scanHeadSha) {
-                core.info(`Skip merge: HEAD moved ${scanHeadSha.slice(0, 7)} → ${check.headSha.slice(0, 7)}`);
-                await removeLabels();
-                return;
-              }
-
-              const method = process.env.MERGE_METHOD || 'merge';
-              await github.rest.pulls.merge({
-                owner, repo, pull_number: prNumber, merge_method: method,
-              });
-
-              await github.rest.issues.addLabels({
-                owner, repo, issue_number: prNumber, labels: ['auto-merged-by-gate'],
-              });
-
-              await github.rest.issues.createComment({
-                owner, repo, issue_number: prNumber,
-                body: [
-                  'Merged by **Claude PR merge gate** (`claude-merge-gate.yml`).',
-                  `Verdict: MERGE_GATE_VERDICT: APPROVE`,
-                  `SHA: \`${check.headSha.slice(0, 7)}\``,
-                  `Method: ${method}`,
-                ].join('\n'),
-              });
-
-              core.info(`Merged PR #${prNumber}`);
-            } finally {
-              await removeLabels();
-            }
@@ -123,8 +123,6 @@ jobs:
      github.event.action != 'opened' &&
      (github.event.action != 'labeled' || github.event.label.name == 'claude') &&
      (github.event_name != 'issue_comment' || contains(github.event.comment.body, '@claude')) &&
-      (github.event_name != 'pull_request_review' || contains(github.event.review.body, '@claude')) &&
-      (github.event_name != 'pull_request_review_comment' || contains(github.event.comment.body, '@claude')) &&
      (
        !contains(github.actor, '[bot]') ||
        github.actor == 'github-actions[bot]' ||
@@ -153,67 +151,26 @@ jobs:
        id: check_fork
        uses: actions/github-script@v7
        with:
-          github-token: ${{ steps.app-token.outputs.token }}
          script: |
-            const isPR = context.eventName === 'pull_request'
-              || context.eventName === 'pull_request_review'
-              || context.eventName === 'pull_request_review_comment'
-              || (context.payload.issue && context.payload.issue.pull_request);
-
-            if (!isPR) {
-              core.setOutput('is_pr', 'false');
-              core.setOutput('is_fork', 'false');
-              core.setOutput('can_push_fork', 'false');
-              return;
-            }
-
-            const prNumber = context.payload.pull_request
-              ? context.payload.pull_request.number
-              : context.payload.issue.number;
-            const pr = await github.rest.pulls.get({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              pull_number: prNumber
-            });
-
-            let mergeState = '';
-            try {
-              const gql = await github.graphql(
-                `query($owner: String!, $repo: String!, $number: Int!) {
-                  repository(owner: $owner, name: $repo) {
-                    pullRequest(number: $number) { mergeStateStatus }
-                  }
-                }`,
-                { owner: context.repo.owner, repo: context.repo.repo, number: prNumber }
-              );
-              mergeState = (gql.repository.pullRequest.mergeStateStatus || '').toUpperCase();
-            } catch (e) {
-              core.warning(`GraphQL mergeStateStatus failed: ${e.message}`);
-            }
-            if (!mergeState) {
-              mergeState = (pr.data.mergeable_state || pr.data.merge_state_status || '').toUpperCase();
-            }
-
-            core.setOutput('is_pr', 'true');
-            core.setOutput('pr_number', String(prNumber));
-            core.setOutput('pr_branch', pr.data.head.ref);
-            core.setOutput('head_repo', pr.data.head.repo.full_name);
-            core.setOutput('merge_state', mergeState);
-
-            const baseRepo = `${context.repo.owner}/${context.repo.repo}`;
-            const maintainerCanModify = pr.data.maintainer_can_modify === true;
-            core.setOutput('maintainer_can_modify', String(maintainerCanModify));
-
-            if (pr.data.head.repo && pr.data.head.repo.full_name !== baseRepo) {
-              core.setOutput('is_fork', 'true');
-              core.setOutput('can_push_fork', String(maintainerCanModify));
-              core.setOutput('fork_repo', pr.data.head.repo.full_name);
-              core.setOutput('clone_url', pr.data.head.repo.clone_url);
-              core.setOutput('branch', pr.data.head.ref);
-              return;
+            const isPR = context.eventName === 'pull_request' || (context.payload.issue && context.payload.issue.pull_request);
+            if (isPR) {
+              const prNumber = context.payload.pull_request ? context.payload.pull_request.number : context.payload.issue.number;
+              const pr = await github.rest.pulls.get({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                pull_number: prNumber
+              });
+              
+              core.setOutput('pr_branch', pr.data.head.ref);
+              
+              if (pr.data.head.repo && pr.data.head.repo.full_name !== context.repo.full_name) {
+                core.setOutput('is_fork', 'true');
+                core.setOutput('clone_url', pr.data.head.repo.clone_url);
+                core.setOutput('branch', pr.data.head.ref);
+                return;
+              }
            }
            core.setOutput('is_fork', 'false');
-            core.setOutput('can_push_fork', 'false');

      - name: Checkout repository
        uses: actions/checkout@v4
@@ -223,40 +180,31 @@ jobs:
          token: ${{ steps.app-token.outputs.token }}

      - name: Fetch PR branch and Setup Remote
-        if: steps.check_fork.outputs.is_pr == 'true'
+        if: github.event.issue.pull_request
        env:
          PR_BRANCH: ${{ steps.check_fork.outputs.pr_branch }}
-          PR_NUMBER: ${{ steps.check_fork.outputs.pr_number }}
          IS_FORK: ${{ steps.check_fork.outputs.is_fork }}
-          CAN_PUSH_FORK: ${{ steps.check_fork.outputs.can_push_fork }}
-          FORK_REPO: ${{ steps.check_fork.outputs.fork_repo }}
-          GH_TOKEN: ${{ secrets.GH_TOKEN }}
        run: |
-          git fetch origin "pull/${PR_NUMBER}/head:${PR_BRANCH}"
-          git checkout "${PR_BRANCH}"
-          git fetch origin main
-
-          if [ "${IS_FORK}" = "true" ] && [ "${CAN_PUSH_FORK}" = "true" ]; then
-            git remote set-url origin "https://x-access-token:${GH_TOKEN}@github.com/${FORK_REPO}.git"
-          elif [ "${IS_FORK}" = "true" ]; then
-            # Block push: trick claude-code-action's `git fetch origin <branch>`
-            git remote set-url origin "file://$(pwd)"
+          # Fetch PR head from base repo and put it into local branch
+          git fetch origin pull/${{ github.event.issue.number }}/head:"${PR_BRANCH}"
+          
+          # If it's a fork, make origin point to local to trick claude-code-action's `git fetch origin <branch>`
+          if [ "${IS_FORK}" = "true" ]; then
+            git remote set-url origin file://$(pwd)
          fi

      - uses: anthropics/claude-code-action@beta
        env:
-          GH_TOKEN: ${{ steps.check_fork.outputs.can_push_fork == 'true' && secrets.GH_TOKEN || steps.app-token.outputs.token }}
+          GH_TOKEN: ${{ steps.app-token.outputs.token }}
        with:
          allowed_bots: 'praisonai-triage-agent[bot]'
          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
          github_token: ${{ steps.app-token.outputs.token }}
-          model: claude-opus-4-8
          trigger_phrase: "@claude"
          label_trigger: "claude"

          direct_prompt: |
-            ${{ steps.check_fork.outputs.is_fork == 'true' && steps.check_fork.outputs.can_push_fork != 'true' && 'CRITICAL: THIS IS A PULL REQUEST FROM A FORK WITHOUT MAINTAINER EDITS. YOU DO NOT HAVE PUSH PERMISSIONS. ONLY READ FILES, VALIDATE CODE, AND PROVIDE COMMENTS/FEEDBACK ON THIS PR. DO NOT ATTEMPT TO PUSH OR CREATE PULL REQUESTS.\n\n' || (steps.check_fork.outputs.can_push_fork == 'true' && format('NOTE: Fork PR with maintainer edits allowed. Push directly to fork {0} branch {1}.\n\n', steps.check_fork.outputs.fork_repo, steps.check_fork.outputs.pr_branch) || (steps.check_fork.outputs.is_pr == 'true' && steps.check_fork.outputs.is_fork != 'true' && 'NOTE: The branch is under MervinPraison/PraisonAI (not a fork). You are able to make modifications and push directly to this branch.\n\n' || (steps.check_fork.outputs.is_pr != 'true' && 'NOTE: This is an issue (not a PR). Create a new fix branch per STEP 3 below.\n\n' || ''))) }}
-            ${{ steps.check_fork.outputs.is_pr == 'true' && format('CONTEXT: PR #{0} on branch {1}. head.repo={2}, is_fork={3}, can_push_fork={4}, merge_state={5}. NEVER treat cursor/*, claude/* branch names or bot authors as forks — only is_fork=true means fork.\n\n', steps.check_fork.outputs.pr_number, steps.check_fork.outputs.pr_branch, steps.check_fork.outputs.head_repo, steps.check_fork.outputs.is_fork, steps.check_fork.outputs.can_push_fork, steps.check_fork.outputs.merge_state) || '' }}
+            ${{ steps.check_fork.outputs.is_fork == 'true' && 'CRITICAL: THIS IS A PULL REQUEST FROM A FORK. YOU DO NOT HAVE PUSH PERMISSIONS TO THIS REPOSITORY. ONLY READ FILES, VALIDATE CODE, AND PROVIDE YOUR COMMENTS/FEEDBACK DIRECTLY IN THIS PR VIA COMMENTS. DO NOT ATTEMPT TO PUSH OR CREATE PULL REQUESTS.\n\n' || 'NOTE: The branch is under MervinPraison/PraisonAI (not a fork). You are able to make modifications and push directly to this branch.\n\n' }}
            You are working on the PraisonAI SDK. Follow AGENTS.md strictly.

            STEP 0 — SETUP GIT IDENTITY & AUTH (GLOBAL — required for all repos):
@@ -279,16 +227,21 @@ jobs:
              5. Plugins (PraisonAI-Plugins): Framework lifecycle extensions replacing systemic behavior (e.g. tracing, logging, metrics, hooks, guardrails).
              For items routed to an EXTERNAL repository (Tools, Docs, Plugins), follow STEP 3-ALT below.
            - Does it duplicate existing functionality? Check if Agent already supports this via existing params (reflection, planning, tools, hooks, memory).
-            - Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.
            - Does it inherit from Agent properly? New agent types MUST inherit Agent, not wrap it with composition.
            - Does it add new dependencies? Only optional deps allowed, must be lazy-imported.
-            - Will agent.py grow larger? If the change adds >50 lines to agent.py, find a way to extract instead. PRs with >100 lines added to agent.py require manual review (merge gate will not auto-merge).
+            - Will agent.py grow larger? If the change adds >50 lines to agent.py, find a way to extract instead.
            - Is there a name collision with existing exports in __init__.py?
            If ANY of these conceptual checks fail (excluding routing), add a comment to the issue explaining why and close it. Do NOT create a PR.

-            ${{ steps.check_fork.outputs.is_pr == 'true' && (steps.check_fork.outputs.is_fork != 'true' || steps.check_fork.outputs.can_push_fork == 'true') && steps.check_fork.outputs.merge_state == 'DIRTY' && format('STEP 3 — RESOLVE MERGE CONFLICTS ON PR (branch: {0}, PR #{1}{2}):\n- DO NOT create a new branch or PR\n- git fetch origin main && git rebase origin/main\n- Resolve conflicts: keep this PR intent, merge in main newer logic\n- git add -A && git rebase --continue (repeat until done)\n- git push --force-with-lease origin {0}\n- Run targeted pytest on changed paths\n- Comment files resolved and what you skipped\n', steps.check_fork.outputs.pr_branch, steps.check_fork.outputs.pr_number, steps.check_fork.outputs.can_push_fork == 'true' && format(', fork: {0}', steps.check_fork.outputs.fork_repo) || '') || '' }}
-            ${{ steps.check_fork.outputs.is_pr == 'true' && (steps.check_fork.outputs.is_fork != 'true' || steps.check_fork.outputs.can_push_fork == 'true') && steps.check_fork.outputs.merge_state != 'DIRTY' && format('STEP 3 — IMPLEMENT ON EXISTING PR (branch: {0}, PR #{1}{2}):\n- DO NOT create a new branch or PR\n- git checkout {0}\n- Implement minimal, focused fixes from reviewer feedback\n- Follow protocol-driven design: protocols in core SDK, heavy implementations in wrapper\n- Keep changes small and backward-compatible\n- Commit: git add -A && git commit -m "fix: <description>"\n- Push: git push origin {0}\n- Comment a summary of exact files modified and what you skipped\n', steps.check_fork.outputs.pr_branch, steps.check_fork.outputs.pr_number, steps.check_fork.outputs.can_push_fork == 'true' && format(', fork: {0}', steps.check_fork.outputs.fork_repo) || '') || '' }}
-            ${{ steps.check_fork.outputs.is_pr != 'true' && 'STEP 3 — IMPLEMENT (for changes in THIS repo — PraisonAI):\n- Create a fix branch: git checkout -b claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)\n- Implement a minimal, focused fix\n- Follow protocol-driven design: protocols in core SDK, heavy implementations in wrapper\n- Keep changes small and backward-compatible\n- Commit: git add -A && git commit -m "fix: <description> (fixes #$ISSUE_NUMBER)"\n- Push: git push origin claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)\nCRITICAL: After pushing, you MUST immediately run: gh pr create --title "fix: <title>" --body "Fixes #$ISSUE_NUMBER" --head claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)\nDo NOT stop after pushing. Do NOT say "manually create a PR". Always create the PR automatically.\n' || '' }}
+            STEP 3 — IMPLEMENT (for changes in THIS repo — PraisonAI):
+            - Create a fix branch: git checkout -b claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)
+            - Implement a minimal, focused fix
+            - Follow protocol-driven design: protocols in core SDK, heavy implementations in wrapper
+            - Keep changes small and backward-compatible
+            - Commit: git add -A && git commit -m "fix: <description> (fixes #$ISSUE_NUMBER)"
+            - Push: git push origin claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)
+            CRITICAL: After pushing, you MUST immediately run: gh pr create --title "fix: <title>" --body "Fixes #$ISSUE_NUMBER" --head claude/issue-$ISSUE_NUMBER-$(date +%Y%m%d)
+            Do NOT stop after pushing. Do NOT say "manually create a PR". Always create the PR automatically.

            STEP 3-ALT — IMPLEMENT IN EXTERNAL REPO (for PraisonAI-Tools, PraisonAIDocs, PraisonAI-Plugins):
            When work must happen in a different repository, follow these steps EXACTLY. Do NOT attempt to use `cd`, as directory state is not preserved. Use `git -C` for all commands.
@@ -314,7 +267,10 @@ jobs:
            - For docs changes: verify files exist and are valid markdown/MDX
            - Ensure no regressions

-            ${{ steps.check_fork.outputs.is_pr == 'true' && 'STEP 5 — SKIP: Already on an open PR. Do NOT run gh pr create.\n' || 'STEP 5 — CREATE PR:\n- Commit with descriptive message, push, and create PR using `gh pr create`\n- For external repos, use `gh pr create -R MervinPraison/<REPO_NAME>`\nCRITICAL: You MUST create the PR automatically using `gh pr create`. Do NOT just provide a link or say "manual push required".\n' }}
+            STEP 5 — CREATE PR:
+            - Commit with descriptive message, push, and create PR using `gh pr create`
+            - For external repos, use `gh pr create -R MervinPraison/<REPO_NAME>`
+            CRITICAL: You MUST create the PR automatically using `gh pr create`. Do NOT just provide a link or say "manual push required".
            NOTE: If you worked on an external repo, the wrapper action will report "No commits" for the main repo at the end. This is expected, do not worry about it.
          allowed_tools: |
            Bash(git:*)
@@ -336,60 +292,3 @@ jobs:
            mcp__github__get_issue_comments
            mcp__github__update_issue
          timeout_minutes: 30
-
-      - name: Kick review chain for bot-created PR
-        if: |
-          steps.check_fork.outputs.is_pr != 'true' &&
-          (github.event_name == 'issues' || github.event_name == 'issue_comment') &&
-          github.event.issue != null
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const chain = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/bot-pr-review-chain.js'));
-            const issueNumber = context.payload.issue.number;
-            await chain.kickReviewChainForIssue(
-              github, context.repo.owner, context.repo.repo, issueNumber, core
-            );
-
-      - name: Sync pipeline status labels for PR
-        if: steps.check_fork.outputs.is_pr == 'true'
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            const prNumber = parseInt('${{ steps.check_fork.outputs.pr_number }}', 10);
-            await ps.ensurePipelineLabels(github, context.repo.owner, context.repo.repo, core);
-            await ps.syncPipelineLabels(github, context.repo.owner, context.repo.repo, prNumber, core);
-
-      - name: Dispatch merge gate when PR pipeline ready
-        if: steps.check_fork.outputs.is_pr == 'true'
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const mergeGate = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/merge-gate.js'));
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const prNumber = parseInt('${{ steps.check_fork.outputs.pr_number }}', 10);
-            const check = await mergeGate.evaluatePipelineQuiescent(
-              github, owner, repo, prNumber, core, {
-                skipGlobalClaudeRunCheck: true,
-                skipRecentClaudeCooldown: true,
-              }
-            );
-            if (!check.ready) {
-              core.info(`PR #${prNumber} not ready for merge gate: ${check.reasons.join(', ')}`);
-              return;
-            }
-            await github.rest.repos.createDispatchEvent({
-              owner,
-              repo,
-              event_type: 'claude-merge-gate',
-              client_payload: { pr_number: prNumber },
-            });
-            core.info(`Dispatched merge gate for PR #${prNumber}`);
@@ -1,52 +0,0 @@
-name: Deploy install scripts to praison.ai
-
-# When install.sh or install.ps1 change on main, trigger praison.ai CI which
-# syncs scripts at build time and deploys to Azure AKS (existing azure-pipeline.yaml).
-
-on:
-  push:
-    branches: [main]
-    paths:
-      - 'src/praisonai/scripts/install.sh'
-      - 'src/praisonai/scripts/install.ps1'
-  workflow_dispatch:
-
-concurrency:
-  group: deploy-install-scripts
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  trigger-website-deploy:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Trigger praison.ai CI
-        env:
-          GH_TOKEN: ${{ secrets.SECRETS_ADMIN_PAT }}
-          SOURCE_SHA: ${{ github.sha }}
-        run: |
-          set -euo pipefail
-          if [ -z "${GH_TOKEN:-}" ]; then
-            echo "SECRETS_ADMIN_PAT is not configured in this repository"
-            exit 1
-          fi
-
-          gh workflow run CI \
-            --repo MervinPraison/praison.ai \
-            --ref master
-
-          echo "Triggered praison.ai CI for PraisonAI commit ${SOURCE_SHA}"
-          echo "Monitor: https://github.com/MervinPraison/praison.ai/actions/workflows/azure-pipeline.yaml"
-
-          sleep 10
-          RUN_URL=$(gh run list \
-            --repo MervinPraison/praison.ai \
-            --workflow CI \
-            --limit 1 \
-            --json url,status,createdAt \
-            --jq '.[0].url' 2>/dev/null || true)
-          if [ -n "${RUN_URL}" ]; then
-            echo "Latest praison.ai CI run: ${RUN_URL}"
-          fi
@@ -22,6 +22,8 @@ jobs:
        uses: actions/checkout@v4
        with:
          persist-credentials: false
+        with:
+          persist-credentials: false

      - name: Set up Docker Buildx
        uses: docker/setup-buildx-action@v3
@@ -36,6 +36,9 @@ jobs:
        uses: actions/checkout@v4
        with:
          persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
          token: ${{ steps.generate_token.outputs.token }}

      - name: Run Gemini Issue Triage
@@ -37,6 +37,9 @@ jobs:
        uses: actions/checkout@v4
        with:
          persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
          token: ${{ steps.generate_token.outputs.token }}

      - name: Set up Git
@@ -48,6 +48,9 @@ jobs:
        uses: actions/checkout@v4
        with:
          persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
          token: ${{ steps.generate_token.outputs.token }}
          ref: ${{ steps.checkout_ref.outputs.ref }}
          fetch-depth: 0
@@ -60,16 +60,16 @@ jobs:
    - name: Run PraisonAI integration tests
      run: |
        cd PraisonAI/src/praisonai
-        python -m pytest tests/integration/test_aiui_* -v
+        python -m pytest tests/integration/test_aiui_* -v --timeout=300
        
    - name: Run PraisonAIUI integration tests
      run: |
        cd PraisonAIUI
        if [ -f tests/integration/test_agentic_roundtrip.py ]; then
-          python -m pytest tests/integration/test_agentic_roundtrip.py -v
+          python -m pytest tests/integration/test_agentic_roundtrip.py -v --timeout=300
        fi
        if [ -f tests/test_feature_sdk_backends.py ]; then
-          python -m pytest tests/test_feature_sdk_backends.py -v
+          python -m pytest tests/test_feature_sdk_backends.py -v --timeout=300
        fi
        
    - name: Test Pattern C CLI
@@ -152,7 +152,7 @@ jobs:
      run: |
        cd PraisonAI/src/praisonai
        if [ -n "$OPENAI_API_KEY" ]; then
-          python -m pytest tests/integration/test_aiui_host_agentic.py -v
+          python -m pytest tests/integration/test_aiui_host_agentic.py -v --timeout=600
        else
          echo "OPENAI_API_KEY not available, skipping agentic tests"
        fi
@@ -1,184 +0,0 @@
-name: Auto Merge Conflict → Claude
-
-# Detects internal PRs with mergeStateStatus=DIRTY and posts @claude to trigger claude.yml.
-
-on:
-  push:
-    branches: [main]
-  pull_request:
-    types: [opened, synchronize, reopened]
-  workflow_dispatch:
-  schedule:
-    - cron: '0 */6 * * *'
-
-concurrency:
-  group: merge-conflict-scan-${{ github.repository }}
-  cancel-in-progress: true
-
-env:
-  # PAT posts as MervinPraison so issue_comment triggers claude.yml (GITHUB_TOKEN cannot chain workflows)
-  GH_TOKEN: ${{ secrets.GH_TOKEN || github.token }}
-
-permissions:
-  pull-requests: write
-  issues: write
-
-jobs:
-  detect-and-trigger:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Detect conflicting PRs and trigger Claude
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ env.GH_TOKEN }}
-          script: |
-            const owner = context.repo.owner;
-            const repo = context.repo.repo;
-            const baseRepo = `${owner}/${repo}`;
-            const COOLDOWN_MS = 12 * 60 * 60 * 1000;
-            const AUTO_ACTORS = ['github-actions[bot]', 'MervinPraison'];
-            const LABEL = 'claude-conflict-pending';
-            const COMMENT_BODY = [
-              '@claude this PR has merge conflicts with `main`.',
-              'Please rebase onto latest `main`, resolve conflicts (keep this PR\'s intent, merge in newer main logic),',
-              'run targeted tests, and force-push with `--force-with-lease`.',
-              'Comment which files you resolved.',
-              'Do not bloat the Agent class with additional params — only if absolutely required; we already support many params.',
-            ].join(' ');
-
-            const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
-
-            async function getMergeState(prNumber) {
-              const query = `
-                query($owner: String!, $repo: String!, $number: Int!) {
-                  repository(owner: $owner, name: $repo) {
-                    pullRequest(number: $number) {
-                      mergeStateStatus
-                      maintainerCanModify
-                      isDraft
-                      headRef { repository { nameWithOwner } }
-                    }
-                  }
-                }
-              `;
-              for (let attempt = 0; attempt < 3; attempt++) {
-                const result = await github.graphql(query, { owner, repo, number: prNumber });
-                const prGql = result.repository.pullRequest;
-                const status = (prGql?.mergeStateStatus || '').toUpperCase();
-                if (status && status !== 'UNKNOWN') {
-                  return {
-                    status,
-                    isDraft: prGql.isDraft,
-                    headRepo: prGql.headRef?.repository?.nameWithOwner,
-                    maintainerCanModify: prGql.maintainerCanModify === true,
-                  };
-                }
-                if (attempt < 2) await sleep(10000);
-              }
-              const pr = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
-              return {
-                status: (pr.data.mergeable_state || '').toUpperCase(),
-                isDraft: pr.data.draft,
-                headRepo: pr.data.head.repo?.full_name,
-                maintainerCanModify: pr.data.maintainer_can_modify === true,
-              };
-            }
-
-            function hasFinalClaudeReviewTrigger(comments) {
-              return comments.some((c) => {
-                const body = (c.body || '').toLowerCase();
-                if (!AUTO_ACTORS.includes(c.user.login)) return false;
-                if (!body.includes('@claude')) return false;
-                if (body.includes('merge conflict')) return false;
-                return body.includes('final architecture reviewer') || body.includes('lead engineer');
-              });
-            }
-
-            function hasRecentAutoComment(comments) {
-              const cutoff = Date.now() - COOLDOWN_MS;
-              return comments.some((c) => {
-                if (!AUTO_ACTORS.includes(c.user.login)) return false;
-                const body = (c.body || '').toLowerCase();
-                if (!body.includes('@claude') || !body.includes('merge conflict')) return false;
-                return new Date(c.created_at).getTime() > cutoff;
-              });
-            }
-
-            async function processPR(prNumber) {
-              const { status, isDraft, headRepo, maintainerCanModify } = await getMergeState(prNumber);
-              core.info(`PR #${prNumber}: mergeState=${status}, head=${headRepo}, draft=${isDraft}, maintainerCanModify=${maintainerCanModify}`);
-
-              const { data: issue } = await github.rest.issues.get({
-                owner,
-                repo,
-                issue_number: prNumber,
-              });
-              const labels = issue.labels.map((l) => l.name);
-
-              if (status !== 'DIRTY' && labels.includes(LABEL)) {
-                await github.rest.issues.removeLabel({
-                  owner,
-                  repo,
-                  issue_number: prNumber,
-                  name: LABEL,
-                });
-                core.info(`Removed ${LABEL} from PR #${prNumber}`);
-              }
-
-              if (status !== 'DIRTY') return { skipped: true, reason: 'not dirty' };
-              if (isDraft) return { skipped: true, reason: 'draft' };
-              if (headRepo && headRepo !== baseRepo && !maintainerCanModify) {
-                return { skipped: true, reason: 'fork without maintainer edits' };
-              }
-              if (labels.includes(LABEL)) {
-                return { skipped: true, reason: 'conflict resolution pending' };
-              }
-
-              const { data: comments } = await github.rest.issues.listComments({
-                owner,
-                repo,
-                issue_number: prNumber,
-                per_page: 100,
-              });
-              if (hasRecentAutoComment(comments)) {
-                return { skipped: true, reason: 'recent auto comment' };
-              }
-
-              if (!hasFinalClaudeReviewTrigger(comments)) {
-                return { skipped: true, reason: 'awaiting final claude review trigger' };
-              }
-
-              await github.rest.issues.addLabels({
-                owner,
-                repo,
-                issue_number: prNumber,
-                labels: [LABEL],
-              });
-              await github.rest.issues.createComment({
-                owner,
-                repo,
-                issue_number: prNumber,
-                body: COMMENT_BODY,
-              });
-              return { triggered: true };
-            }
-
-            let prNumbers = [];
-            if (context.eventName === 'pull_request') {
-              prNumbers = [context.payload.pull_request.number];
-            } else {
-              const prs = await github.paginate(github.rest.pulls.list, {
-                owner,
-                repo,
-                state: 'open',
-                per_page: 100,
-              });
-              prNumbers = prs.map((p) => p.number);
-            }
-
-            const results = [];
-            for (const num of prNumbers) {
-              results.push({ pr: num, ...(await processPR(num)) });
-            }
-            core.summary.addRaw('```json\n' + JSON.stringify(results, null, 2) + '\n```');
-            await core.summary.write();
@@ -1,126 +0,0 @@
-name: Nightly Release Gate
-
-# Preflight checks before dispatching PyPI Release (patch bump).
-# Runs nightly at 00:00 UTC (midnight).
-
-on:
-  schedule:
-    - cron: '0 0 * * *'
-  workflow_dispatch:
-    inputs:
-      dry_run:
-        description: 'Run preflight only; do not dispatch pypi-release.yml'
-        required: false
-        default: false
-        type: boolean
-
-permissions:
-  contents: read
-  actions: write
-
-concurrency:
-  group: nightly-release-gate
-  cancel-in-progress: false
-
-jobs:
-  preflight:
-    if: github.ref == 'refs/heads/main' || github.event_name == 'workflow_dispatch'
-    runs-on: ubuntu-latest
-    env:
-      GH_TOKEN: ${{ secrets.GH_TOKEN }}
-    steps:
-      - name: Checkout main
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5  # v4
-        with:
-          ref: main
-          fetch-depth: 0
-
-      - name: Preflight — path changes and CI green
-        id: preflight
-        run: |
-          set -euo pipefail
-
-          git fetch --tags origin
-
-          LAST_TAG=$(git describe --tags --match 'v*' --abbrev=0 2>/dev/null || true)
-          if [ -z "$LAST_TAG" ]; then
-            echo "No v* tag found — skip automated release (first release must be manual)"
-            echo "should_release=false" >> "$GITHUB_OUTPUT"
-            echo "reason=no_tag" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          CHANGED=$(git diff --name-only "${LAST_TAG}" HEAD -- \
-            src/praisonai \
-            src/praisonai-agents || true)
-
-          if [ -z "$CHANGED" ]; then
-            echo "No changes in src/praisonai or src/praisonai-agents since ${LAST_TAG} — skip"
-            echo "should_release=false" >> "$GITHUB_OUTPUT"
-            echo "reason=no_path_changes" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          echo "Changes since ${LAST_TAG}:"
-          echo "$CHANGED"
-
-          HEAD_SHA=$(git rev-parse HEAD)
-          echo "main HEAD: ${HEAD_SHA}"
-
-          GREEN_SHA=$(gh run list \
-            --repo "${GITHUB_REPOSITORY}" \
-            --workflow "Core Tests" \
-            --branch main \
-            --status success \
-            --json headSha \
-            --jq '.[0].headSha // empty')
-
-          if [ "$GREEN_SHA" != "$HEAD_SHA" ]; then
-            echo "main@${HEAD_SHA} is not covered by a green Core Tests run (last green: ${GREEN_SHA:-none}) — skip"
-            echo "should_release=false" >> "$GITHUB_OUTPUT"
-            echo "reason=ci_not_green" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          IN_PROGRESS=$(gh run list \
-            --repo "${GITHUB_REPOSITORY}" \
-            --workflow "Core Tests" \
-            --branch main \
-            --json headSha,status \
-            --jq "[.[] | select(.headSha == \"${HEAD_SHA}\" and .status == \"in_progress\")] | length")
-
-          if [ "${IN_PROGRESS:-0}" != "0" ]; then
-            echo "Core Tests still in progress on ${HEAD_SHA} — skip"
-            echo "should_release=false" >> "$GITHUB_OUTPUT"
-            echo "reason=ci_in_progress" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          echo "Preflight passed — ready to dispatch patch release"
-          echo "should_release=true" >> "$GITHUB_OUTPUT"
-          echo "reason=ready" >> "$GITHUB_OUTPUT"
-          echo "last_tag=${LAST_TAG}" >> "$GITHUB_OUTPUT"
-          echo "head_sha=${HEAD_SHA}" >> "$GITHUB_OUTPUT"
-
-      - name: Dry run summary
-        if: inputs.dry_run == true
-        run: |
-          echo "Dry run — preflight only, no pypi-release dispatch."
-          echo "should_release=${{ steps.preflight.outputs.should_release }}"
-          echo "reason=${{ steps.preflight.outputs.reason }}"
-          if [ "${{ steps.preflight.outputs.should_release }}" = "true" ]; then
-            echo "Would dispatch: pypi-release.yml bump=patch source=nightly"
-          fi
-
-      - name: Dispatch PyPI Release (patch)
-        if: steps.preflight.outputs.should_release == 'true' && inputs.dry_run != true
-        run: |
-          set -euo pipefail
-          gh workflow run pypi-release.yml \
-            --repo "${GITHUB_REPOSITORY}" \
-            --ref main \
-            -f bump=patch \
-            -f source=nightly \
-            -f dry_run=false
-          echo "Dispatched pypi-release.yml (patch, source=nightly)"
-          echo "Approve the pypi environment when prompted."
@@ -1,31 +0,0 @@
-name: Pipeline Status Sync
-
-# Sync pipeline/* stage and blocker labels on open internal PRs.
-
-on:
-  schedule:
-    - cron: '*/15 * * * *'
-  workflow_dispatch:
-
-permissions:
-  pull-requests: read
-  issues: write
-
-jobs:
-  sync:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
-
-      - name: Sync pipeline labels on open PRs
-        uses: actions/github-script@v7
-        with:
-          github-token: ${{ secrets.GH_TOKEN }}
-          script: |
-            const path = require('path');
-            const ps = require(path.join(process.env.GITHUB_WORKSPACE, '.github/scripts/pipeline-status.js'));
-            await ps.syncOpenPullRequests(
-              github, context.repo.owner, context.repo.repo, { maxPrs: 30 }, core
-            );
@@ -1,305 +0,0 @@
-name: PyPI Release
-
-# Requires a GitHub Environment named "pypi" (Settings → Environments → pypi)
-# with required reviewers before secrets are exposed to this job.
-
-on:
-  workflow_dispatch:
-    inputs:
-      bump:
-        description: 'Version bump type for both packages'
-        required: true
-        default: patch
-        type: choice
-        options:
-          - patch
-          - minor
-          - major
-      dry_run:
-        description: 'Preview versions and steps without publishing or pushing'
-        required: false
-        default: false
-        type: boolean
-      skip_agents:
-        description: 'Skip agents publish (recovery if already on PyPI)'
-        required: false
-        default: false
-        type: boolean
-      skip_wrapper:
-        description: 'Publish agents only'
-        required: false
-        default: false
-        type: boolean
-      agents_version:
-        description: 'Override agents version (empty = auto bump from pyproject.toml)'
-        required: false
-        default: ''
-        type: string
-      wrapper_version:
-        description: 'Override wrapper version (empty = auto bump from version.py)'
-        required: false
-        default: ''
-        type: string
-      source:
-        description: 'Trigger source (audit trail)'
-        required: false
-        default: manual
-        type: choice
-        options:
-          - manual
-          - nightly
-
-permissions:
-  contents: write
-
-concurrency:
-  group: pypi-release
-  cancel-in-progress: false
-
-jobs:
-  release:
-    if: github.ref == 'refs/heads/main'
-    runs-on: ubuntu-latest
-    environment: pypi
-    env:
-      UV_PUBLISH_TOKEN: ${{ secrets.PYPI_TOKEN || secrets.PYPI_API_TOKEN }}
-      GH_TOKEN: ${{ secrets.GH_TOKEN }}
-    steps:
-      - name: Checkout main
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5  # v4
-        with:
-          ref: main
-          fetch-depth: 0
-          token: ${{ secrets.GH_TOKEN }}
-
-      - name: Set up Python
-        uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065  # v5
-        with:
-          python-version: '3.11'
-
-      - name: Set up uv
-        uses: astral-sh/setup-uv@38f3f104447c67c051c4a08e39b64a148898af3a  # v4
-
-      - name: Configure git
-        run: |
-          git config user.name "MervinPraison"
-          git config user.email "[email protected]"
-          gh auth setup-git
-
-      - name: Release summary
-        run: |
-          echo "Source: ${{ inputs.source }}"
-          echo "Bump:   ${{ inputs.bump }}"
-          echo "Dry run: ${{ inputs.dry_run }}"
-
-      - name: Compute release versions
-        id: versions
-        env:
-          BUMP: ${{ inputs.bump }}
-          AGENTS_OVERRIDE: ${{ inputs.agents_version }}
-          WRAPPER_OVERRIDE: ${{ inputs.wrapper_version }}
-          SKIP_AGENTS: ${{ inputs.skip_agents }}
-        run: |
-          python <<'PY'
-          import os
-          import re
-          import sys
-          from pathlib import Path
-
-          SEMVER = re.compile(r"^\d+\.\d+\.\d+$")
-          root = Path(".")
-          bump = os.environ["BUMP"]
-
-          def bump_version(version: str) -> str:
-              parts = version.split(".")
-              if len(parts) != 3:
-                  print(f"Invalid version: {version}", file=sys.stderr)
-                  sys.exit(1)
-              maj, min_, patch = (int(parts[0]), int(parts[1]), int(parts[2]))
-              if bump == "major":
-                  return f"{maj + 1}.0.0"
-              if bump == "minor":
-                  return f"{maj}.{min_ + 1}.0"
-              return f"{maj}.{min_}.{patch + 1}"
-
-          def validate_version(label: str, value: str) -> None:
-              if value and not SEMVER.match(value):
-                  print(f"Invalid {label} override: {value!r} (expected X.Y.Z)", file=sys.stderr)
-                  sys.exit(1)
-
-          agents_override = os.environ.get("AGENTS_OVERRIDE", "").strip()
-          wrapper_override = os.environ.get("WRAPPER_OVERRIDE", "").strip()
-          validate_version("agents_version", agents_override)
-          validate_version("wrapper_version", wrapper_override)
-
-          agents_pyproject = root / "src/praisonai-agents/pyproject.toml"
-          agents_content = agents_pyproject.read_text()
-          agents_match = re.search(r'^version\s*=\s*"([^"]+)"', agents_content, re.MULTILINE)
-          if not agents_match:
-              print("Could not read agents version from pyproject.toml", file=sys.stderr)
-              sys.exit(1)
-          current_agents = agents_match.group(1)
-          if not SEMVER.match(current_agents):
-              print(f"Invalid current agents version: {current_agents}", file=sys.stderr)
-              sys.exit(1)
-
-          wrapper_version_py = root / "src/praisonai/praisonai/version.py"
-          wrapper_content = wrapper_version_py.read_text()
-          wrapper_match = re.search(r'__version__ = "([^"]+)"', wrapper_content)
-          if not wrapper_match:
-              print("Could not read wrapper version from version.py", file=sys.stderr)
-              sys.exit(1)
-          current_wrapper = wrapper_match.group(1)
-          if not SEMVER.match(current_wrapper):
-              print(f"Invalid current wrapper version: {current_wrapper}", file=sys.stderr)
-              sys.exit(1)
-
-          skip_agents = os.environ.get("SKIP_AGENTS", "false").lower() == "true"
-
-          if agents_override:
-              agents_version = agents_override
-          elif skip_agents:
-              agents_version = current_agents
-          else:
-              agents_version = bump_version(current_agents)
-
-          if wrapper_override:
-              wrapper_version = wrapper_override
-          else:
-              wrapper_version = bump_version(current_wrapper)
-
-          github_output = os.environ["GITHUB_OUTPUT"]
-          with open(github_output, "a", encoding="utf-8") as fh:
-              fh.write(f"agents_version={agents_version}\n")
-              fh.write(f"wrapper_version={wrapper_version}\n")
-              fh.write(f"current_agents={current_agents}\n")
-              fh.write(f"current_wrapper={current_wrapper}\n")
-
-          print(f"Agents:  {current_agents} -> {agents_version}")
-          print(f"Wrapper: {current_wrapper} -> {wrapper_version}")
-          PY
-
-      - name: Dry run summary
-        if: inputs.dry_run
-        run: |
-          echo "Dry run — no publish, commit, or push."
-          echo "Agents version:  ${{ steps.versions.outputs.agents_version }}"
-          echo "Wrapper version: ${{ steps.versions.outputs.wrapper_version }}"
-          echo ""
-          if [ "${{ inputs.skip_agents }}" != "true" ]; then
-            echo "Would publish praisonaiagents ${{ steps.versions.outputs.agents_version }}"
-            echo "Would commit src/praisonai-agents/pyproject.toml and uv.lock"
-          fi
-          if [ "${{ inputs.skip_wrapper }}" != "true" ]; then
-            echo "Would run bump_and_release.py ${{ steps.versions.outputs.wrapper_version }} --agents ${{ steps.versions.outputs.agents_version }}"
-            echo "Would uv publish praisonai ${{ steps.versions.outputs.wrapper_version }}"
-          fi
-
-      - name: Publish praisonaiagents
-        if: inputs.dry_run != true && inputs.skip_agents != true
-        working-directory: src/praisonai-agents
-        env:
-          NEW_VERSION: ${{ steps.versions.outputs.agents_version }}
-          CURRENT: ${{ steps.versions.outputs.current_agents }}
-        run: |
-          set -euo pipefail
-          python <<'PY'
-          import os
-          from pathlib import Path
-          path = Path("pyproject.toml")
-          content = path.read_text()
-          current = os.environ["CURRENT"]
-          new = os.environ["NEW_VERSION"]
-          path.write_text(content.replace(f'version = "{current}"', f'version = "{new}"', 1))
-          PY
-
-          rm -rf dist
-          uv lock
-          uv build
-          uv publish
-
-      - name: Wait for praisonaiagents on PyPI
-        if: inputs.dry_run != true && inputs.skip_agents != true
-        run: |
-          set -euo pipefail
-          PACKAGE="praisonaiagents"
-          VERSION="${{ steps.versions.outputs.agents_version }}"
-          MAX_WAIT=600
-          INTERVAL=30
-          START=$(date +%s)
-
-          while true; do
-            if curl -fsSL "https://pypi.org/pypi/${PACKAGE}/${VERSION}/json" >/dev/null; then
-              echo "✅ ${PACKAGE}==${VERSION} is available on PyPI"
-              exit 0
-            fi
-            ELAPSED=$(( $(date +%s) - START ))
-            if [ "$ELAPSED" -ge "$MAX_WAIT" ]; then
-              echo "❌ Timeout waiting for ${PACKAGE}==${VERSION} on PyPI"
-              exit 1
-            fi
-            echo "⏳ Waiting for ${PACKAGE}==${VERSION} ($(( MAX_WAIT - ELAPSED ))s remaining...)"
-            sleep "$INTERVAL"
-          done
-
-      - name: Commit agents version bump
-        if: inputs.dry_run != true && inputs.skip_agents != true
-        run: |
-          set -euo pipefail
-          git add src/praisonai-agents/pyproject.toml src/praisonai-agents/uv.lock
-          git diff --cached --quiet && { echo "No agents files to commit"; exit 0; }
-          git commit -m "Bump praisonaiagents to ${{ steps.versions.outputs.agents_version }}"
-          git pull --rebase origin main
-          git push origin main
-
-      - name: Bump and release wrapper
-        if: inputs.dry_run != true && inputs.skip_wrapper != true
-        run: |
-          set -euo pipefail
-          python src/praisonai/scripts/bump_and_release.py "${{ steps.versions.outputs.wrapper_version }}" \
-            --agents "${{ steps.versions.outputs.agents_version }}" \
-            --wait --max-wait 600 --no-add-all
-
-      - name: Publish praisonai wrapper to PyPI
-        if: inputs.dry_run != true && inputs.skip_wrapper != true
-        working-directory: src/praisonai
-        run: uv publish
-
-      - name: Wait for praisonai on PyPI
-        if: inputs.dry_run != true && inputs.skip_wrapper != true
-        run: |
-          set -euo pipefail
-          PACKAGE="praisonai"
-          VERSION="${{ steps.versions.outputs.wrapper_version }}"
-          MAX_WAIT=600
-          INTERVAL=30
-          START=$(date +%s)
-
-          while true; do
-            if curl -fsSL "https://pypi.org/pypi/${PACKAGE}/${VERSION}/json" >/dev/null; then
-              echo "✅ ${PACKAGE}==${VERSION} is available on PyPI"
-              exit 0
-            fi
-            ELAPSED=$(( $(date +%s) - START ))
-            if [ "$ELAPSED" -ge "$MAX_WAIT" ]; then
-              echo "❌ Timeout waiting for ${PACKAGE}==${VERSION} on PyPI"
-              exit 1
-            fi
-            echo "⏳ Waiting for ${PACKAGE}==${VERSION} ($(( MAX_WAIT - ELAPSED ))s remaining...)"
-            sleep "$INTERVAL"
-          done
-
-      - name: Verify PyPI versions
-        if: inputs.dry_run != true
-        run: |
-          set -euo pipefail
-          AGENTS="${{ steps.versions.outputs.agents_version }}"
-          WRAPPER="${{ steps.versions.outputs.wrapper_version }}"
-
-          if [ "${{ inputs.skip_agents }}" != "true" ]; then
-            echo "✅ praisonaiagents==${AGENTS}"
-          fi
-
-          if [ "${{ inputs.skip_wrapper }}" != "true" ]; then
-            echo "✅ praisonai==${WRAPPER}"
-          fi
@@ -15,8 +15,8 @@ jobs:

    steps:
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
      with:
@@ -8,18 +8,15 @@

 name: Upload Python Package

-# Superseded by pypi-release.yml (manual workflow_dispatch with uv publish).
-# Kept as workflow_dispatch-only fallback; release trigger removed to avoid
-# duplicate PyPI publish when bump_and_release creates a GitHub release.
 on:
-  workflow_dispatch:
+  release:
+    types: [published]

 permissions:
  contents: read

 jobs:
  deploy:
-    if: false

    runs-on: ubuntu-latest

@@ -26,8 +26,11 @@ jobs:
    # First checkout the upstream repo at the specified tag
    - name: Checkout upstream repo
      uses: actions/checkout@v4
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false
      with:
-        persist-credentials: false
        repository: MervinPraison/PraisonAI
        ref: ${{ github.event.inputs.tag }}
        fetch-depth: 0
@@ -45,8 +48,11 @@ jobs:
    # Now checkout the fork at main branch
    - name: Checkout fork
      uses: actions/checkout@v4
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false
      with:
-        persist-credentials: false
        repository: ${{ github.event.inputs.fork_owner }}/PraisonAI
        ref: main
        fetch-depth: 0
@@ -443,8 +449,11 @@ jobs:
    # First checkout the fork's main branch
    - name: Checkout fork repo
      uses: actions/checkout@v4
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false
      with:
-        persist-credentials: false
        fetch-depth: 0
        token: ${{ secrets.PAT_TOKEN }}
        repository: ${{ github.event.inputs.fork_owner }}/PraisonAI
@@ -1,31 +0,0 @@
-name: Sync Claude Secrets to PraisonAIUI
-
-on:
-  workflow_dispatch:
-
-permissions:
-  contents: read
-
-jobs:
-  sync:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Sync secrets to PraisonAIUI
-        env:
-          GH_TOKEN: ${{ secrets.SECRETS_ADMIN_PAT }}
-          TARGET: MervinPraison/PraisonAIUI
-          CLAUDE_APP_ID: ${{ secrets.CLAUDE_APP_ID }}
-          CLAUDE_APP_PRIVATE_KEY: ${{ secrets.CLAUDE_APP_PRIVATE_KEY }}
-          CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
-          GH_TOKEN_VALUE: ${{ secrets.GH_TOKEN }}
-        run: |
-          set -euo pipefail
-          if [ -z "${GH_TOKEN:-}" ]; then
-            echo "SECRETS_ADMIN_PAT is not configured in this repository"
-            exit 1
-          fi
-          echo -n "$CLAUDE_APP_ID" | gh secret set CLAUDE_APP_ID --repo "$TARGET"
-          echo -n "$CLAUDE_APP_PRIVATE_KEY" | gh secret set CLAUDE_APP_PRIVATE_KEY --repo "$TARGET"
-          echo -n "$CLAUDE_CODE_OAUTH_TOKEN" | gh secret set CLAUDE_CODE_OAUTH_TOKEN --repo "$TARGET"
-          echo -n "$GH_TOKEN_VALUE" | gh secret set GH_TOKEN --repo "$TARGET"
-          echo "Synced CLAUDE_APP_ID, CLAUDE_APP_PRIVATE_KEY, CLAUDE_CODE_OAUTH_TOKEN, GH_TOKEN → $TARGET"
@@ -37,8 +37,10 @@ jobs:
    steps:
    - name: Checkout code
      uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false

    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
@@ -22,8 +22,8 @@ jobs:

    steps:
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
    - name: Set up Python 3.11
      uses: actions/setup-python@v5
      with:
@@ -13,8 +13,10 @@ jobs:
    steps:
    - name: Checkout code
      uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false

    - name: Set up Python
      uses: actions/setup-python@v5
@@ -79,8 +81,10 @@ jobs:
    steps:
    - name: Checkout code
      uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false

    - name: Set up Python
      uses: actions/setup-python@v5
@@ -33,8 +33,10 @@ jobs:
    steps:
    - name: Checkout code
      uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false

    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
@@ -45,8 +45,8 @@ jobs:
    timeout-minutes: 5
    steps:
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
    - name: Set up Python 3.11
      uses: actions/setup-python@v5
      with:
@@ -105,8 +105,8 @@ jobs:
    
    steps:
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
      with:
@@ -187,8 +187,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -259,8 +259,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -318,8 +318,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -377,8 +377,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -436,8 +436,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -495,8 +495,8 @@ jobs:
        fi
    
    - uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
      if: steps.check-key.outputs.has_key == 'true'
    
    - name: Set up Python 3.11
@@ -45,8 +45,10 @@ jobs:
        
    - name: Checkout code
      uses: actions/checkout@v4
-      with:
-        persist-credentials: false
+        with:
+          persist-credentials: false
+        with:
+          persist-credentials: false

    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
@@ -88,9 +88,6 @@ issue_details.json
 .internal
 .praisonai

-# Backup files
-*.bak
-
 # Rust / Cargo
 target/
 Cargo.lock
@@ -1,10 +0,0 @@
-# Agent Instructions
-
-You are working on the PraisonAI project.
-
-## Project Guidelines
- Follow the existing code style and conventions
- Be concise and helpful in responses  
- Test implementation thoroughly
- Ensure backward compatibility with existing APIs
- Follow protocol-driven design: core protocols in praisonaiagents/, heavy implementations in praisonai/
@@ -1,437 +0,0 @@
-# PraisonAI Architecture
-
-> **Last updated:** 2026-06-24
->
-> Strategic architecture document for PraisonAI — a multi-agent AI framework.
-> Covers system design, runtime architecture, data contracts, reliability,
-> observability, and the 2-quarter road map.
-
---
-
-## Table of Contents
-
-1. [Executive Summary](#1-executive-summary)
-2. [System Overview](#2-system-overview)
-3. [Layered Architecture](#3-layered-architecture)
-4. [Core Data Contracts](#4-core-data-contracts)
-5. [Reliability by Design](#5-reliability-by-design)
-6. [Observability & Telemetry](#6-observability--telemetry)
-7. [Replay & Checkpointing](#7-replay--checkpointing)
-8. [Implementation Roadmap](#8-implementation-roadmap)
-9. [Success Metrics](#9-success-metrics)
-
---
-
-## 1. Executive Summary
-
-PraisonAI is a multi-agent AI framework with broad capability surface across
-Python, TypeScript, and Rust SDKs. The framework's core strength is its
-feature breadth — agent abstractions, integrations, workflows, and active
-release cadence.
-
-The strategic bet for the next two quarters is building a **Reliability + 
-Orchestration + Observability** core to unlock adoption and trust:
-
-| Dimension | Current State | Target |
-|-----------|--------------|--------|
-| Reliability | Cross-platform mismatches, optional dep fragility | Deterministic gating, adapter abstraction, parity CI |
-| Determinism | Flaky tests, global state collisions | Isolated test fixtures, traceable tenant IDs |
-| Developer UX | Complex onboarding, implicit config | Golden-path CLI, Doctor Auto-Fix |
-| Observability | Minimal runtime tracing | Structured event bus, replay engine, failure classifier |
-
-### Design Principles
-
- **Deterministic by default**, flexible by opt-in
- **Policy-enforced execution boundaries** at every layer
- **Structured events** for all lifecycle transitions
- **Capability isolation** per agent and per tool
-
---
-
-## 2. System Overview
-
-```
-User/SDK/CLI
-  → Workflow Compiler (graph + policy + schemas)
-    → Execution Orchestrator (state machine, retries, fallbacks)
-      → Tool Runtime (sandbox + approval + capability scope)
-      → Model Runtime (provider abstraction + rate/timeout policy)
-    → Observability Bus (trace events + metrics + cost)
-    → Replay Engine (checkpoint + deterministic re-run)
-    → Persistence (session, memory, artifacts, run ledger)
-```
-
-### Runtime Components
-
-| Component | Responsibility | Interface |
-|-----------|---------------|-----------|
-| API/CLI Gateway | Accepts run requests, validates config/profile | `RunRequest`, `RunProfile` |
-| Workflow Compiler | Converts YAML/DSL to normalized execution graph | `CompiledGraph` |
-| Orchestrator | Executes graph state machine with retries/fallbacks | `ExecutionState`, `StepTransition` |
-| Tool Runtime | Runs tools with approval/sandbox policies | `ToolCall`, `ToolResult` |
-| Model Runtime | Provider routing, timeout, budget control | `ModelRequest`, `ModelResponse` |
-| Observability Bus | Streams structured events + metrics | `RunEvent` schema |
-| Replay Engine | Restarts from checkpoint with deterministic inputs | `ReplayRequest` |
-
-### Existing Architecture (Python SDK)
-
-The Python SDK is the primary runtime and is organised as follows:
-
- `src/praisonai/` — Core runtime and agent abstractions
-  - `agents/` — Agent types (chat, code, realtime, audio, vision, etc.)
-  - `llm/` — Model runtime with provider routing, rate limiting, failover
-  - `tools/` — Tool runtime with sandbox, approval, retry
-  - `memory/` — Memory runtime (in-memory, SQLite, MongoDB, Mem0 adapters)
-  - `knowledge/` — Knowledge management (indexing, retrieval, chunking)
-  - `workflow/` — Workflow engine (YAML/SDK-based orchestration)
-  - `telemetry/` — Observability, monitoring, token tracking
-  - `reliability/` — Error classification, circuit breaker, failover
-  - `skill/` — Plugin/skill system
-  - `loop_detection/` — Doom loop detection and prevention
-  - `mcp/` — MCP protocol support
-  - `a2a/`, `a2ui/` — Agent-to-agent and agent-to-UI protocols
-
-### Multi-SDK Layout
-
- **Python SDK** — Primary runtime (`src/praisonai/`)
- **TypeScript SDK** — JS/TS runtime (`ts-sdk/`)
- **Rust SDK** — High-performance Rust runtime (`rust-sdk/`)
- **CLI** — Command-line interface (`cli/`)
- **UI** — Web UI (`ui/`)
-
---
-
-## 3. Layered Architecture
-
-```mermaid
-flowchart TB
-  subgraph UX["Interface Layer"]
-    CLI[CLI]
-    SDK[SDK]
-    API[API Gateway]
-  end
-
-  subgraph Control["Control Plane"]
-    Compiler[Workflow Compiler]
-    Policy[Policy Engine]
-    Orchestrator[Execution Orchestrator]
-  end
-
-  subgraph Runtime["Runtime Plane"]
-    ModelRT[Model Runtime]
-    ToolRT[Tool Runtime]
-    MemoryRT[Memory Runtime]
-  end
-
-  subgraph Observe["Observability Plane"]
-    EventBus[Run Event Bus]
-    Metrics[Metrics + Cost]
-    Replay[Replay Engine]
-  end
-
-  subgraph Data["Data Plane"]
-    Ledger[Run Ledger]
-    Checkpoints[Checkpoints]
-    Artifacts[Artifacts Store]
-  end
-
-  CLI --> API
-  SDK --> API
-  API --> Compiler
-  API --> Policy
-  Compiler --> Orchestrator
-  Policy --> Orchestrator
-  Orchestrator --> ModelRT
-  Orchestrator --> ToolRT
-  Orchestrator --> MemoryRT
-  Orchestrator --> EventBus
-  EventBus --> Metrics
-  EventBus --> Replay
-  Orchestrator --> Ledger
-  Orchestrator --> Checkpoints
-  ToolRT --> Artifacts
-  Replay --> Checkpoints
-```
-
-### Interface Layer
-
- **CLI** — The `praisonai` command-line entry point
- **SDK** — Python library API (`from praisonai import Agent`)
- **API Gateway** — HTTP/WebSocket REST API (`praisonai api` or `a2a`/`a2ui`)
-
-### Control Plane
-
- **Workflow Compiler** — Converts YAML workflows and SDK-defined graphs into
-  a normalized `CompiledGraph` with adjacency validation and cycle detection.
- **Policy Engine** — Enforces runtime policies (budget, timeout, approval
-  gates, capability validation) before and during execution.
- **Execution Orchestrator** — Drives the state machine through graph nodes,
-  handling retry, fallback, and failure classification.
-
-### Runtime Plane
-
- **Model Runtime** — Provider abstraction layer. Routes to OpenAI, Anthropic,
-  Gemini, Ollama, DeepSeek, etc. Handles rate limiting, token tracking, and
-  automatic failover between providers.
- **Tool Runtime** — Executes tool calls with configurable sandboxing,
-  approval gates, retry policies, and capability scope validation.
- **Memory Runtime** — Manages session memory, persistent memory stores,
-  and knowledge retrieval across in-memory, SQLite, and MongoDB backends.
-
-### Observability Plane
-
- **Run Event Bus** — Structured event streaming for all lifecycle transitions
-  (START, INPUT, MODEL_CALL, TOOL_CALL, ERROR, RETRY, COMPLETE).
- **Metrics + Cost** — Token counting, cost tracking, and performance metrics.
- **Replay Engine** — Deterministic replay from the last stable checkpoint,
-  with state hash verification.
-
-### Data Plane
-
- **Run Ledger** — Persistent record of all runs with state, metrics, and
-  failure classifications.
- **Checkpoints** — Point-in-time snapshots of run state, memory, and
-  artifacts for replay and recovery.
- **Artifacts Store** — Tool outputs, generated files, and intermediate
-  results.
-
---
-
-## 4. Core Data Contracts
-
-### RunRequest
-
-```text
-RunRequest {
-  run_id: UUID,
-  workflow_ref: string,
-  inputs: object,
-  profile: {safe_mode, budget, timeout, approval_policy},
-  context: {user_id, workspace_id, env}
-}
-```
-
-### RunEvent
-
-```text
-RunEvent {
-  ts: ISO8601,
-  run_id: UUID,
-  step_id: string,
-  type: START | INPUT | MODEL_CALL | TOOL_CALL |
-        ERROR | RETRY | COMPLETE,
-  payload: object,
-  cost: {tokens_in, tokens_out, usd},
-  latency_ms: number
-}
-```
-
-### Checkpoint
-
-```text
-Checkpoint {
-  run_id: UUID,
-  step_id: string,
-  state_hash: string,
-  memory_snapshot_ref: string,
-  artifact_refs: string[]
-}
-```
-
-### Execution Sequence
-
-```text
-1) Client submits RunRequest
-2) Gateway validates profile + schema
-3) Compiler resolves workflow → CompiledGraph
-4) Orchestrator starts node execution
-5) Node may call Model Runtime or Tool Runtime
-6) Every transition emits RunEvent
-7) On failure: classifier decides retry/fallback/abort
-8) Checkpoint stored after each critical step
-9) Final state + artifacts persisted to run ledger
-10) Replay can resume from last stable checkpoint
-```
-
---
-
-## 5. Reliability by Design
-
-### Error Taxonomy
-
-All failures are classified into one of five categories for automatic
-remediation:
-
-| Category | Description | Default Action |
-|----------|-------------|----------------|
-| `config` | Invalid configuration or profile | Abort with remediation hint |
-| `dependency` | Missing optional module or capability | Degrade / skip optional path |
-| `tool` | Tool execution failure | Retry with backoff and policy cap |
-| `model` | Model API error or timeout | Fallback to alternate provider/route |
-| `infra` | Network or resource exhaustion | Circuit-breaker + replay from checkpoint |
-
-```mermaid
-flowchart TD
-  S[Step Failure] --> T{Failure Type?}
-  T -->|Config| C1[Abort + config remediation hint]
-  T -->|Dependency| C2[Degrade capability / skip optional path]
-  T -->|Tool| C3[Retry with backoff and policy cap]
-  T -->|Model| C4[Fallback model/provider route]
-  T -->|Infra| C5[Circuit-breaker + replay from checkpoint]
-```
-
-### Hardening Checklist
-
- **Dependency gates** — Optional modules declared as capabilities;
-  unavailable capability results in skip/degrade, not crash.
- **Cross-platform adapters** — OS-specific implementations behind a single
-  interface (e.g., file lock adapter for Windows/Linux/macOS).
- **Idempotent fixtures** — Integration tests use unique tenant/workspace IDs
-  and guaranteed teardown.
- **Fail-safe output mode** — CLI rendering falls back to ASCII-safe mode on
-  encoding mismatch.
- **CI parity matrix** — Run smoke workflows across Windows, Linux, and macOS.
-
-### Deterministic Test Pattern
-
-```python
-# Integration test pattern — always use unique, traceable IDs
-import uuid
-
-def test_agent_workflow():
-    tenant_id = f"test-{uuid.uuid4().hex[:8]}"
-    agent = Agent(name=f"agent-{tenant_id}", ...)
-    result = agent.run("task")
-    assert result.status == "success"
-    # teardown is guaranteed via fixture or context manager
-```
-
---
-
-## 6. Observability & Telemetry
-
-### Current Telemetry Stack
-
-PraisonAI already includes:
-
- **OpenTelemetry integration** — Manual and auto-instrumentation for
-  traces, metrics, and logs.
- **Token tracking** — Per-call and cumulative token/cost tracking.
- **Performance monitoring** — Real-time dashboards and monitoring views.
- **LangTrace integration** — LangTrace provider for tracing.
- **Run outcomes** — Structured `RunOutcome` objects with status, duration,
-  token usage, and error classification.
-
-### Planned Enhancements
-
- **Run Event Bus** — Stream all lifecycle events as structured `RunEvent`
-  payloads.
- **Failure classifier** — Automatic failure category detection with
-  remediation hints.
- **Run ledger visualization** — Timeline view of runs with step-by-step
-  event drill-down.
- **Replay integration** — Checkpoint-based run replay from the
-  observability dashboard.
-
---
-
-## 7. Replay & Checkpointing
-
-### Design
-
-```mermaid
-sequenceDiagram
-  participant U as User/Client
-  participant G as Gateway
-  participant C as Compiler
-  participant O as Orchestrator
-  participant R as Model/Tool Runtime
-  participant E as Event Bus
-  participant L as Run Ledger
-  participant K as Checkpoint Store
-
-  U->>G: Submit RunRequest
-  G->>C: Validate + compile workflow
-  C-->>O: CompiledGraph
-  O->>R: Execute next step
-  R-->>O: Step output / error
-  O->>E: Emit structured RunEvent
-  O->>K: Persist checkpoint
-  O->>L: Persist run state + metrics
-  O-->>U: Final result / failure summary
-```
-
-### Current State
-
- Basic run outcome persistence exists
- Session persistence for bot/multi-turn conversations
- Token and cost tracking per run
-
-### Roadmap
-
- Full checkpoint creation at configurable step granularity
- State hash verification for deterministic replay
- Replay API endpoint (`POST /runs/{id}/replay`)
- Checkpoint pruning and retention policy
-
---
-
-## 8. Implementation Roadmap
-
-### Quarter 1: Trust Foundation (Q3 2026)
-
-| Priority | Project | Description | Status |
-|----------|---------|-------------|--------|
-| P0 | Reliability Core | Cross-platform hardening, optional-dep gating, deterministic fixtures | In progress |
-| P0 | Doctor Auto-Fix | Automated environment diagnosis and repairs | Planned |
-| P1 | Golden-Path CLI | Single canonical flow: init → run → test → deploy | Planned |
-| P1 | CI Parity Matrix | Cross-platform smoke tests in CI pipeline | Planned |
-
-### Quarter 2: Operational Excellence (Q4 2026)
-
-| Priority | Project | Description | Status |
-|----------|---------|-------------|--------|
-| P1 | Trace + Replay | Run timeline, checkpoint replay, failure classifier | Planned |
-| P1 | Failure Classifier | Automatic failure category detection with remediation | Planned |
-| P2 | Graph Studio UX | Visual deterministic orchestration editor/inspector | Planned |
-| P2 | Safe Production Profiles | Policy presets: guardrails, approvals, cost/time caps | Planned |
-
-```mermaid
-gantt
-    title PraisonAI Architecture Program (2 Quarters)
-    dateFormat  YYYY-MM-DD
-    section Quarter 1
-    Reliability hardening + parity CI       :a1, 2026-07-01, 45d
-    Deterministic test isolation            :a2, 2026-07-10, 50d
-    Doctor Auto-Fix MVP                     :a3, 2026-07-20, 40d
-    Golden-path CLI                         :a4, 2026-08-01, 45d
-    section Quarter 2
-    Trace bus + run ledger                  :b1, 2026-10-01, 45d
-    Replay engine                           :b2, 2026-10-15, 45d
-    Failure classifier                      :b3, 2026-11-01, 30d
-    Graph UX beta + Safe profiles           :b4, 2026-11-10, 50d
-```
-
---
-
-## 9. Success Metrics
-
-### KPI Targets
-
-| KPI | Target | Primary Workstream |
-|-----|--------|-------------------|
-| Time-to-first-successful run | 30-40% reduction | Golden-path CLI + Doctor Auto-Fix |
-| Cross-platform issue rate | 50% reduction | Reliability Core + adapter layer |
-| Incident triage time | 40% reduction | Trace + Replay + failure classifier |
-| CI confidence | Flaky test rate <2% | Deterministic tests + fixture isolation |
-
-### Governance
-
- **Architecture Review Council** — Reviews runtime-contract changes
- **Release Quality Gate** — Platform matrix + deterministic test thresholds
- **Monthly telemetry review** — DX, reliability, adoption trends
-
-
---
-
-*This document is a living reference. Update as the architecture evolves.*
@@ -770,11 +770,11 @@ agent.start('Write a movie script about a robot in Mars');

 ## ⚡ Performance

-PraisonAI is built for speed, with agent instantiation in around 14μs. This reduces overhead, improves responsiveness, and helps multi-agent systems scale efficiently in real-world production workloads.
+PraisonAI is built for speed, with agent instantiation in under 4μs. This reduces overhead, improves responsiveness, and helps multi-agent systems scale efficiently in real-world production workloads.

 | Performance Metric | PraisonAI |
 |--------------------|-----------|
-| Avg Instantiation Time | **14 μs** |
+| Avg Instantiation Time | **3.77 μs** |

 ---

@@ -1,66 +0,0 @@
-# Security Policy
-
-PraisonAI takes security seriously. We appreciate responsible disclosure from researchers and users.
-
-## Supported packages
-
-| Package | Registry | Source path |
-|---------|----------|-------------|
-| `praisonaiagents` | [PyPI](https://pypi.org/project/praisonaiagents/) | `src/praisonai-agents/` |
-| `praisonai` | [PyPI](https://pypi.org/project/praisonai/) | `src/praisonai/` |
-| `praisonai-platform` | [PyPI](https://pypi.org/project/praisonai-platform/) | `src/praisonai-platform/` |
-| `praisonai` | [npm](https://www.npmjs.com/package/praisonai) | `src/praisonai-ts/` |
-
-Report issues against the **specific package** where the vulnerability exists. A platform-only issue does not necessarily affect the core SDK, and vice versa.
-
-## In scope
-
- Authentication, authorisation, and multi-tenant isolation (e.g. workspace boundaries)
- Tool execution, sandbox escape, and code execution paths
- SSRF, injection, and unsafe deserialisation in shipped code
- MCP, gateway, and bot/webhook integrations
- Secrets handling, JWT/session configuration, and default credentials
- Supply-chain issues in published package artefacts
-
-## Out of scope
-
- Issues in forks or deployments not maintained in [MervinPraison/PraisonAI](https://github.com/MervinPraison/PraisonAI)
- Vulnerabilities requiring physical access or fully compromised operator machines
- Reports against versions no longer published or without a supported upgrade path
- Theoretical issues with no practical impact in documented, supported configurations
- Behaviour covered by an explicit, documented opt-out (e.g. dev-only env vars)
-
-## How to report
-
-**Preferred:** open a **private** security advisory:
-
-[https://github.com/MervinPraison/PraisonAI/security/advisories/new](https://github.com/MervinPraison/PraisonAI/security/advisories/new)
-
-Include:
-
-1. Affected package and version
-2. Clear reproduction steps
-3. Impact assessment (confidentiality, integrity, availability)
-4. Suggested fix (optional)
-
-Please allow reasonable time for triage and a patched release before public disclosure.
-
-## What to expect
-
-1. **Triage** — we validate the report and assign a disposition (`real`, `duplicate`, `wontfix`, etc.)
-2. **Fix** — minimal, backward-compatible patch on `main` where appropriate
-3. **Release** — patched version published to PyPI/npm **before** the advisory is published
-4. **Advisory** — GitHub Security Advisory with reporter credit and CVE where applicable
-
-## Resources
-
- **Repository:** https://github.com/MervinPraison/PraisonAI
- **Documentation:** https://docs.praison.ai
- **Website:** https://praison.ai
- **Advisories:** https://github.com/MervinPraison/PraisonAI/security/advisories
-
-## Maintainer workflow
-
-Contributors and agents handling audits should follow:
-
-`src/praisonai-agents/.agent/workflows/security-audit.md`
@@ -30,7 +30,6 @@ from praisonaiagents import Agent, AutoAgents, AutoRagAgent, ContextAgent, DeepR
 Methods:

 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">aclose</a>() -> None</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">add_mcp_server</a>(name: str, mcp: Any) -> Any</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">agent_id</a>() -> str</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">analyze_prompt</a>(prompt: str) -> set</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">auto_memory</a>() -> Optional[bool]</code>
@@ -60,9 +59,6 @@ Methods:
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">handoff_to</a>(target_agent: 'Agent', prompt: str, context: Optional[Dict[str, Any]] = None, config: Optional['HandoffConfig'] = None) -> 'HandoffResult'</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">handoff_to_async</a>(target_agent: 'Agent', prompt: str, context: Optional[Dict[str, Any]] = None, config: Optional['HandoffConfig'] = None) -> 'HandoffResult'</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">is_closed</a>() -> bool</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">list_mcp_servers</a>() -> List[str]</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">llm_instance</a>()</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">llm_instance</a>(value)</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">llm_model</a>() -> Optional[str]</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">output_style</a>() -> Optional[str]</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">output_style</a>(value: Optional[str]) -> None</code>
@@ -72,8 +68,6 @@ Methods:
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">rag</a>() -> Optional[Any]</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">rag_query</a>(question: str, **kwargs) -> 'RAGResult'</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">redo</a>() -> bool</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">refresh_tools</a>() -> List[Any]</code>
-* <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">remove_mcp_server</a>(name: str) -> bool</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">retrieval_config</a>() -> Optional[Any]</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">retrieve</a>(query: str, **kwargs) -> 'ContextPack'</code>
 * <code title="class Agent">Agent.<a href="./src/praisonai-agents/praisonaiagents/agent/agent.py">rules_manager</a>() -> Optional[Any]</code>
@@ -246,7 +240,6 @@ from praisonaiagents import Memory, MemoryBackend, MemoryConfig

 Methods:

-* <code title="class Memory">Memory.<a href="./src/praisonai-agents/praisonaiagents/memory/memory.py">build_cache_optimized_context</a>(task_descr: str, user_id: Optional[str] = None, additional: str = '', max_items: int = 3, include_cache_boundary: bool = True, include_in_output: Optional[bool] = None) -> Dict[str, str]</code>
 * <code title="class Memory">Memory.<a href="./src/praisonai-agents/praisonaiagents/memory/memory.py">build_context_for_task</a>(task_descr: str, user_id: Optional[str] = None, additional: str = '', max_items: int = 3, include_in_output: Optional[bool] = None) -> str</code>
 * <code title="class Memory">Memory.<a href="./src/praisonai-agents/praisonaiagents/memory/memory.py">calculate_quality_metrics</a>(output: str, expected_output: str, llm: Optional[str] = None, custom_prompt: Optional[str] = None) -> Dict[str, float]</code>
 * <code title="class Memory">Memory.<a href="./src/praisonai-agents/praisonaiagents/memory/memory.py">close_connections</a>()</code>
@@ -483,16 +476,11 @@ Methods:
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">activate</a>(skill: LoadedSkill) -> bool</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">activate_by_name</a>(name: str) -> bool</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">add_skill</a>(skill_path: str) -> Optional[LoadedSkill]</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">approve</a>(identifier: str) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">archive_skill</a>(name: str) -> dict</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">clear</a>() -> None</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">create</a>(name: str, content: str, category: Optional[str] = None, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">create_skill</a>(name: str, content: str, category: str = None, agent_created: bool = True, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">delete</a>(name: str, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">delete_skill</a>(name: str, hard: bool = False, propose: Optional[bool] = None) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">create_skill</a>(name: str, content: str, category: str = None) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">delete_skill</a>(name: str) -> dict</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">discover</a>(skill_dirs: Optional[List[str]] = None, include_defaults: bool = True) -> int</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">edit</a>(name: str, content: str, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">edit_skill</a>(name: str, content: str, propose: Optional[bool] = None) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">edit_skill</a>(name: str, content: str) -> dict</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">get_allowed_tools</a>(name: str) -> List[str]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">get_available_skills</a>() -> List[SkillMetadata]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">get_available_skills_by_state</a>(state: SkillState) -> List[LoadedSkill]</code>
@@ -501,24 +489,15 @@ Methods:
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">get_skills_diagnostics</a>() -> Dict[str, ValidationResult]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">get_user_invocable_skills</a>() -> List[LoadedSkill]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">invoke</a>(name: str, raw_args: str = '', session_id: Optional[str] = None, shell_exec: bool = False) -> Optional[str]</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">list_archived_skills</a>() -> List[str]</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">list_pending</a>() -> List[dict]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">load_resources</a>(name: str) -> bool</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">patch</a>(name: str, old_string: str, new_string: str, file_path: Optional[str] = None, replace_all: bool = False, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">patch_skill</a>(name: str, old_string: str, new_string: str, file_path: str = None, replace_all: bool = False, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">reject</a>(identifier: str) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">remove_file</a>(name: str, file_path: str, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">remove_skill_file</a>(name: str, file_path: str, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">restore_skill</a>(name: str, skill_dir: Optional[str] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">rollback_skill</a>(name: str) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">patch_skill</a>(name: str, old_string: str, new_string: str, file_path: str = None, replace_all: bool = False) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">remove_skill_file</a>(name: str, file_path: str) -> dict</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">skill_names</a>() -> List[str]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">skills</a>() -> List[LoadedSkill]</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">to_prompt</a>() -> str</code>
 * <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">validate_skill_capabilities</a>(skill_name: str, force_refresh: bool = False) -> ValidationResult</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">write_file</a>(name: str, file_path: str, file_content: str, propose: Optional[bool] = None) -> dict</code>
-* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">write_skill_file</a>(name: str, file_path: str, file_content: str, propose: Optional[bool] = None) -> dict</code>
+* <code title="class SkillManager">SkillManager.<a href="./src/praisonai-agents/praisonaiagents/skills/manager.py">write_skill_file</a>(name: str, file_path: str, file_content: str) -> dict</code>
 * <code title="class SkillMetadata">SkillMetadata.<a href="./src/praisonai-agents/praisonaiagents/skills/models.py">from_properties</a>(props: SkillProperties) -> 'SkillMetadata'</code>
-* <code title="class SkillProperties">SkillProperties.<a href="./src/praisonai-agents/praisonaiagents/skills/models.py">idle_days</a>() -> Optional[float]</code>
 * <code title="class SkillProperties">SkillProperties.<a href="./src/praisonai-agents/praisonaiagents/skills/models.py">to_dict</a>() -> dict</code>
 * <code title="class SkillsConfig">SkillsConfig.<a href="./src/praisonai-agents/praisonaiagents/config/feature_configs.py">to_dict</a>() -> Dict[str, Any]</code>

@@ -536,12 +515,10 @@ Methods:
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">add_memory</a>(text: str, memory_type: str = 'long', **metadata) -> None</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">chat</a>(message: str, **kwargs) -> str</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">clear_memory</a>(memory_type: str = 'all') -> None</code>
-* <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">close</a>() -> None</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">create_agent</a>(*args, **kwargs) -> 'Agent'</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">get_context</a>(query: str, max_items: int = 3) -> str</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">get_state</a>(key: str, default: Any = None) -> Any</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">increment_state</a>(key: str, increment: int = 1, default: int = 0) -> None</code>
-* <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">is_expired</a>() -> bool</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">knowledge</a>() -> 'Knowledge'</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">memory</a>() -> 'Memory'</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">restore_state</a>() -> Dict[str, Any]</code>
@@ -550,7 +527,6 @@ Methods:
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">search_memory</a>(query: str, memory_type: str = 'long', limit: int = 5) -> List[Dict[str, Any]]</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">send_message</a>(message: str, **kwargs) -> str</code>
 * <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">set_state</a>(key: str, value: Any) -> None</code>
-* <code title="class Session">Session.<a href="./src/praisonai-agents/praisonaiagents/session/api.py">time_to_expiry</a>() -> Optional[float]</code>

 # MCP

@@ -633,7 +609,7 @@ Methods:
 * <code title="class ContextManager">ContextManager.<a href="./src/praisonai-agents/praisonaiagents/context/manager.py">register_snapshot_callback</a>(callback: Callable[[SnapshotHookData], None]) -> None</code>
 * <code title="class ContextManager">ContextManager.<a href="./src/praisonai-agents/praisonaiagents/context/manager.py">reset</a>() -> None</code>
 * <code title="class ContextManager">ContextManager.<a href="./src/praisonai-agents/praisonaiagents/context/manager.py">set_tool_budget</a>(tool_name: str, max_tokens: int, protected: bool = False) -> None</code>
-* <code title="class ContextManager">ContextManager.<a href="./src/praisonai-agents/praisonaiagents/context/manager.py">truncate_tool_output</a>(tool_name: str, output: str, tool_call_id: str = None, run_id: str = None) -> str</code>
+* <code title="class ContextManager">ContextManager.<a href="./src/praisonai-agents/praisonaiagents/context/manager.py">truncate_tool_output</a>(tool_name: str, output: str) -> str</code>

 # UI

@@ -722,7 +698,7 @@ Methods:

 Types:
 ```python
-from praisonaiagents import AgentAppConfig, AgentAppProtocol, AgentFlow, AgentManager, AgentOSConfig, AgentOSProtocol, AgentTeam, AutoApproveBackend, EmbeddingResult, RetryBackoffConfig, __version__, aembedding, aembeddings, embedding, embeddings, get_dimensions
+from praisonaiagents import AgentAppConfig, AgentAppProtocol, AgentFlow, AgentManager, AgentOSConfig, AgentOSProtocol, AgentTeam, AutoApproveBackend, EmbeddingResult, aembedding, aembeddings, embedding, embeddings, get_dimensions
 ```

 Methods:
@@ -813,18 +789,11 @@ Methods:

 * <code title="cli">praisonai <a href="./src/praisonai/praisonai/cli/main.py">--help</a></code>
 * <code title="cli">praisonai acp acp-main <a href="./src/praisonai/praisonai/cli/commands/acp.py">--help</a></code>
-* <code title="cli">praisonai agent list <a href="./src/praisonai/praisonai/cli/commands/agent.py">--help</a></code>
-* <code title="cli">praisonai agent show <a href="./src/praisonai/praisonai/cli/commands/agent.py">--help</a></code>
 * <code title="cli">praisonai agents create <a href="./src/praisonai/praisonai/cli/commands/agents.py">--help</a></code>
 * <code title="cli">praisonai agents info <a href="./src/praisonai/praisonai/cli/commands/agents.py">--help</a></code>
 * <code title="cli">praisonai agents list <a href="./src/praisonai/praisonai/cli/commands/agents.py">--help</a></code>
 * <code title="cli">praisonai app app <a href="./src/praisonai/praisonai/cli/commands/app.py">--help</a></code>
-* <code title="cli">praisonai attach attach-main <a href="./src/praisonai/praisonai/cli/commands/attach.py">--help</a></code>
 * <code title="cli">praisonai audit agent-centric <a href="./src/praisonai/praisonai/cli/commands/audit.py">--help</a></code>
-* <code title="cli">praisonai auth list <a href="./src/praisonai/praisonai/cli/commands/auth.py">--help</a></code>
-* <code title="cli">praisonai auth login <a href="./src/praisonai/praisonai/cli/commands/auth.py">--help</a></code>
-* <code title="cli">praisonai auth logout <a href="./src/praisonai/praisonai/cli/commands/auth.py">--help</a></code>
-* <code title="cli">praisonai auth status <a href="./src/praisonai/praisonai/cli/commands/auth.py">--help</a></code>
 * <code title="cli">praisonai batch batch-run <a href="./src/praisonai/praisonai/cli/commands/batch.py">--help</a></code>
 * <code title="cli">praisonai batch list <a href="./src/praisonai/praisonai/cli/commands/batch.py">--help</a></code>
 * <code title="cli">praisonai batch report <a href="./src/praisonai/praisonai/cli/commands/batch.py">--help</a></code>
@@ -862,16 +831,9 @@ Methods:
 * <code title="cli">praisonai call call-main <a href="./src/praisonai/praisonai/cli/commands/call.py">--help</a></code>
 * <code title="cli">praisonai chat <a href="./src/praisonai/praisonai/cli/main.py">--help</a></code>
 * <code title="cli">praisonai chat chat-main <a href="./src/praisonai/praisonai/cli/commands/chat.py">--help</a></code>
-* <code title="cli">praisonai checkpoint delete <a href="./src/praisonai/praisonai/cli/commands/checkpoint.py">--help</a></code>
-* <code title="cli">praisonai checkpoint diff <a href="./src/praisonai/praisonai/cli/commands/checkpoint.py">--help</a></code>
-* <code title="cli">praisonai checkpoint list <a href="./src/praisonai/praisonai/cli/commands/checkpoint.py">--help</a></code>
-* <code title="cli">praisonai checkpoint restore <a href="./src/praisonai/praisonai/cli/commands/checkpoint.py">--help</a></code>
-* <code title="cli">praisonai checkpoint save <a href="./src/praisonai/praisonai/cli/commands/checkpoint.py">--help</a></code>
 * <code title="cli">praisonai claw claw <a href="./src/praisonai/praisonai/cli/commands/claw.py">--help</a></code>
 * <code title="cli">praisonai code <a href="./src/praisonai/praisonai/cli/main.py">--help</a></code>
 * <code title="cli">praisonai code code-main <a href="./src/praisonai/praisonai/cli/commands/code.py">--help</a></code>
-* <code title="cli">praisonai command list <a href="./src/praisonai/praisonai/cli/commands/command.py">--help</a></code>
-* <code title="cli">praisonai command show <a href="./src/praisonai/praisonai/cli/commands/command.py">--help</a></code>
 * <code title="cli">praisonai commit commit-main <a href="./src/praisonai/praisonai/cli/commands/commit.py">--help</a></code>
 * <code title="cli">praisonai completion bash <a href="./src/praisonai/praisonai/cli/commands/completion.py">--help</a></code>
 * <code title="cli">praisonai completion completion-callback <a href="./src/praisonai/praisonai/cli/commands/completion.py">--help</a></code>
@@ -884,9 +846,6 @@ Methods:
 * <code title="cli">praisonai config path <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
 * <code title="cli">praisonai config reset <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
 * <code title="cli">praisonai config set <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
-* <code title="cli">praisonai config show <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
-* <code title="cli">praisonai config sources <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
-* <code title="cli">praisonai config validate <a href="./src/praisonai/praisonai/cli/commands/config.py">--help</a></code>
 * <code title="cli">praisonai context add <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
 * <code title="cli">praisonai context clear <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
 * <code title="cli">praisonai context compact <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
@@ -896,10 +855,6 @@ Methods:
 * <code title="cli">praisonai context show <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
 * <code title="cli">praisonai context stats <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
 * <code title="cli">praisonai context tail <a href="./src/praisonai/praisonai/cli/commands/context.py">--help</a></code>
-* <code title="cli">praisonai daemon daemon-main <a href="./src/praisonai/praisonai/cli/commands/daemon.py">--help</a></code>
-* <code title="cli">praisonai daemon start <a href="./src/praisonai/praisonai/cli/commands/daemon.py">--help</a></code>
-* <code title="cli">praisonai daemon status <a href="./src/praisonai/praisonai/cli/commands/daemon.py">--help</a></code>
-* <code title="cli">praisonai daemon stop <a href="./src/praisonai/praisonai/cli/commands/daemon.py">--help</a></code>
 * <code title="cli">praisonai dashboard unified <a href="./src/praisonai/praisonai/cli/commands/dashboard.py">--help</a></code>
 * <code title="cli">praisonai debug acp <a href="./src/praisonai/praisonai/cli/commands/debug.py">--help</a></code>
 * <code title="cli">praisonai debug debug-callback <a href="./src/praisonai/praisonai/cli/commands/debug.py">--help</a></code>
@@ -921,18 +876,17 @@ Methods:
 * <code title="cli">praisonai docs serve <a href="./src/praisonai/praisonai/cli/commands/docs.py">--help</a></code>
 * <code title="cli">praisonai docs stats <a href="./src/praisonai/praisonai/cli/commands/docs.py">--help</a></code>
 * <code title="cli">praisonai doctor cleanup <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
-* <code title="cli">praisonai doctor docker <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
+* <code title="cli">praisonai doctor config <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
+* <code title="cli">praisonai doctor db <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
+* <code title="cli">praisonai doctor doctor-callback <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor env <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
-* <code title="cli">praisonai doctor llm-providers <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor mcp <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
-* <code title="cli">praisonai doctor memory-store <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
-* <code title="cli">praisonai doctor metadata-store <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor network <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor packaging <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor performance <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
-* <code title="cli">praisonai doctor runtime <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor selftest <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor skills <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
+* <code title="cli">praisonai doctor tools <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai doctor troubleshoot <a href="./src/praisonai/praisonai/cli/commands/doctor.py">--help</a></code>
 * <code title="cli">praisonai endpoints list <a href="./src/praisonai/praisonai/cli/commands/endpoints.py">--help</a></code>
 * <code title="cli">praisonai endpoints test <a href="./src/praisonai/praisonai/cli/commands/endpoints.py">--help</a></code>
@@ -960,9 +914,6 @@ Methods:
 * <code title="cli">praisonai gateway install <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
 * <code title="cli">praisonai gateway logs <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
 * <code title="cli">praisonai gateway mint-link <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
-* <code title="cli">praisonai gateway pause <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
-* <code title="cli">praisonai gateway reconnect <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
-* <code title="cli">praisonai gateway resume <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
 * <code title="cli">praisonai gateway send <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
 * <code title="cli">praisonai gateway start <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
 * <code title="cli">praisonai gateway status <a href="./src/praisonai/praisonai/cli/commands/gateway.py">--help</a></code>
@@ -972,11 +923,6 @@ Methods:
 * <code title="cli">praisonai hooks add <a href="./src/praisonai/praisonai/cli/commands/hooks.py">--help</a></code>
 * <code title="cli">praisonai hooks list <a href="./src/praisonai/praisonai/cli/commands/hooks.py">--help</a></code>
 * <code title="cli">praisonai hooks remove <a href="./src/praisonai/praisonai/cli/commands/hooks.py">--help</a></code>
-* <code title="cli">praisonai identity import <a href="./src/praisonai/praisonai/cli/commands/identity.py">--help</a></code>
-* <code title="cli">praisonai identity link <a href="./src/praisonai/praisonai/cli/commands/identity.py">--help</a></code>
-* <code title="cli">praisonai identity list <a href="./src/praisonai/praisonai/cli/commands/identity.py">--help</a></code>
-* <code title="cli">praisonai identity unlink <a href="./src/praisonai/praisonai/cli/commands/identity.py">--help</a></code>
-* <code title="cli">praisonai init init <a href="./src/praisonai/praisonai/cli/commands/init.py">--help</a></code>
 * <code title="cli">praisonai kanban block <a href="./src/praisonai/praisonai/cli/commands/kanban.py">--help</a></code>
 * <code title="cli">praisonai kanban boards <a href="./src/praisonai/praisonai/cli/commands/kanban.py">--help</a></code>
 * <code title="cli">praisonai kanban comment <a href="./src/praisonai/praisonai/cli/commands/kanban.py">--help</a></code>
@@ -1041,9 +987,6 @@ Methods:
 * <code title="cli">praisonai memory search <a href="./src/praisonai/praisonai/cli/commands/memory.py">--help</a></code>
 * <code title="cli">praisonai memory show <a href="./src/praisonai/praisonai/cli/commands/memory.py">--help</a></code>
 * <code title="cli">praisonai memory status <a href="./src/praisonai/praisonai/cli/commands/memory.py">--help</a></code>
-* <code title="cli">praisonai models describe <a href="./src/praisonai/praisonai/cli/commands/models.py">--help</a></code>
-* <code title="cli">praisonai models list <a href="./src/praisonai/praisonai/cli/commands/models.py">--help</a></code>
-* <code title="cli">praisonai models validate <a href="./src/praisonai/praisonai/cli/commands/models.py">--help</a></code>
 * <code title="cli">praisonai n8n export <a href="./src/praisonai/praisonai/cli/commands/n8n.py">--help</a></code>
 * <code title="cli">praisonai n8n import <a href="./src/praisonai/praisonai/cli/commands/n8n.py">--help</a></code>
 * <code title="cli">praisonai n8n list <a href="./src/praisonai/praisonai/cli/commands/n8n.py">--help</a></code>
@@ -1061,14 +1004,6 @@ Methods:
 * <code title="cli">praisonai pairing list <a href="./src/praisonai/praisonai/cli/commands/pairing.py">--help</a></code>
 * <code title="cli">praisonai pairing revoke <a href="./src/praisonai/praisonai/cli/commands/pairing.py">--help</a></code>
 * <code title="cli">praisonai paths show <a href="./src/praisonai/praisonai/cli/commands/paths.py">--help</a></code>
-* <code title="cli">praisonai permissions allow <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions ask <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions deny <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions export <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions import-rules <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions list <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions remove <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
-* <code title="cli">praisonai permissions reset <a href="./src/praisonai/praisonai/cli/commands/permissions.py">--help</a></code>
 * <code title="cli">praisonai plugins create <a href="./src/praisonai/praisonai/cli/commands/plugins.py">--help</a></code>
 * <code title="cli">praisonai plugins disable <a href="./src/praisonai/praisonai/cli/commands/plugins.py">--help</a></code>
 * <code title="cli">praisonai plugins discover <a href="./src/praisonai/praisonai/cli/commands/plugins.py">--help</a></code>
@@ -1131,8 +1066,6 @@ Methods:
 * <code title="cli">praisonai sandbox sandbox-callback <a href="./src/praisonai/praisonai/cli/commands/sandbox.py">--help</a></code>
 * <code title="cli">praisonai sandbox status <a href="./src/praisonai/praisonai/cli/commands/sandbox.py">--help</a></code>
 * <code title="cli">praisonai schedule add <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule blueprint <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule blueprint-list <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai schedule delete <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai schedule describe <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai schedule list <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
@@ -1142,10 +1075,6 @@ Methods:
 * <code title="cli">praisonai schedule start <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai schedule stats <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai schedule stop <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule suggestion-accept <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule suggestion-dismiss <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule suggestion-propose <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
-* <code title="cli">praisonai schedule suggestions <a href="./src/praisonai/praisonai/cli/commands/schedule.py">--help</a></code>
 * <code title="cli">praisonai serve a2a <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve a2u <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve acp <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
@@ -1154,7 +1083,6 @@ Methods:
 * <code title="cli">praisonai serve gateway <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve lsp <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve mcp <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
-* <code title="cli">praisonai serve openai <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve rag <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve recipe <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
 * <code title="cli">praisonai serve registry <a href="./src/praisonai/praisonai/cli/commands/serve.py">--help</a></code>
@@ -1181,7 +1109,6 @@ Methods:
 * <code title="cli">praisonai skills eligible <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
 * <code title="cli">praisonai skills info <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
 * <code title="cli">praisonai skills install <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
-* <code title="cli">praisonai skills learn <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
 * <code title="cli">praisonai skills list <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
 * <code title="cli">praisonai skills search <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
 * <code title="cli">praisonai skills validate <a href="./src/praisonai/praisonai/cli/commands/skills.py">--help</a></code>
@@ -1223,9 +1150,6 @@ Methods:
 * <code title="cli">praisonai up logs <a href="./src/praisonai/praisonai/cli/commands/up.py">--help</a></code>
 * <code title="cli">praisonai up status <a href="./src/praisonai/praisonai/cli/commands/up.py">--help</a></code>
 * <code title="cli">praisonai up up-start <a href="./src/praisonai/praisonai/cli/commands/up.py">--help</a></code>
-* <code title="cli">praisonai validate check <a href="./src/praisonai/praisonai/cli/commands/validate.py">--help</a></code>
-* <code title="cli">praisonai validate schema <a href="./src/praisonai/praisonai/cli/commands/validate.py">--help</a></code>
-* <code title="cli">praisonai validate validate <a href="./src/praisonai/praisonai/cli/commands/validate.py">--help</a></code>
 * <code title="cli">praisonai version check <a href="./src/praisonai/praisonai/cli/commands/version.py">--help</a></code>
 * <code title="cli">praisonai version show <a href="./src/praisonai/praisonai/cli/commands/version.py">--help</a></code>
 * <code title="cli">praisonai version version-callback <a href="./src/praisonai/praisonai/cli/commands/version.py">--help</a></code>
@@ -1377,8 +1301,8 @@ export { // Python parity additions
  MinimalTelemetry, AgentTelemetry, PerformanceMonitor, TelemetryCollector, TelemetryIntegration, cleanupTelemetry, cleanupTelemetryResources, createAgentTelemetry, createConsoleSink, createHTTPSink, createPerformanceMonitor, createTelemetryIntegration, disablePerformanceMode, disableTelemetry, enablePerformanceMode, enableTelemetry, getMinimalTelemetry, getTelemetry } from "./telemetry";
 export { // Subagent Tool (agent-as-tool pattern)
  SubagentTool, BaseTool, FunctionTool, ToolRegistry, ToolResult, ToolValidationError, createDelegator, createSubagentTool, createSubagentTools, createTool, getRegistry, getTool, registerTool, tool, validateTool } from "./tools";
-export { airweaveSearch, bedrockBrowserClick, bedrockBrowserFill, bedrockBrowserNavigate, bedrockCodeInterpreter, codeExecution, codeMode, createCustomTool, crwCrawl, crwScrape, exaSearch, firecrawlCrawl, firecrawlScrape, parallelSearch, perplexitySearch, registerCustomTool, registerLocalTool, registerNpmTool, superagentGuard, superagentRedact, superagentVerify, tavilyCrawl, tavilyExtract, tavilySearch, valyuBioSearch, valyuCompanyResearch, valyuEconomicsSearch, valyuFinanceSearch, valyuPaperSearch, valyuPatentSearch, valyuSecSearch, valyuWebSearch } from "./tools/builtins";
-export { BudgetExceededError, MissingDependencyError, MissingEnvVarError, ToolsRegistry, composeMiddleware, createLoggingMiddleware, createRateLimitMiddleware, createRedactionMiddleware, createRetryMiddleware, createTimeoutMiddleware, createToolsRegistry, createTracingMiddleware, createValidationMiddleware, getToolsRegistry, get_registry, get_tool, register_tool, resetToolsRegistry, validate_tool } from "./tools/registry";
+export { airweaveSearch, bedrockBrowserClick, bedrockBrowserFill, bedrockBrowserNavigate, bedrockCodeInterpreter, codeExecution, codeMode, createCustomTool, exaSearch, firecrawlCrawl, firecrawlScrape, parallelSearch, perplexitySearch, registerCustomTool, registerLocalTool, registerNpmTool, superagentGuard, superagentRedact, superagentVerify, tavilyCrawl, tavilyExtract, tavilySearch, valyuBioSearch, valyuCompanyResearch, valyuEconomicsSearch, valyuFinanceSearch, valyuPaperSearch, valyuPatentSearch, valyuSecSearch, valyuWebSearch } from "./tools/builtins";
+export { BudgetExceededError, MissingDependencyError, MissingEnvVarError, ToolsRegistry, composeMiddleware, createLoggingMiddleware, createRateLimitMiddleware, createRedactionMiddleware, createRetryMiddleware, createTimeoutMiddleware, createToolsRegistry, createTracingMiddleware, createValidationMiddleware, getRegistry, getTool, getToolsRegistry, get_registry, get_tool, registerTool, register_tool, resetToolsRegistry, validateTool, validate_tool } from "./tools/registry";
 export type { InstallHints, PraisonTool, RedactionHooks, RegisteredTool, ToolCapabilities, ToolExecutionContext, ToolExecutionResult, ToolFactory, ToolHooks, ToolInstallStatus, ToolLimits, ToolLogger, ToolMetadata, ToolMiddleware, ToolParameterProperty, ToolParameterSchema } from "./tools/registry";
 export { registerBuiltinTools, tools } from "./tools/tools";
 export { // Classes
@@ -16,7 +16,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
    praisonai_tools \
-    "praisonai>=4.6.80" \
+    "praisonai>=4.6.55" \
    "praisonai[chat]" \
    "embedchain[github,youtube]"

@@ -20,7 +20,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
    praisonai_tools \
-    "praisonai>=4.6.80" \
+    "praisonai>=4.6.55" \
    "praisonai[ui]" \
    "praisonai[chat]" \
    "praisonai[realtime]" \
@@ -16,7 +16,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
    praisonai_tools \
-    "praisonai>=4.6.80" \
+    "praisonai>=4.6.55" \
    "praisonai[ui]" \
    "praisonai[crewai]"

@@ -5,9 +5,9 @@ Demonstrates web search with consolidated params.
 Presets: duckduckgo, tavily, google, bing, serper, search_only, fetch_only
 """

-from praisonaiagents import Agent, WebConfig
+from praisonaiagents import Agent

-# Basic: Enable web search with preset (search only — fetch needs Claude)
+# Basic: Enable web search with preset
 agent = Agent(
    instructions="You are a research assistant with web access.",
    web="search_only",  # Compatible across models without web_fetch
@@ -1,216 +0,0 @@
-# Industry Templates for PraisonAI
-
-This directory contains industry-specific agent templates based on the SRAO Framework, providing pre-built workflows for rapid deployment of agent workforces across different industries.
-
-## Overview
-
-These templates demonstrate how PraisonAI can be used to create specialized agent teams for various industries, with **70% code reuse** across different sectors through shared patterns and abstractions.
-
-## Available Templates
-
-### 1. Manufacturing (`manufacturing_template.py`)
-**Key Agents:**
- **ParseOrder**: Extracts and structures order information from various formats
- **CheckInventory**: Validates material availability and supply chain status  
- **OptimizeSchedule**: Optimizes production scheduling based on constraints
- **DefectDetect**: Quality control and defect detection
-
-**Features:**
- Order processing with multiple input formats
- Real-time inventory management
- Production optimization algorithms
- Quality inspection with defect detection
- Fallback strategies for each critical step
-
-### 2. Energy (`energy_template.py`)
-**Key Agents:**
- **SCADAReader**: Processes SCADA telemetry data from wind turbines
- **VibrationAnalyzer**: Analyzes vibration patterns for predictive maintenance
- **PowerForecaster**: Predicts power generation based on weather and history
- **MaintenanceScheduler**: Schedules predictive maintenance
-
-**Features:**
- Real-time wind farm monitoring
- Vibration-based fault detection
- Power generation forecasting
- Grid integration and load balancing
- Predictive maintenance scheduling
-
-### 3. Healthcare (`healthcare_template.py`)
-**Key Agents:**
- **VitalSignsCapture**: Processes patient vital signs from monitoring devices
- **EMRRetrieval**: Retrieves electronic medical records with HIPAA compliance
- **TriageRecommendation**: Provides ESI-based triage recommendations
- **ResourceAllocator**: Manages hospital resource allocation
-
-**Features:**
- Emergency triage workflow
- HIPAA-compliant data handling
- Patient safety protocols
- Resource optimization
- Cross-department coordination
-
-### 4. Agriculture (`agriculture_template.py`)
-**Key Agents:**
- **MultispectralAnalyzer**: Analyzes multispectral imagery for crop health
- **DiseaseIdentifier**: Identifies crop diseases and pests using AI
- **SprayRecommender**: Recommends targeted spraying strategies
- **YieldPredictor**: Predicts crop yield based on conditions
-
-**Features:**
- Precision agriculture with drone/satellite imagery
- Pest and disease early warning system
- Optimized chemical application
- Yield forecasting
- Sustainable farming patterns
-
-### 5. Transportation (`transportation_template.py`)
-**Key Agents:**
- **LiDARFusion**: Processes LiDAR point cloud data for infrastructure
- **DisplacementCalculator**: Calculates structural displacement
- **HeatmapGenerator**: Generates safety visualization heatmaps
- **MaintenancePlanner**: Plans infrastructure maintenance
-
-**Features:**
- Tunnel/bridge safety monitoring
- LiDAR-based structural analysis
- Real-time safety assessment
- Traffic flow optimization
- Predictive maintenance scheduling
-
-## Key Features Across All Templates
-
-### 1. I/O Schemas
-Each template includes Pydantic models for structured data handling:
- Input validation
- Type safety
- Clear data contracts between agents
-
-### 2. SLA Requirements
-Every agent has defined Service Level Agreements:
- Processing time requirements
- Response time guarantees
- Performance benchmarks
-
-### 3. Fallback Strategies
-Robust error handling with fallback mechanisms:
- Graceful degradation
- Manual intervention queues
- Conservative estimates when systems fail
-
-### 4. Cross-Industry Patterns (70% Reuse)
-Reusable base patterns that work across industries:
- Data parsing agents
- Resource availability checkers
- Optimization engines
- Inspection/quality control agents
-
-## Usage Example
-
-```python
-from manufacturing_template import (
-    order_parser,
-    inventory_checker,
-    schedule_optimizer,
-    quality_inspector,
-    manufacturing_workflow
-)
-
-# Process a manufacturing order
-order_text = "Customer ABC needs 100 units of Product-XYZ by March 15th, high priority"
-result = manufacturing_workflow(order_text)
-print(result)
-```
-
-## Adapting Templates for Your Industry
-
-### Using Base Patterns
-Each template includes reusable pattern classes that can be adapted:
-
-```python
-from manufacturing_template import IndustryAgentPattern
-
-# Create a custom data parser for any domain
-custom_parser = IndustryAgentPattern.create_data_parser(
-    name="CustomParser",
-    domain="retail",
-    sla_seconds=20
-)
-
-# Create a resource checker for any resource type
-resource_checker = IndustryAgentPattern.create_resource_checker(
-    name="StaffAvailability",
-    resource_type="customer_service_staff",
-    sla_seconds=3
-)
-```
-
-### Creating Custom Workflows
-Combine agents from different templates or create new ones:
-
-```python
-from praisonaiagents import Agent, tool
-
-# Define custom tools
-@tool
-def custom_analysis(data: str) -> Dict:
-    """Your custom analysis logic"""
-    return {"result": "analyzed"}
-
-# Create custom agent
-custom_agent = Agent(
-    name="CustomAnalyzer",
-    instructions="Your specific instructions",
-    tools=[custom_analysis]
-)
-```
-
-## Integration with PraisonAI Core
-
-These templates are built on top of PraisonAI's core Agent framework:
-
-```python
-from praisonaiagents import Agent, tool, AgentTeam, Task
-
-# Combine multiple industry agents into a team
-team = AgentTeam(
-    agents=[order_parser, inventory_checker, schedule_optimizer],
-    tasks=[
-        Task(name="parse", agent=order_parser),
-        Task(name="check", agent=inventory_checker),
-        Task(name="optimize", agent=schedule_optimizer)
-    ]
-)
-```
-
-## Best Practices
-
-1. **Start with Templates**: Use these as starting points and customize for your specific needs
-2. **Maintain SLAs**: Define clear performance requirements for each agent
-3. **Implement Fallbacks**: Always have fallback strategies for critical operations
-4. **Use Type Safety**: Leverage Pydantic models for data validation
-5. **Monitor Performance**: Track agent performance against SLAs
-6. **Iterate and Improve**: Continuously refine based on real-world usage
-
-## Contributing
-
-To add a new industry template:
-
-1. Follow the existing template structure
-2. Include all key components:
-   - I/O Schemas (Pydantic models)
-   - Agent definitions with SLAs
-   - Tools for agent capabilities
-   - Complete workflow function
-   - Fallback strategies
-   - Cross-industry patterns
-3. Add comprehensive documentation
-4. Include usage examples
-
-## License
-
-These templates are part of the PraisonAI project and follow the same licensing terms.
-
-## References
-
-Based on the [SRAO Framework](https://github.com/beixuan577/SRAO-Framework) (MIT License)
@@ -1,38 +0,0 @@
-"""
-Industry Templates for PraisonAI
-================================
-
-Pre-built agent templates for rapid deployment across different industries.
-Based on SRAO Framework with 70% cross-industry code reuse.
-
-Available templates:
- Manufacturing: Order processing, inventory, scheduling, quality control
- Energy: Wind farm monitoring, predictive maintenance, power forecasting
- Healthcare: Emergency triage, EMR retrieval, resource allocation
- Agriculture: Precision farming, pest detection, yield prediction
- Transportation: Infrastructure monitoring, safety assessment, maintenance
-
-Example usage:
-    from Industry_Templates.manufacturing_template import manufacturing_workflow
-    
-    result = manufacturing_workflow("Customer order text")
-"""
-
-__version__ = "1.0.0"
-__author__ = "PraisonAI Team"
-
-# Import the modules so they're accessible via the package
-from . import manufacturing_template
-from . import energy_template
-from . import healthcare_template
-from . import agriculture_template
-from . import transportation_template
-
-# Export commonly used items for convenience
-__all__ = [
-    "manufacturing_template",
-    "energy_template", 
-    "healthcare_template",
-    "agriculture_template",
-    "transportation_template"
-]
@@ -1,551 +0,0 @@
-"""
-Agriculture Industry Template
-=============================
-Precision agriculture and pest management workflow
-Based on SRAO Framework with IoT sensor integration
-
-Key agents:
- MultispectralAnalyzer: Analyzes multispectral imagery
- DiseaseIdentifier: Identifies crop diseases and pests
- SprayRecommender: Recommends targeted spraying strategies
- YieldPredictor: Predicts crop yield based on conditions
-"""
-
-from praisonaiagents import Agent, tool
-from typing import Dict, List, Any, Tuple, Optional
-from pydantic import BaseModel
-from datetime import datetime, timedelta
-from enum import Enum
-
-
-# Crop health levels
-class CropHealthLevel(str, Enum):
-    EXCELLENT = "excellent"
-    GOOD = "good"
-    FAIR = "fair"
-    POOR = "poor"
-    CRITICAL = "critical"
-
-
-# Pest/Disease severity
-class SeverityLevel(str, Enum):
-    NONE = "none"
-    LOW = "low"
-    MODERATE = "moderate"
-    HIGH = "high"
-    SEVERE = "severe"
-
-
-# I/O Schemas
-class MultispectralData(BaseModel):
-    """Multispectral imagery analysis data"""
-    field_id: str
-    capture_time: str
-    ndvi_index: float  # Normalized Difference Vegetation Index
-    ndre_index: float  # Normalized Difference Red Edge
-    ccci_index: float  # Canopy Chlorophyll Content Index
-    moisture_level: float  # Soil moisture percentage
-    temperature: float  # Celsius
-    affected_area: float  # Hectares
-    gps_coordinates: Tuple[float, float]
-
-
-class PestDiseaseReport(BaseModel):
-    """Pest and disease identification report"""
-    report_id: str
-    field_id: str
-    detection_time: str
-    pest_type: Optional[str]
-    disease_type: Optional[str]
-    severity: SeverityLevel
-    affected_crops: List[str]
-    spread_rate: float  # % per day
-    environmental_factors: Dict[str, Any]
-    confidence_score: float
-
-
-class SprayRecommendation(BaseModel):
-    """Targeted spraying recommendation"""
-    recommendation_id: str
-    field_id: str
-    treatment_type: str  # pesticide, fungicide, herbicide, fertilizer
-    product_name: str
-    application_rate: float  # L/hectare
-    target_zones: List[Dict[str, Any]]  # GPS zones
-    optimal_time: str
-    weather_window: Dict[str, Any]
-    cost_estimate: float
-    environmental_impact: str  # low, medium, high
-
-
-class YieldForecast(BaseModel):
-    """Crop yield prediction"""
-    forecast_id: str
-    field_id: str
-    crop_type: str
-    predicted_yield: float  # tons/hectare
-    confidence_interval: Tuple[float, float]
-    harvest_window: Tuple[str, str]
-    quality_grade: str  # A, B, C
-    market_price_estimate: float
-    risk_factors: List[str]
-
-
-# Agriculture-specific tools
-@tool
-def analyze_multispectral_imagery(field_id: str, image_data: Dict) -> Dict:
-    """Analyze multispectral drone/satellite imagery for crop health"""
-    # Simulate multispectral analysis
-    ndvi = 0.75  # Healthy vegetation typically 0.6-0.9
-    
-    # Determine health based on NDVI
-    if ndvi > 0.8:
-        health = CropHealthLevel.EXCELLENT
-        affected_area = 0.0
-    elif ndvi > 0.6:
-        health = CropHealthLevel.GOOD
-        affected_area = 2.5
-    elif ndvi > 0.4:
-        health = CropHealthLevel.FAIR
-        affected_area = 5.0
-    else:
-        health = CropHealthLevel.POOR
-        affected_area = 10.0
-    
-    return {
-        "field_id": field_id,
-        "capture_time": datetime.now().isoformat(),
-        "ndvi_index": ndvi,
-        "ndre_index": 0.68,
-        "ccci_index": 0.72,
-        "moisture_level": 35.0,
-        "temperature": 24.5,
-        "affected_area": affected_area,
-        "gps_coordinates": (40.7128, -74.0060),
-        "health_level": health.value
-    }
-
-
-@tool
-def identify_pest_disease(spectral_data: Dict, historical_data: List) -> Dict:
-    """Identify pests and diseases using AI image recognition"""
-    # Simulate pest/disease detection
-    ndvi = spectral_data.get("ndvi_index", 0.75)
-    affected_area = spectral_data.get("affected_area", 0)
-    
-    if affected_area > 8.0:
-        pest_type = "aphids"
-        disease_type = "powdery_mildew"
-        severity = SeverityLevel.HIGH
-        spread_rate = 2.5
-    elif affected_area > 4.0:
-        pest_type = "whiteflies"
-        disease_type = "leaf_spot"
-        severity = SeverityLevel.MODERATE
-        spread_rate = 1.5
-    elif affected_area > 0:
-        pest_type = None
-        disease_type = "early_blight"
-        severity = SeverityLevel.LOW
-        spread_rate = 0.5
-    else:
-        pest_type = None
-        disease_type = None
-        severity = SeverityLevel.NONE
-        spread_rate = 0.0
-    
-    return {
-        "report_id": f"PDR-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "field_id": spectral_data["field_id"],
-        "detection_time": datetime.now().isoformat(),
-        "pest_type": pest_type,
-        "disease_type": disease_type,
-        "severity": severity.value,
-        "affected_crops": ["wheat", "corn"],
-        "spread_rate": spread_rate,
-        "environmental_factors": {
-            "humidity": 65,
-            "temperature": spectral_data.get("temperature", 25),
-            "rainfall_last_week": 15  # mm
-        },
-        "confidence_score": 0.85
-    }
-
-
-@tool
-def calculate_spray_recommendation(pest_report: Dict, field_data: Dict, weather: Dict) -> Dict:
-    """Calculate optimal spraying strategy for pest/disease control"""
-    severity = pest_report.get("severity", "none")
-    
-    if severity == "high" or severity == "severe":
-        treatment = "pesticide"
-        product = "Imidacloprid"
-        rate = 0.5
-        impact = "medium"
-    elif severity == "moderate":
-        treatment = "fungicide"
-        product = "Azoxystrobin"
-        rate = 0.3
-        impact = "low"
-    elif severity == "low":
-        treatment = "organic_treatment"
-        product = "Neem Oil"
-        rate = 0.2
-        impact = "minimal"
-    else:
-        treatment = "fertilizer"
-        product = "NPK 20-20-20"
-        rate = 0.4
-        impact = "positive"
-    
-    # Calculate target zones based on affected area
-    affected_area = field_data.get("affected_area", 0)
-    if affected_area > 0:
-        target_zones = [
-            {
-                "zone_id": "A1",
-                "gps_boundary": [(40.71, -74.00), (40.72, -74.01)],
-                "priority": "high"
-            }
-        ]
-    else:
-        target_zones = []
-    
-    return {
-        "recommendation_id": f"SPR-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "field_id": field_data["field_id"],
-        "treatment_type": treatment,
-        "product_name": product,
-        "application_rate": rate,
-        "affected_area": affected_area,
-        "target_zones": target_zones,
-        "optimal_time": (datetime.now() + timedelta(days=1)).isoformat(),
-        "weather_window": {
-            "wind_speed": weather.get("wind_speed", 5),
-            "precipitation": weather.get("precipitation", 0),
-            "suitable": True
-        },
-        "cost_estimate": rate * affected_area * 50,  # $/hectare
-        "environmental_impact": impact
-    }
-
-
-@tool
-def predict_crop_yield(field_id: str, crop_type: str, current_health: Dict, weather_forecast: Dict) -> Dict:
-    """Predict crop yield based on current conditions and forecasts"""
-    # Simulate yield prediction
-    health_level = current_health.get("health_level", "good")
-    
-    base_yield = {
-        "wheat": 3.5,
-        "corn": 9.5,
-        "soybeans": 2.8,
-        "rice": 6.0
-    }.get(crop_type, 5.0)
-    
-    # Adjust based on health
-    health_multiplier = {
-        "excellent": 1.2,
-        "good": 1.0,
-        "fair": 0.8,
-        "poor": 0.6,
-        "critical": 0.4
-    }.get(health_level, 1.0)
-    
-    predicted_yield = base_yield * health_multiplier
-    
-    # Quality grade based on conditions
-    if health_level in ["excellent", "good"]:
-        quality = "A"
-        price_multiplier = 1.2
-    elif health_level == "fair":
-        quality = "B"
-        price_multiplier = 1.0
-    else:
-        quality = "C"
-        price_multiplier = 0.8
-    
-    return {
-        "forecast_id": f"YLD-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "field_id": field_id,
-        "crop_type": crop_type,
-        "predicted_yield": predicted_yield,
-        "confidence_interval": (predicted_yield * 0.9, predicted_yield * 1.1),
-        "harvest_window": (
-            (datetime.now() + timedelta(days=60)).isoformat(),
-            (datetime.now() + timedelta(days=75)).isoformat()
-        ),
-        "quality_grade": quality,
-        "market_price_estimate": predicted_yield * 250 * price_multiplier,  # $/ton
-        "risk_factors": ["weather_variability", "pest_pressure", "market_volatility"]
-    }
-
-
-# Agriculture agent definitions
-multispectral_agent = Agent(
-    name="MultispectralAnalyzer",
-    instructions="""You are a remote sensing specialist for precision agriculture.
-    Analyze multispectral imagery from drones and satellites.
-    Calculate vegetation indices (NDVI, NDRE, CCCI) to assess crop health.
-    Identify stress zones and anomalies in field conditions.
-    SLA: Process imagery within 2 minutes per field.""",
-    tools=[analyze_multispectral_imagery]
-)
-
-disease_agent = Agent(
-    name="DiseaseIdentifier",
-    instructions="""You are a plant pathology expert with AI vision capabilities.
-    Identify crop diseases, pests, and nutrient deficiencies.
-    Assess severity levels and predict spread patterns.
-    Consider environmental factors affecting disease development.
-    SLA: Complete identification within 30 seconds.""",
-    tools=[identify_pest_disease]
-)
-
-spray_agent = Agent(
-    name="SprayRecommender",
-    instructions="""You are a precision application specialist.
-    Recommend targeted spraying strategies for pest and disease control.
-    Optimize chemical usage to minimize environmental impact.
-    Consider weather windows and application regulations.
-    SLA: Generate recommendations within 1 minute.""",
-    tools=[calculate_spray_recommendation]
-)
-
-yield_agent = Agent(
-    name="YieldPredictor",
-    instructions="""You are a crop yield forecasting specialist.
-    Predict harvest yields based on current conditions and historical data.
-    Assess crop quality grades and market value estimates.
-    Identify risk factors that could impact final yield.
-    SLA: Generate forecast within 45 seconds.""",
-    tools=[predict_crop_yield]
-)
-
-
-# Precision agriculture workflow
-def precision_agriculture_workflow(field_ids: List[str], crop_type: str, weather_data: Dict):
-    """
-    Complete precision agriculture workflow from monitoring to treatment
-    Includes early warning system and sustainable farming practices
-    """
-    
-    workflow_results = {
-        "timestamp": datetime.now().isoformat(),
-        "fields_monitored": len(field_ids),
-        "alerts": [],
-        "treatments_needed": [],
-        "yield_forecasts": [],
-        "sustainability_score": 0.0
-    }
-    
-    for field_id in field_ids:
-        try:
-            # Step 1: Multispectral analysis
-            spectral_analysis = multispectral_agent.start(
-                f"Analyze multispectral imagery for field {field_id}"
-            )
-            
-            # Step 2: Disease and pest detection
-            pest_report = disease_agent.start(
-                f"Identify pests/diseases from spectral data: {spectral_analysis}"
-            )
-            
-            # Check for alerts
-            if pest_report.get("severity") in ["high", "severe"]:
-                workflow_results["alerts"].append({
-                    "field_id": field_id,
-                    "type": "pest_disease_alert",
-                    "severity": pest_report["severity"],
-                    "pest": pest_report.get("pest_type"),
-                    "disease": pest_report.get("disease_type"),
-                    "action": "immediate_treatment_required"
-                })
-                
-                # Step 3: Generate spray recommendation
-                spray_recommendation = spray_agent.start(
-                    f"Calculate spray strategy for pest report: {pest_report}, "
-                    f"field: {spectral_analysis}, weather: {weather_data}"
-                )
-                workflow_results["treatments_needed"].append(spray_recommendation)
-            
-            # Step 4: Yield prediction
-            yield_forecast = yield_agent.start(
-                f"Predict yield for field {field_id}, crop {crop_type}, "
-                f"health: {spectral_analysis}, weather: {weather_data}"
-            )
-            workflow_results["yield_forecasts"].append(yield_forecast)
-            
-        except Exception as e:
-            # Fallback: Schedule manual inspection
-            workflow_results["alerts"].append({
-                "field_id": field_id,
-                "type": "monitoring_failure",
-                "error": str(e),
-                "action": "schedule_manual_inspection"
-            })
-    
-    # Calculate sustainability score
-    total_chemical = sum(
-        t.get("application_rate", 0) * t.get("affected_area", 0) 
-        for t in workflow_results["treatments_needed"]
-    )
-    workflow_results["sustainability_score"] = max(0, 100 - (total_chemical * 10))
-    
-    return workflow_results
-
-
-# Sustainable farming patterns
-class SustainableFarmingPatterns:
-    """
-    Patterns for sustainable and regenerative agriculture
-    """
-    
-    @staticmethod
-    def integrated_pest_management(pest_level: str) -> Dict:
-        """IPM strategy based on pest pressure"""
-        strategies = {
-            "low": {
-                "method": "biological_control",
-                "actions": ["release_beneficial_insects", "plant_trap_crops"],
-                "chemical_use": 0
-            },
-            "moderate": {
-                "method": "targeted_intervention",
-                "actions": ["spot_treatment", "pheromone_traps", "minimal_spray"],
-                "chemical_use": 30
-            },
-            "high": {
-                "method": "integrated_approach",
-                "actions": ["systematic_spray", "crop_rotation", "resistant_varieties"],
-                "chemical_use": 70
-            }
-        }
-        return strategies.get(pest_level, strategies["moderate"])
-    
-    @staticmethod
-    def precision_irrigation(moisture_level: float, crop_stage: str) -> Dict:
-        """Calculate optimal irrigation based on soil moisture and crop stage"""
-        # Water requirements by growth stage (mm/day)
-        stage_requirements = {
-            "germination": 3.0,
-            "vegetative": 5.0,
-            "flowering": 7.0,
-            "grain_filling": 6.0,
-            "maturation": 2.0
-        }
-        
-        required = stage_requirements.get(crop_stage, 5.0)
-        deficit = max(0, 40 - moisture_level)  # Target 40% moisture
-        
-        return {
-            "irrigation_needed": deficit > 10,
-            "amount_mm": deficit * 2,
-            "method": "drip" if deficit < 20 else "sprinkler",
-            "schedule": "night_irrigation" if deficit > 15 else "morning_irrigation",
-            "water_saved": (100 - deficit) / 100 * 1000  # Liters per hectare
-        }
-    
-    @staticmethod
-    def crop_rotation_optimizer(current_crop: str, soil_data: Dict) -> str:
-        """Recommend next crop for rotation based on soil health"""
-        rotations = {
-            "wheat": ["soybeans", "canola"],  # Nitrogen fixation
-            "corn": ["soybeans", "alfalfa"],  # Break disease cycle
-            "soybeans": ["wheat", "corn"],  # Utilize fixed nitrogen
-            "rice": ["legumes", "vegetables"],  # Soil structure improvement
-        }
-        
-        return rotations.get(current_crop, ["cover_crops"])[0]
-
-
-# IoT sensor integration patterns
-class IoTSensorPatterns:
-    """
-    Patterns for integrating IoT sensors in smart farming
-    """
-    
-    @staticmethod
-    def create_sensor_network_agent(sensor_types: List[str]) -> Agent:
-        """Create agent for managing IoT sensor networks"""
-        return Agent(
-            name="SensorNetworkManager",
-            instructions=f"""Manage IoT sensor network with {', '.join(sensor_types)}.
-            Collect real-time data from field sensors.
-            Detect sensor anomalies and maintenance needs.
-            Aggregate data for precision agriculture decisions.
-            SLA: Data collection every 15 minutes."""
-        )
-    
-    @staticmethod
-    def process_sensor_data(sensor_readings: List[Dict]) -> Dict:
-        """Process and aggregate IoT sensor data"""
-        aggregated = {
-            "timestamp": datetime.now().isoformat(),
-            "sensor_count": len(sensor_readings),
-            "averages": {},
-            "alerts": []
-        }
-        
-        # Calculate averages
-        for reading in sensor_readings:
-            for key, value in reading.items():
-                if isinstance(value, (int, float)):
-                    if key not in aggregated["averages"]:
-                        aggregated["averages"][key] = []
-                    aggregated["averages"][key].append(value)
-        
-        # Convert to means
-        for key in aggregated["averages"]:
-            values = aggregated["averages"][key]
-            aggregated["averages"][key] = sum(values) / len(values)
-            
-            # Check for anomalies
-            if key == "soil_ph" and (aggregated["averages"][key] < 6.0 or aggregated["averages"][key] > 7.5):
-                aggregated["alerts"].append({
-                    "type": "soil_ph_anomaly",
-                    "value": aggregated["averages"][key],
-                    "action": "adjust_soil_treatment"
-                })
-        
-        return aggregated
-
-
-# Example usage
-if __name__ == "__main__":
-    # Monitor multiple fields
-    fields = ["FIELD-001", "FIELD-002", "FIELD-003"]
-    weather = {
-        "temperature": 25,
-        "humidity": 60,
-        "wind_speed": 8,
-        "precipitation": 0,
-        "forecast": "partly_cloudy"
-    }
-    
-    result = precision_agriculture_workflow(fields, "wheat", weather)
-    print("Agriculture workflow result:", result)
-    
-    # Sustainable farming example
-    ipm_strategy = SustainableFarmingPatterns.integrated_pest_management("moderate")
-    print("IPM Strategy:", ipm_strategy)
-    
-    irrigation_plan = SustainableFarmingPatterns.precision_irrigation(25.0, "flowering")
-    print("Irrigation Plan:", irrigation_plan)
-    
-    next_crop = SustainableFarmingPatterns.crop_rotation_optimizer("wheat", {})
-    print("Recommended next crop:", next_crop)
-    
-    # IoT integration example
-    sensor_agent = IoTSensorPatterns.create_sensor_network_agent(
-        ["soil_moisture", "temperature", "humidity", "light", "ph"]
-    )
-    
-    sensor_data = [
-        {"sensor_id": "S001", "soil_moisture": 35, "temperature": 24, "soil_ph": 6.8},
-        {"sensor_id": "S002", "soil_moisture": 32, "temperature": 25, "soil_ph": 7.2},
-        {"sensor_id": "S003", "soil_moisture": 38, "temperature": 23, "soil_ph": 5.8}
-    ]
-    
-    processed_data = IoTSensorPatterns.process_sensor_data(sensor_data)
-    print("Processed sensor data:", processed_data)
@@ -1,372 +0,0 @@
-"""
-Energy Industry Template
-========================
-Wind farm monitoring and power optimization workflow
-Based on SRAO Framework patterns with predictive maintenance
-
-Key agents:
- SCADAReader: Processes SCADA telemetry data
- VibrationAnalyzer: Analyzes turbine vibration patterns
- PowerForecaster: Predicts power generation
- MaintenanceScheduler: Schedules predictive maintenance
-"""
-
-from praisonaiagents import Agent, tool
-from typing import Dict, List, Any, Optional
-from pydantic import BaseModel
-from datetime import datetime, timedelta
-
-
-# I/O Schemas
-class TurbineData(BaseModel):
-    """Wind turbine telemetry data"""
-    turbine_id: str
-    timestamp: str
-    wind_speed: float  # m/s
-    power_output: float  # MW
-    rotor_speed: float  # RPM
-    blade_pitch: float  # degrees
-    nacelle_temperature: float  # Celsius
-    vibration_level: float  # mm/s
-    operational_status: str  # running, stopped, maintenance
-
-
-class VibrationAnalysis(BaseModel):
-    """Vibration analysis results"""
-    turbine_id: str
-    analysis_time: str
-    vibration_rms: float
-    frequency_peaks: List[float]
-    anomaly_detected: bool
-    failure_probability: float
-    recommended_action: str
-
-
-class PowerForecast(BaseModel):
-    """Power generation forecast"""
-    forecast_id: str
-    start_time: str
-    end_time: str
-    predicted_output: List[float]  # MW per hour
-    confidence_interval: float
-    weather_factors: Dict[str, Any]
-    grid_demand: float
-
-
-class MaintenancePlan(BaseModel):
-    """Maintenance scheduling plan"""
-    plan_id: str
-    turbine_id: str
-    maintenance_type: str  # preventive, predictive, corrective
-    scheduled_date: str
-    estimated_duration: int  # hours
-    required_parts: List[str]
-    technician_team: str
-    production_loss: float  # MW
-
-
-# Tools for energy management
-@tool
-def read_scada_telemetry(turbine_id: str, time_range: str = "last_hour") -> Dict:
-    """Read SCADA system telemetry for wind turbines"""
-    # Simulate SCADA data reading
-    return {
-        "turbine_id": turbine_id,
-        "timestamp": datetime.now().isoformat(),
-        "wind_speed": 12.5,
-        "power_output": 2.3,
-        "rotor_speed": 16.8,
-        "blade_pitch": 8.5,
-        "nacelle_temperature": 45.2,
-        "vibration_level": 3.2,
-        "operational_status": "running"
-    }
-
-
-@tool
-def analyze_vibration_patterns(turbine_data: Dict) -> Dict:
-    """Analyze vibration patterns for early fault detection"""
-    # Simulate vibration analysis
-    vibration_level = turbine_data.get("vibration_level", 0)
-    anomaly = vibration_level > 5.0  # Threshold for anomaly
-    
-    return {
-        "turbine_id": turbine_data["turbine_id"],
-        "analysis_time": datetime.now().isoformat(),
-        "vibration_rms": vibration_level,
-        "frequency_peaks": [10.5, 25.3, 50.0],  # Hz
-        "anomaly_detected": anomaly,
-        "failure_probability": min(vibration_level / 10.0, 1.0),
-        "recommended_action": "immediate_inspection" if anomaly else "continue_monitoring"
-    }
-
-
-@tool
-def forecast_power_generation(weather_data: Dict, historical_data: List) -> Dict:
-    """Forecast power generation based on weather and historical patterns"""
-    # Simulate power forecasting
-    base_forecast = [2.0 + (i * 0.1) for i in range(24)]  # 24-hour forecast
-    
-    return {
-        "forecast_id": f"FCST-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "start_time": datetime.now().isoformat(),
-        "end_time": (datetime.now() + timedelta(hours=24)).isoformat(),
-        "predicted_output": base_forecast,
-        "confidence_interval": 0.85,
-        "weather_factors": weather_data,
-        "grid_demand": sum(base_forecast)
-    }
-
-
-@tool
-def schedule_maintenance(turbine_id: str, maintenance_type: str, urgency: str = "normal") -> Dict:
-    """Schedule maintenance based on condition monitoring"""
-    # Calculate maintenance window
-    if urgency == "urgent":
-        scheduled_date = (datetime.now() + timedelta(days=1)).isoformat()
-    else:
-        scheduled_date = (datetime.now() + timedelta(days=7)).isoformat()
-    
-    return {
-        "plan_id": f"MNT-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "turbine_id": turbine_id,
-        "maintenance_type": maintenance_type,
-        "scheduled_date": scheduled_date,
-        "estimated_duration": 4,
-        "required_parts": ["bearing_set", "lubricant", "filter"],
-        "technician_team": "Team-A",
-        "production_loss": 8.0  # MW lost during maintenance
-    }
-
-
-# Agent Definitions with SLA requirements
-scada_reader = Agent(
-    name="SCADAReader",
-    instructions="""You are a SCADA system specialist for wind farms.
-    Process real-time telemetry data from wind turbines.
-    Identify operational anomalies and performance deviations.
-    SLA: Process telemetry within 1 second for real-time monitoring.""",
-    tools=[read_scada_telemetry]
-)
-
-vibration_analyzer = Agent(
-    name="VibrationAnalyzer",
-    instructions="""You are a vibration analysis expert for rotating machinery.
-    Analyze vibration patterns to detect early signs of mechanical failure.
-    Use FFT analysis and pattern recognition for fault diagnosis.
-    SLA: Complete analysis within 5 seconds per turbine.""",
-    tools=[analyze_vibration_patterns]
-)
-
-power_forecaster = Agent(
-    name="PowerForecaster",
-    instructions="""You are a power generation forecasting specialist.
-    Predict power output based on weather forecasts and historical patterns.
-    Consider grid demand and optimize generation schedules.
-    SLA: Generate 24-hour forecast within 30 seconds.""",
-    tools=[forecast_power_generation]
-)
-
-maintenance_scheduler = Agent(
-    name="MaintenanceScheduler",
-    instructions="""You are a predictive maintenance planning expert.
-    Schedule maintenance based on condition monitoring and failure predictions.
-    Minimize production loss while ensuring turbine reliability.
-    SLA: Create maintenance plan within 1 minute.""",
-    tools=[schedule_maintenance]
-)
-
-
-# Energy workflow with grid integration
-def energy_monitoring_workflow(turbine_ids: List[str], weather_forecast: Dict):
-    """
-    Complete energy monitoring and optimization workflow
-    Includes predictive maintenance and grid integration
-    """
-    
-    results = {
-        "timestamp": datetime.now().isoformat(),
-        "turbines_monitored": len(turbine_ids),
-        "alerts": [],
-        "maintenance_required": [],
-        "power_forecast": None
-    }
-    
-    # Step 1: Monitor all turbines
-    for turbine_id in turbine_ids:
-        try:
-            # Read SCADA data
-            scada_data = scada_reader.start(
-                f"Read telemetry for turbine {turbine_id}"
-            )
-            
-            # Analyze vibrations
-            vibration_result = vibration_analyzer.start(
-                f"Analyze vibrations for data: {scada_data}"
-            )
-            
-            # Check for anomalies
-            if vibration_result.get("anomaly_detected"):
-                results["alerts"].append({
-                    "turbine_id": turbine_id,
-                    "type": "vibration_anomaly",
-                    "severity": "high" if vibration_result.get("failure_probability", 0) > 0.7 else "medium"
-                })
-                
-                # Schedule maintenance if needed
-                maintenance_plan = maintenance_scheduler.start(
-                    f"Schedule maintenance for turbine {turbine_id} with urgency based on {vibration_result}"
-                )
-                results["maintenance_required"].append(maintenance_plan)
-                
-        except Exception as e:
-            # Fallback: Mark turbine for manual inspection
-            results["alerts"].append({
-                "turbine_id": turbine_id,
-                "type": "monitoring_failure",
-                "error": str(e),
-                "action": "manual_inspection_required"
-            })
-    
-    # Step 2: Power generation forecast
-    try:
-        historical_data = []  # Would be fetched from database
-        power_forecast = power_forecaster.start(
-            f"Forecast power with weather {weather_forecast} and history {historical_data}"
-        )
-        results["power_forecast"] = power_forecast
-        
-        # Grid integration check
-        if power_forecast.get("predicted_output"):
-            total_predicted = sum(power_forecast["predicted_output"])
-            grid_demand = power_forecast.get("grid_demand", 0)
-            
-            if total_predicted < grid_demand * 0.9:
-                results["alerts"].append({
-                    "type": "supply_shortage",
-                    "predicted": total_predicted,
-                    "demand": grid_demand,
-                    "action": "consider_backup_sources"
-                })
-                
-    except Exception as e:
-        # Fallback: Use conservative forecast
-        results["power_forecast"] = {
-            "status": "fallback_mode",
-            "conservative_estimate": len(turbine_ids) * 1.5,  # MW
-            "error": str(e)
-        }
-    
-    return results
-
-
-# Cross-industry pattern adaptation
-class EnergyPatternAdapter:
-    """
-    Adapts manufacturing patterns for energy sector
-    Demonstrates 70% code reuse across industries
-    """
-    
-    @staticmethod
-    def adapt_for_solar_farm():
-        """Adapt wind farm agents for solar farm monitoring"""
-        return Agent(
-            name="SolarMonitor",
-            instructions="""Monitor solar panel efficiency and temperature.
-            Track irradiance levels and predict power output.
-            Detect panel degradation and soiling issues.
-            SLA: Real-time monitoring with 2-second updates."""
-        )
-    
-    @staticmethod
-    def adapt_for_power_grid():
-        """Adapt for power grid management"""
-        return Agent(
-            name="GridBalancer",
-            instructions="""Balance power supply and demand across the grid.
-            Manage load distribution and prevent blackouts.
-            Coordinate with renewable and traditional sources.
-            SLA: Sub-second response for grid stability."""
-        )
-    
-    @staticmethod
-    def adapt_for_battery_storage():
-        """Adapt for battery energy storage systems"""
-        return Agent(
-            name="BatteryOptimizer",
-            instructions="""Optimize battery charge/discharge cycles.
-            Predict battery degradation and schedule replacements.
-            Maximize energy arbitrage opportunities.
-            SLA: Optimization decision within 10 seconds."""
-        )
-
-
-# Fallback strategies for critical operations
-class EnergyFallbackStrategies:
-    """
-    Fallback strategies for maintaining grid stability
-    """
-    
-    @staticmethod
-    def turbine_failure_fallback(failed_turbine_id: str, available_turbines: List[str]):
-        """Handle turbine failure by redistributing load"""
-        return {
-            "strategy": "load_redistribution",
-            "failed_turbine": failed_turbine_id,
-            "compensating_turbines": available_turbines[:3],  # Use top 3 turbines
-            "increase_output_by": 0.3,  # MW per turbine
-            "duration": "until_maintenance_complete"
-        }
-    
-    @staticmethod
-    def forecast_failure_fallback(last_known_forecast: Optional[Dict]):
-        """Handle forecasting system failure"""
-        if last_known_forecast:
-            return {
-                "strategy": "use_last_known",
-                "forecast": last_known_forecast,
-                "confidence": 0.6,
-                "validity": "4_hours"
-            }
-        else:
-            return {
-                "strategy": "seasonal_average",
-                "estimated_output": 50.0,  # MW based on seasonal average
-                "confidence": 0.4,
-                "recommendation": "increase_reserves"
-            }
-    
-    @staticmethod
-    def communication_failure_fallback():
-        """Handle SCADA communication failure"""
-        return {
-            "strategy": "local_control_mode",
-            "actions": [
-                "Switch turbines to autonomous mode",
-                "Use local safety thresholds",
-                "Log data locally for later sync",
-                "Alert maintenance teams"
-            ],
-            "recovery_check_interval": 300  # seconds
-        }
-
-
-# Example usage
-if __name__ == "__main__":
-    # Monitor wind farm
-    turbine_list = [f"WT-{i:03d}" for i in range(1, 11)]  # 10 turbines
-    weather = {"wind_speed": 14.0, "direction": "NW", "temperature": 15.0}
-    
-    result = energy_monitoring_workflow(turbine_list, weather)
-    print("Energy monitoring result:", result)
-    
-    # Demonstrate cross-industry adaptation
-    solar_monitor = EnergyPatternAdapter.adapt_for_solar_farm()
-    grid_balancer = EnergyPatternAdapter.adapt_for_power_grid()
-    battery_optimizer = EnergyPatternAdapter.adapt_for_battery_storage()
-    
-    # Fallback examples
-    turbine_fallback = EnergyFallbackStrategies.turbine_failure_fallback(
-        "WT-003", ["WT-001", "WT-002", "WT-004", "WT-005"]
-    )
-    print("Turbine failure fallback:", turbine_fallback)
@@ -1,478 +0,0 @@
-"""
-Healthcare Industry Template
-============================
-Emergency triage and patient care coordination workflow
-Based on SRAO Framework with HIPAA-compliant patterns
-
-Key agents:
- VitalSignsCapture: Processes patient vital signs
- EMRRetrieval: Retrieves electronic medical records
- TriageRecommendation: Provides triage recommendations
- ResourceAllocator: Manages hospital resource allocation
-"""
-
-from praisonaiagents import Agent, tool
-from typing import Dict, List, Any, Optional
-from pydantic import BaseModel
-from datetime import datetime
-from enum import Enum
-
-
-# Triage levels following Emergency Severity Index (ESI)
-class TriageLevel(str, Enum):
-    RESUSCITATION = "1_resuscitation"  # Immediate life-saving intervention
-    EMERGENT = "2_emergent"  # High risk, severe pain/distress
-    URGENT = "3_urgent"  # Stable but needs multiple resources
-    LESS_URGENT = "4_less_urgent"  # Stable, one resource needed
-    NON_URGENT = "5_non_urgent"  # Stable, no resources needed
-
-
-# I/O Schemas
-class VitalSigns(BaseModel):
-    """Patient vital signs data"""
-    patient_id: str
-    timestamp: str
-    heart_rate: int  # bpm
-    blood_pressure_systolic: int  # mmHg
-    blood_pressure_diastolic: int  # mmHg
-    respiratory_rate: int  # breaths/min
-    temperature: float  # Celsius
-    oxygen_saturation: int  # percentage
-    pain_scale: int  # 0-10
-    consciousness_level: str  # alert, verbal, pain, unresponsive (AVPU)
-
-
-class MedicalHistory(BaseModel):
-    """Patient medical history from EMR"""
-    patient_id: str
-    allergies: List[str]
-    chronic_conditions: List[str]
-    current_medications: List[Dict[str, str]]
-    recent_visits: List[Dict[str, Any]]
-    emergency_contacts: List[Dict[str, str]]
-    insurance_status: str
-
-
-class TriageAssessment(BaseModel):
-    """Triage assessment and recommendations"""
-    assessment_id: str
-    patient_id: str
-    triage_level: TriageLevel
-    chief_complaint: str
-    recommended_department: str
-    estimated_wait_time: int  # minutes
-    required_resources: List[str]
-    clinical_notes: str
-    red_flags: List[str]
-
-
-class ResourceAllocation(BaseModel):
-    """Hospital resource allocation"""
-    allocation_id: str
-    patient_id: str
-    assigned_bed: Optional[str]
-    assigned_staff: List[str]
-    equipment_needed: List[str]
-    department: str
-    priority_score: float
-    estimated_treatment_time: int  # minutes
-
-
-# Healthcare-specific tools
-@tool
-def capture_vital_signs(patient_id: str) -> Dict:
-    """Capture patient vital signs from monitoring devices"""
-    # Simulate vital signs capture
-    return {
-        "patient_id": patient_id,
-        "timestamp": datetime.now().isoformat(),
-        "heart_rate": 78,
-        "blood_pressure_systolic": 130,
-        "blood_pressure_diastolic": 85,
-        "respiratory_rate": 16,
-        "temperature": 37.2,
-        "oxygen_saturation": 98,
-        "pain_scale": 6,
-        "consciousness_level": "alert"
-    }
-
-
-@tool
-def retrieve_medical_history(patient_id: str, consent_verified: bool = True) -> Dict:
-    """Retrieve patient medical history from EMR system"""
-    if not consent_verified:
-        return {"error": "consent_required", "message": "Patient consent required for EMR access"}
-    
-    # Simulate EMR retrieval
-    return {
-        "patient_id": patient_id,
-        "allergies": ["penicillin", "latex"],
-        "chronic_conditions": ["hypertension", "diabetes_type2"],
-        "current_medications": [
-            {"name": "metformin", "dose": "500mg", "frequency": "twice daily"},
-            {"name": "lisinopril", "dose": "10mg", "frequency": "once daily"}
-        ],
-        "recent_visits": [
-            {"date": "2024-02-15", "reason": "routine_checkup", "department": "primary_care"}
-        ],
-        "emergency_contacts": [
-            {"name": "Jane Doe", "relationship": "spouse", "phone": "555-0123"}
-        ],
-        "insurance_status": "active"
-    }
-
-
-@tool
-def calculate_triage_priority(vital_signs: Dict, medical_history: Dict, chief_complaint: str) -> Dict:
-    """Calculate triage priority based on clinical indicators"""
-    # Simulate triage algorithm
-    priority_score = 0
-    red_flags = []
-    
-    # Check vital signs
-    if vital_signs["heart_rate"] > 120 or vital_signs["heart_rate"] < 50:
-        priority_score += 3
-        red_flags.append("abnormal_heart_rate")
-    
-    if vital_signs["oxygen_saturation"] < 92:
-        priority_score += 4
-        red_flags.append("low_oxygen")
-    
-    if vital_signs["consciousness_level"] != "alert":
-        priority_score += 5
-        red_flags.append("altered_consciousness")
-    
-    # Determine triage level
-    if priority_score >= 8:
-        triage_level = TriageLevel.RESUSCITATION
-        department = "emergency"
-        wait_time = 0
-    elif priority_score >= 5:
-        triage_level = TriageLevel.EMERGENT
-        department = "emergency"
-        wait_time = 10
-    elif priority_score >= 3:
-        triage_level = TriageLevel.URGENT
-        department = "urgent_care"
-        wait_time = 30
-    else:
-        triage_level = TriageLevel.LESS_URGENT
-        department = "primary_care"
-        wait_time = 60
-    
-    return {
-        "assessment_id": f"TRG-{datetime.now().strftime('%Y%m%d%H%M%S')}",
-        "patient_id": vital_signs["patient_id"],
-        "triage_level": triage_level.value,
-        "chief_complaint": chief_complaint,
-        "recommended_department": department,
-        "estimated_wait_time": wait_time,
-        "required_resources": ["physician", "nurse", "monitoring"],
-        "clinical_notes": f"Priority score: {priority_score}",
-        "red_flags": red_flags
-    }
-
-
-@tool
-def allocate_resources(patient_id: str, triage_level: str, department: str) -> Dict:
-    """Allocate hospital resources based on triage assessment"""
-    # Simulate resource allocation
-    if "resuscitation" in triage_level:
-        bed = "ER-01"
-        staff = ["Dr. Smith", "Nurse Johnson", "Nurse Williams"]
-        equipment = ["cardiac_monitor", "defibrillator", "crash_cart"]
-        priority = 10.0
-    elif "emergent" in triage_level:
-        bed = "ER-05"
-        staff = ["Dr. Brown", "Nurse Davis"]
-        equipment = ["vital_signs_monitor", "IV_pump"]
-        priority = 8.0
-    else:
-        bed = "UC-12"
-        staff = ["PA Miller", "Nurse Garcia"]
-        equipment = ["basic_monitor"]
-        priority = 5.0
-    
-    return {
-        "allocation_id": f"RES-{datetime.now().strftime('%Y%m%d%H%M%S')}",
-        "patient_id": patient_id,
-        "assigned_bed": bed,
-        "assigned_staff": staff,
-        "equipment_needed": equipment,
-        "department": department,
-        "priority_score": priority,
-        "estimated_treatment_time": 45
-    }
-
-
-# Healthcare agent definitions with SLA requirements
-vital_signs_agent = Agent(
-    name="VitalSignsCapture",
-    instructions="""You are a clinical vital signs specialist.
-    Capture and interpret patient vital signs from various monitoring devices.
-    Identify abnormal values and trends that require immediate attention.
-    Follow HIPAA guidelines for patient data handling.
-    SLA: Capture and process within 30 seconds.""",
-    tools=[capture_vital_signs]
-)
-
-emr_agent = Agent(
-    name="EMRRetrieval",
-    instructions="""You are an EMR system specialist with HIPAA compliance expertise.
-    Retrieve relevant patient medical history while ensuring data privacy.
-    Verify patient consent and maintain audit trails.
-    Highlight relevant allergies and contraindications.
-    SLA: Retrieve records within 5 seconds.""",
-    tools=[retrieve_medical_history]
-)
-
-triage_agent = Agent(
-    name="TriageRecommendation",
-    instructions="""You are an emergency triage specialist following ESI guidelines.
-    Assess patient severity and provide triage recommendations.
-    Consider vital signs, medical history, and presenting symptoms.
-    Identify red flags that require immediate intervention.
-    SLA: Complete assessment within 1 minute.""",
-    tools=[calculate_triage_priority]
-)
-
-resource_agent = Agent(
-    name="ResourceAllocator",
-    instructions="""You are a hospital resource management specialist.
-    Allocate beds, staff, and equipment based on patient needs and priority.
-    Optimize resource utilization while maintaining quality of care.
-    Coordinate between departments for efficient patient flow.
-    SLA: Allocate resources within 30 seconds.""",
-    tools=[allocate_resources]
-)
-
-
-# Healthcare workflow with safety protocols
-def emergency_triage_workflow(patient_id: str, chief_complaint: str):
-    """
-    Complete emergency triage workflow from arrival to resource allocation
-    Includes safety checks and fallback protocols
-    """
-    
-    workflow_result = {
-        "patient_id": patient_id,
-        "timestamp": datetime.now().isoformat(),
-        "status": "in_progress",
-        "safety_checks": [],
-        "clinical_pathway": []
-    }
-    
-    # Step 1: Vital signs capture with validation
-    try:
-        vital_signs = vital_signs_agent.start(
-            f"Capture vital signs for patient {patient_id}"
-        )
-        workflow_result["clinical_pathway"].append({
-            "step": "vital_signs",
-            "status": "completed",
-            "data": vital_signs
-        })
-        
-        # Safety check: Validate vital signs are within possible ranges
-        if vital_signs.get("heart_rate", 0) < 20 or vital_signs.get("heart_rate", 0) > 300:
-            workflow_result["safety_checks"].append({
-                "type": "vital_signs_validation",
-                "issue": "impossible_heart_rate",
-                "action": "manual_verification_required"
-            })
-    except Exception as e:
-        # Fallback: Manual vital signs entry
-        workflow_result["clinical_pathway"].append({
-            "step": "vital_signs",
-            "status": "manual_entry_required",
-            "error": str(e)
-        })
-        vital_signs = {"manual_entry": True}
-    
-    # Step 2: EMR retrieval with consent check
-    try:
-        medical_history = emr_agent.start(
-            f"Retrieve medical history for patient {patient_id} with verified consent"
-        )
-        workflow_result["clinical_pathway"].append({
-            "step": "emr_retrieval",
-            "status": "completed",
-            "records_retrieved": True
-        })
-        
-        # Safety check: Allergy alerts
-        if medical_history.get("allergies"):
-            workflow_result["safety_checks"].append({
-                "type": "allergy_alert",
-                "allergies": medical_history["allergies"],
-                "action": "notify_all_providers"
-            })
-    except Exception as e:
-        # Fallback: Proceed with limited information
-        medical_history = {
-            "patient_id": patient_id,
-            "allergies": [],
-            "note": "EMR_unavailable_proceed_with_caution"
-        }
-        workflow_result["safety_checks"].append({
-            "type": "emr_unavailable",
-            "action": "collect_history_manually"
-        })
-    
-    # Step 3: Triage assessment
-    try:
-        triage_assessment = triage_agent.start(
-            f"Assess triage for patient with vitals {vital_signs}, "
-            f"history {medical_history}, complaint: {chief_complaint}"
-        )
-        workflow_result["clinical_pathway"].append({
-            "step": "triage_assessment",
-            "status": "completed",
-            "triage_level": triage_assessment.get("triage_level")
-        })
-        
-        # Safety check: Red flags requiring immediate action
-        if triage_assessment.get("red_flags"):
-            workflow_result["safety_checks"].append({
-                "type": "clinical_red_flags",
-                "flags": triage_assessment["red_flags"],
-                "action": "immediate_physician_notification"
-            })
-    except Exception as e:
-        # Fallback: Conservative triage (assume urgent)
-        triage_assessment = {
-            "triage_level": TriageLevel.URGENT.value,
-            "recommended_department": "emergency",
-            "fallback": True
-        }
-    
-    # Step 4: Resource allocation
-    try:
-        resource_allocation = resource_agent.start(
-            f"Allocate resources for patient {patient_id} "
-            f"with triage {triage_assessment.get('triage_level')} "
-            f"to {triage_assessment.get('recommended_department')}"
-        )
-        workflow_result["clinical_pathway"].append({
-            "step": "resource_allocation",
-            "status": "completed",
-            "bed_assigned": resource_allocation.get("assigned_bed"),
-            "staff_assigned": resource_allocation.get("assigned_staff")
-        })
-    except Exception as e:
-        # Fallback: Queue for next available
-        workflow_result["clinical_pathway"].append({
-            "step": "resource_allocation",
-            "status": "queued",
-            "queue_position": "next_available",
-            "error": str(e)
-        })
-    
-    workflow_result["status"] = "completed"
-    return workflow_result
-
-
-# HIPAA-compliant data handling patterns
-class HIPAACompliancePatterns:
-    """
-    Reusable patterns for HIPAA-compliant healthcare workflows
-    """
-    
-    @staticmethod
-    def anonymize_patient_data(data: Dict) -> Dict:
-        """Remove PHI for analytics and reporting"""
-        safe_fields = ["triage_level", "department", "wait_time", "resource_type"]
-        return {k: v for k, v in data.items() if k in safe_fields}
-    
-    @staticmethod
-    def audit_log_access(user_id: str, patient_id: str, action: str, reason: str):
-        """Create audit trail for EMR access"""
-        return {
-            "timestamp": datetime.now().isoformat(),
-            "user_id": user_id,
-            "patient_id": patient_id,
-            "action": action,
-            "reason": reason,
-            "ip_address": "10.0.0.1",  # Would be actual IP
-            "session_id": "SEC-" + datetime.now().strftime('%Y%m%d%H%M%S')
-        }
-    
-    @staticmethod
-    def encrypt_sensitive_data(data: Dict) -> Dict:
-        """Encrypt sensitive fields (placeholder for real encryption)"""
-        sensitive_fields = ["ssn", "dob", "address", "phone"]
-        encrypted = data.copy()
-        for field in sensitive_fields:
-            if field in encrypted:
-                encrypted[field] = f"ENCRYPTED_{field.upper()}"
-        return encrypted
-
-
-# Cross-department coordination patterns
-class HealthcareCoordinationPatterns:
-    """
-    Patterns for coordinating across healthcare departments
-    """
-    
-    @staticmethod
-    def coordinate_lab_orders(patient_id: str, triage_level: str) -> Agent:
-        """Create agent for lab order coordination"""
-        return Agent(
-            name="LabCoordinator",
-            instructions=f"""Coordinate laboratory orders for patient {patient_id}.
-            Priority level: {triage_level}.
-            Ensure STAT orders for critical patients.
-            Track specimen collection and result delivery.
-            SLA: Order placement within 2 minutes."""
-        )
-    
-    @staticmethod
-    def coordinate_radiology(patient_id: str, imaging_type: str) -> Agent:
-        """Create agent for radiology coordination"""
-        return Agent(
-            name="RadiologyCoordinator",
-            instructions=f"""Schedule {imaging_type} imaging for patient {patient_id}.
-            Check for contrast allergies and pregnancy status.
-            Coordinate with transport for patient movement.
-            SLA: Schedule within 5 minutes for urgent cases."""
-        )
-    
-    @staticmethod
-    def coordinate_pharmacy(patient_id: str, medications: List[str]) -> Agent:
-        """Create agent for pharmacy coordination"""
-        return Agent(
-            name="PharmacyCoordinator",
-            instructions=f"""Verify and dispense medications for patient {patient_id}.
-            Check for drug interactions and allergies.
-            Ensure proper dosing based on patient parameters.
-            SLA: Medication verification within 3 minutes."""
-        )
-
-
-# Example usage
-if __name__ == "__main__":
-    # Process emergency patient
-    patient_id = "PT-2024-00123"
-    chief_complaint = "severe chest pain radiating to left arm"
-    
-    result = emergency_triage_workflow(patient_id, chief_complaint)
-    print("Triage workflow result:", result)
-    
-    # Create specialized coordinators
-    lab_coordinator = HealthcareCoordinationPatterns.coordinate_lab_orders(
-        patient_id, TriageLevel.EMERGENT.value
-    )
-    
-    radiology_coordinator = HealthcareCoordinationPatterns.coordinate_radiology(
-        patient_id, "chest_xray"
-    )
-    
-    pharmacy_coordinator = HealthcareCoordinationPatterns.coordinate_pharmacy(
-        patient_id, ["aspirin", "nitroglycerin"]
-    )
-    
-    # HIPAA compliance example
-    audit_log = HIPAACompliancePatterns.audit_log_access(
-        "DR-001", patient_id, "EMR_ACCESS", "emergency_triage"
-    )
-    print("Audit log created:", audit_log)
@@ -1,298 +0,0 @@
-"""
-Manufacturing Industry Template
-================================
-Order scheduling, inventory management, and quality inspection workflow
-Based on SRAO Framework patterns with 70% cross-industry reuse
-
-Key agents:
- ParseOrder: Extracts order details from various formats
- CheckInventory: Validates material availability
- OptimizeSchedule: Optimizes production scheduling
- DefectDetect: Quality control and defect detection
-"""
-
-from praisonaiagents import Agent, tool
-from typing import Dict, List, Any
-from pydantic import BaseModel
-
-
-# I/O Schemas
-class OrderDetails(BaseModel):
-    """Structured order information"""
-    order_id: str
-    customer_id: str
-    product_id: str
-    quantity: int
-    priority: str = "normal"  # normal, high, urgent
-    due_date: str
-    requirements: Dict[str, Any] = {}
-
-
-class InventoryStatus(BaseModel):
-    """Material availability status"""
-    material_id: str
-    available_quantity: int
-    reorder_level: int
-    supplier_lead_time: int
-    location: str
-
-
-class ProductionSchedule(BaseModel):
-    """Optimized production schedule"""
-    schedule_id: str
-    production_line: str
-    start_time: str
-    end_time: str
-    assigned_workers: List[str]
-    estimated_completion: str
-    bottleneck_analysis: Dict[str, Any] = {}
-
-
-class QualityReport(BaseModel):
-    """Quality inspection results"""
-    batch_id: str
-    inspection_time: str
-    defect_rate: float
-    defect_types: List[str]
-    passed: bool
-    recommendations: List[str]
-
-
-# Tools for agents
-@tool
-def extract_order_details(text: str) -> Dict:
-    """Extract structured order information from unstructured text"""
-    # Simulate order parsing logic
-    return {
-        "order_id": "ORD-2024-001",
-        "customer_id": "CUST-123",
-        "product_id": "PROD-456",
-        "quantity": 100,
-        "priority": "high",
-        "due_date": "2024-03-15",
-        "requirements": {"material": "steel", "finish": "polished"}
-    }
-
-
-@tool
-def check_material_availability(product_id: str, quantity: int) -> Dict:
-    """Check if materials are available for production"""
-    # Simulate inventory check
-    return {
-        "material_id": "MAT-789",
-        "available_quantity": 500,
-        "reorder_level": 100,
-        "supplier_lead_time": 7,
-        "location": "Warehouse-A"
-    }
-
-
-@tool
-def calculate_optimal_schedule(order: Dict, inventory: Dict) -> Dict:
-    """Calculate optimal production schedule based on constraints"""
-    # Simulate scheduling optimization
-    return {
-        "schedule_id": "SCH-2024-001",
-        "production_line": "Line-3",
-        "start_time": "2024-03-10T08:00:00",
-        "end_time": "2024-03-14T17:00:00",
-        "assigned_workers": ["Worker-001", "Worker-002"],
-        "estimated_completion": "2024-03-14T16:30:00",
-        "bottleneck_analysis": {
-            "critical_path": "Assembly",
-            "buffer_time": 2.5
-        }
-    }
-
-
-@tool
-def perform_quality_inspection(batch_id: str) -> Dict:
-    """Perform automated quality inspection on production batch"""
-    # Simulate quality control
-    return {
-        "batch_id": batch_id,
-        "inspection_time": "2024-03-14T17:00:00",
-        "defect_rate": 0.02,
-        "defect_types": ["surface_scratch", "dimension_variance"],
-        "passed": True,
-        "recommendations": [
-            "Adjust machine calibration on Line-3",
-            "Increase inspection frequency for next batch"
-        ]
-    }
-
-
-# Agent Definitions with SLA requirements
-order_parser = Agent(
-    name="ParseOrder",
-    instructions="""You are an order parsing specialist. Extract and structure order information 
-    from various formats (email, PDF, API). Ensure all required fields are captured and validated.
-    SLA: Process within 30 seconds per order.""",
-    tools=[extract_order_details]
-)
-
-inventory_checker = Agent(
-    name="CheckInventory",
-    instructions="""You are an inventory management specialist. Verify material availability,
-    check reorder points, and identify potential supply chain issues.
-    SLA: Real-time response within 5 seconds.""",
-    tools=[check_material_availability]
-)
-
-schedule_optimizer = Agent(
-    name="OptimizeSchedule",
-    instructions="""You are a production scheduling expert. Optimize production schedules
-    based on order priority, resource availability, and constraints. Minimize idle time
-    and maximize throughput.
-    SLA: Generate schedule within 2 minutes for complex orders.""",
-    tools=[calculate_optimal_schedule]
-)
-
-quality_inspector = Agent(
-    name="DefectDetect",
-    instructions="""You are a quality control specialist. Perform automated inspections,
-    detect defects using vision analysis, and provide improvement recommendations.
-    SLA: Complete inspection within 1 minute per batch.""",
-    tools=[perform_quality_inspection]
-)
-
-
-# Workflow with fallback strategies
-def manufacturing_workflow(order_text: str):
-    """
-    Complete manufacturing workflow from order to quality control
-    Includes fallback strategies for each critical step
-    """
-    
-    # Step 1: Parse order with fallback
-    try:
-        order_details = order_parser.start(f"Parse this order: {order_text}")
-        if not order_details:
-            # Fallback: Manual review queue
-            return {"status": "manual_review_required", "reason": "order_parsing_failed"}
-    except Exception as e:
-        return {"status": "error", "stage": "order_parsing", "error": str(e)}
-    
-    # Step 2: Check inventory with fallback
-    try:
-        inventory_status = inventory_checker.start(
-            f"Check materials for order: {order_details}"
-        )
-        if not inventory_status:
-            # Fallback: Alternative supplier check
-            return {"status": "alternative_supplier_needed", "order": order_details}
-    except Exception as e:
-        # Fallback: Conservative estimate with safety margin
-        inventory_status = {"estimated": True, "safety_margin": 1.5}
-    
-    # Step 3: Optimize schedule with fallback
-    try:
-        schedule = schedule_optimizer.start(
-            f"Create schedule for order {order_details} with inventory {inventory_status}"
-        )
-        if not schedule:
-            # Fallback: Standard scheduling template
-            schedule = {"template": "standard", "priority": "normal"}
-    except Exception as e:
-        # Fallback: Queue for next available slot
-        schedule = {"status": "queued", "estimated_start": "next_available"}
-    
-    # Step 4: Quality inspection with fallback
-    try:
-        quality_report = quality_inspector.start(
-            f"Inspect batch for schedule {schedule}"
-        )
-        if not quality_report:
-            # Fallback: Manual inspection required
-            quality_report = {"requires_manual": True, "inspector_assigned": "QC-Team-1"}
-    except Exception as e:
-        # Fallback: Quarantine for review
-        quality_report = {"status": "quarantined", "review_required": True}
-    
-    return {
-        "status": "completed",
-        "order": order_details,
-        "inventory": inventory_status,
-        "schedule": schedule,
-        "quality": quality_report
-    }
-
-
-# Cross-industry reusable patterns (70% reuse)
-class IndustryAgentPattern:
-    """
-    Base pattern reusable across industries
-    Provides common agent capabilities that can be specialized
-    """
-    
-    @staticmethod
-    def create_data_parser(name: str, domain: str, sla_seconds: int = 30):
-        """Create a data parsing agent for any domain"""
-        return Agent(
-            name=name,
-            instructions=f"""You are a {domain} data parsing specialist.
-            Extract and structure information from various formats.
-            Validate all required fields according to {domain} standards.
-            SLA: Process within {sla_seconds} seconds."""
-        )
-    
-    @staticmethod
-    def create_resource_checker(name: str, resource_type: str, sla_seconds: int = 5):
-        """Create a resource availability checker for any resource type"""
-        return Agent(
-            name=name,
-            instructions=f"""You are a {resource_type} availability specialist.
-            Verify {resource_type} availability and identify constraints.
-            Provide real-time status and alternative options if needed.
-            SLA: Response within {sla_seconds} seconds."""
-        )
-    
-    @staticmethod
-    def create_optimizer(name: str, optimization_target: str, sla_minutes: int = 2):
-        """Create an optimization agent for any target metric"""
-        return Agent(
-            name=name,
-            instructions=f"""You are an optimization expert for {optimization_target}.
-            Find optimal solutions considering all constraints and priorities.
-            Minimize waste and maximize efficiency.
-            SLA: Generate solution within {sla_minutes} minutes."""
-        )
-    
-    @staticmethod
-    def create_inspector(name: str, inspection_type: str, sla_minutes: int = 1):
-        """Create an inspection agent for any quality metric"""
-        return Agent(
-            name=name,
-            instructions=f"""You are a {inspection_type} inspection specialist.
-            Perform automated analysis and detect anomalies.
-            Provide actionable recommendations for improvement.
-            SLA: Complete inspection within {sla_minutes} minute(s)."""
-        )
-
-
-# Example usage
-if __name__ == "__main__":
-    # Create a manufacturing workflow instance
-    order = "Customer ABC needs 100 units of Product-XYZ by March 15th, high priority"
-    
-    # Option 1: Use specialized agents
-    result = manufacturing_workflow(order)
-    print("Manufacturing workflow result:", result)
-    
-    # Option 2: Use reusable patterns for other industries
-    # These patterns can be adapted for Energy, Healthcare, Agriculture, Transportation
-    energy_parser = IndustryAgentPattern.create_data_parser(
-        "EnergyDataParser", "wind farm telemetry", sla_seconds=10
-    )
-    
-    healthcare_checker = IndustryAgentPattern.create_resource_checker(
-        "BedAvailability", "hospital bed", sla_seconds=3
-    )
-    
-    agriculture_optimizer = IndustryAgentPattern.create_optimizer(
-        "IrrigationOptimizer", "water usage", sla_minutes=5
-    )
-    
-    transport_inspector = IndustryAgentPattern.create_inspector(
-        "TunnelSafety", "structural integrity", sla_minutes=2
-    )
@@ -1,599 +0,0 @@
-"""
-Transportation Industry Template
-================================
-Tunnel safety monitoring and infrastructure management workflow
-Based on SRAO Framework with LiDAR and sensor fusion
-
-Key agents:
- LiDARFusion: Processes LiDAR point cloud data
- DisplacementCalculator: Calculates structural displacement
- HeatmapGenerator: Generates safety heatmaps
- MaintenancePlanner: Plans infrastructure maintenance
-"""
-
-from praisonaiagents import Agent, tool
-from typing import Dict, List, Any, Tuple, Optional
-from pydantic import BaseModel
-from datetime import datetime, timedelta
-from enum import Enum
-import math
-
-
-# Safety levels for infrastructure
-class SafetyLevel(str, Enum):
-    SAFE = "safe"
-    CAUTION = "caution"
-    WARNING = "warning"
-    DANGER = "danger"
-    CRITICAL = "critical"
-
-
-# Infrastructure types
-class InfrastructureType(str, Enum):
-    TUNNEL = "tunnel"
-    BRIDGE = "bridge"
-    HIGHWAY = "highway"
-    RAILWAY = "railway"
-    AIRPORT = "airport"
-
-
-# I/O Schemas
-class LiDARScan(BaseModel):
-    """LiDAR scan data for infrastructure"""
-    scan_id: str
-    infrastructure_id: str
-    scan_time: str
-    point_cloud_size: int  # Number of points
-    scan_resolution: float  # Points per square meter
-    coverage_area: float  # Square meters
-    anomaly_points: List[Tuple[float, float, float]]  # XYZ coordinates
-    reference_baseline: Optional[Dict[str, Any]]
-
-
-class DisplacementAnalysis(BaseModel):
-    """Structural displacement analysis"""
-    analysis_id: str
-    infrastructure_id: str
-    analysis_time: str
-    max_displacement: float  # millimeters
-    displacement_vector: Tuple[float, float, float]  # XYZ displacement
-    critical_zones: List[Dict[str, Any]]
-    settlement_rate: float  # mm/month
-    tilt_angle: float  # degrees
-    safety_factor: float
-
-
-class SafetyHeatmap(BaseModel):
-    """Infrastructure safety heatmap"""
-    heatmap_id: str
-    infrastructure_id: str
-    generation_time: str
-    grid_resolution: float  # meters
-    safety_scores: List[List[float]]  # 2D grid of safety scores
-    critical_areas: List[Dict[str, Any]]
-    overall_safety: SafetyLevel
-    risk_trends: Dict[str, Any]
-
-
-class MaintenanceSchedule(BaseModel):
-    """Maintenance planning schedule"""
-    schedule_id: str
-    infrastructure_id: str
-    maintenance_type: str  # preventive, corrective, emergency
-    priority_level: str  # low, medium, high, critical
-    scheduled_date: str
-    estimated_duration: int  # hours
-    required_resources: List[str]
-    traffic_impact: str
-    cost_estimate: float
-
-
-# Transportation-specific tools
-@tool
-def process_lidar_scan(infrastructure_id: str, scan_type: str = "full") -> Dict:
-    """Process LiDAR point cloud data for structural analysis"""
-    # Simulate LiDAR processing
-    return {
-        "scan_id": f"LIDAR-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "infrastructure_id": infrastructure_id,
-        "scan_time": datetime.now().isoformat(),
-        "point_cloud_size": 5000000,  # 5 million points
-        "scan_resolution": 100.0,  # points/m²
-        "coverage_area": 50000.0,  # m²
-        "anomaly_points": [
-            (10.5, 20.3, 5.1),
-            (15.2, 25.8, 5.3),
-            (20.1, 30.5, 5.5)
-        ],
-        "reference_baseline": {
-            "date": "2024-01-01",
-            "baseline_id": "BSL-001"
-        }
-    }
-
-
-@tool
-def calculate_displacement(current_scan: Dict, baseline_scan: Dict) -> Dict:
-    """Calculate structural displacement from LiDAR scans"""
-    # Simulate displacement calculation
-    anomalies = current_scan.get("anomaly_points", [])
-    
-    if len(anomalies) > 5:
-        max_displacement = 15.0
-        safety_factor = 0.6
-        safety_level = SafetyLevel.WARNING
-    elif len(anomalies) > 2:
-        max_displacement = 8.0
-        safety_factor = 0.8
-        safety_level = SafetyLevel.CAUTION
-    else:
-        max_displacement = 3.0
-        safety_factor = 0.95
-        safety_level = SafetyLevel.SAFE
-    
-    return {
-        "analysis_id": f"DISP-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "infrastructure_id": current_scan["infrastructure_id"],
-        "analysis_time": datetime.now().isoformat(),
-        "max_displacement": max_displacement,
-        "displacement_vector": (2.5, 1.8, max_displacement),
-        "critical_zones": [
-            {
-                "zone_id": "CZ-001",
-                "location": "tunnel_crown",
-                "displacement": max_displacement * 0.8,
-                "risk": "high" if max_displacement > 10 else "medium"
-            }
-        ],
-        "settlement_rate": max_displacement / 30,  # mm/month
-        "tilt_angle": math.degrees(math.atan(max_displacement / 1000)),
-        "safety_factor": safety_factor
-    }
-
-
-@tool
-def generate_safety_heatmap(displacement_data: Dict, sensor_data: List[Dict]) -> Dict:
-    """Generate safety heatmap visualization for infrastructure"""
-    # Simulate heatmap generation
-    safety_factor = displacement_data.get("safety_factor", 1.0)
-    
-    # Create grid (simplified 10x10)
-    grid_size = 10
-    safety_grid = []
-    critical_areas = []
-    
-    for i in range(grid_size):
-        row = []
-        for j in range(grid_size):
-            # Generate safety score based on position and displacement
-            base_score = safety_factor
-            distance_factor = math.sqrt((i-5)**2 + (j-5)**2) / 10
-            safety_score = min(1.0, base_score + (0.2 * distance_factor))
-            row.append(safety_score)
-            
-            if safety_score < 0.7:
-                critical_areas.append({
-                    "grid_position": (i, j),
-                    "safety_score": safety_score,
-                    "recommended_action": "immediate_inspection"
-                })
-        safety_grid.append(row)
-    
-    # Determine overall safety
-    avg_safety = sum(sum(row) for row in safety_grid) / (grid_size * grid_size)
-    if avg_safety > 0.9:
-        overall = SafetyLevel.SAFE
-    elif avg_safety > 0.8:
-        overall = SafetyLevel.CAUTION
-    elif avg_safety > 0.7:
-        overall = SafetyLevel.WARNING
-    else:
-        overall = SafetyLevel.DANGER
-    
-    return {
-        "heatmap_id": f"HEAT-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "infrastructure_id": displacement_data["infrastructure_id"],
-        "generation_time": datetime.now().isoformat(),
-        "grid_resolution": 5.0,  # meters
-        "safety_scores": safety_grid,
-        "critical_areas": critical_areas,
-        "overall_safety": overall.value,
-        "risk_trends": {
-            "trend": "deteriorating" if avg_safety < 0.8 else "stable",
-            "rate": -0.02 if avg_safety < 0.8 else 0.0
-        }
-    }
-
-
-@tool
-def plan_maintenance(infrastructure_id: str, safety_analysis: Dict, traffic_data: Dict) -> Dict:
-    """Plan maintenance based on safety analysis and traffic patterns"""
-    safety_level = safety_analysis.get("overall_safety", "safe")
-    
-    # Determine maintenance urgency
-    if safety_level in ["danger", "critical"]:
-        maintenance_type = "emergency"
-        priority = "critical"
-        days_until = 1
-        duration = 48
-    elif safety_level == "warning":
-        maintenance_type = "corrective"
-        priority = "high"
-        days_until = 7
-        duration = 24
-    elif safety_level == "caution":
-        maintenance_type = "preventive"
-        priority = "medium"
-        days_until = 30
-        duration = 12
-    else:
-        maintenance_type = "preventive"
-        priority = "low"
-        days_until = 90
-        duration = 8
-    
-    # Traffic impact assessment
-    peak_traffic = traffic_data.get("peak_volume", 1000)
-    if peak_traffic > 5000:
-        traffic_impact = "severe"
-        cost_multiplier = 2.0
-    elif peak_traffic > 2000:
-        traffic_impact = "moderate"
-        cost_multiplier = 1.5
-    else:
-        traffic_impact = "minimal"
-        cost_multiplier = 1.0
-    
-    return {
-        "schedule_id": f"MAINT-{datetime.now().strftime('%Y%m%d%H%M')}",
-        "infrastructure_id": infrastructure_id,
-        "maintenance_type": maintenance_type,
-        "priority_level": priority,
-        "scheduled_date": (datetime.now() + timedelta(days=days_until)).isoformat(),
-        "estimated_duration": duration,
-        "required_resources": [
-            "inspection_team",
-            "repair_crew",
-            "safety_equipment",
-            "traffic_control"
-        ],
-        "traffic_impact": traffic_impact,
-        "cost_estimate": 50000 * cost_multiplier * (duration / 24)
-    }
-
-
-# Transportation agent definitions
-lidar_agent = Agent(
-    name="LiDARFusion",
-    instructions="""You are a LiDAR data processing specialist for infrastructure.
-    Process point cloud data from mobile and terrestrial LiDAR scanners.
-    Perform data fusion from multiple sensor sources.
-    Identify structural anomalies and deformations.
-    SLA: Process scan data within 5 minutes.""",
-    tools=[process_lidar_scan]
-)
-
-displacement_agent = Agent(
-    name="DisplacementCalculator",
-    instructions="""You are a structural displacement analysis expert.
-    Calculate precise displacement measurements from temporal scans.
-    Analyze settlement patterns and structural movements.
-    Predict future displacement trends using ML models.
-    SLA: Complete analysis within 2 minutes.""",
-    tools=[calculate_displacement]
-)
-
-heatmap_agent = Agent(
-    name="HeatmapGenerator",
-    instructions="""You are a safety visualization specialist.
-    Generate intuitive safety heatmaps for infrastructure monitoring.
-    Integrate multiple data sources for comprehensive risk assessment.
-    Highlight critical areas requiring immediate attention.
-    SLA: Generate heatmap within 1 minute.""",
-    tools=[generate_safety_heatmap]
-)
-
-maintenance_agent = Agent(
-    name="MaintenancePlanner",
-    instructions="""You are an infrastructure maintenance planning expert.
-    Schedule maintenance based on structural health and traffic patterns.
-    Optimize maintenance windows to minimize disruption.
-    Coordinate resources and estimate costs.
-    SLA: Create maintenance plan within 3 minutes.""",
-    tools=[plan_maintenance]
-)
-
-
-# Transportation infrastructure monitoring workflow
-def infrastructure_monitoring_workflow(
-    infrastructure_ids: List[str],
-    infrastructure_type: InfrastructureType,
-    traffic_data: Dict
-):
-    """
-    Complete infrastructure monitoring workflow from scanning to maintenance
-    Includes safety assessment and predictive maintenance
-    """
-    
-    monitoring_results = {
-        "timestamp": datetime.now().isoformat(),
-        "infrastructure_type": infrastructure_type.value,
-        "structures_monitored": len(infrastructure_ids),
-        "safety_assessments": [],
-        "maintenance_required": [],
-        "emergency_alerts": [],
-        "network_health_score": 0.0
-    }
-    
-    total_safety_score = 0.0
-    
-    for infra_id in infrastructure_ids:
-        try:
-            # Step 1: LiDAR scanning
-            lidar_scan = lidar_agent.start(
-                f"Process LiDAR scan for {infrastructure_type.value} {infra_id}"
-            )
-            
-            # Step 2: Displacement analysis
-            # Get baseline (would be from database in real scenario)
-            baseline = {"scan_id": "BSL-001", "point_cloud": []}
-            
-            displacement_analysis = displacement_agent.start(
-                f"Calculate displacement for scan: {lidar_scan} against baseline: {baseline}"
-            )
-            
-            # Step 3: Generate safety heatmap
-            sensor_data = []  # Would include IoT sensor readings
-            safety_heatmap = heatmap_agent.start(
-                f"Generate heatmap for displacement: {displacement_analysis} "
-                f"with sensors: {sensor_data}"
-            )
-            
-            monitoring_results["safety_assessments"].append({
-                "infrastructure_id": infra_id,
-                "safety_level": safety_heatmap.get("overall_safety"),
-                "max_displacement": displacement_analysis.get("max_displacement"),
-                "critical_zones": len(safety_heatmap.get("critical_areas", []))
-            })
-            
-            # Check for emergency conditions
-            if safety_heatmap.get("overall_safety") in ["danger", "critical"]:
-                monitoring_results["emergency_alerts"].append({
-                    "infrastructure_id": infra_id,
-                    "alert_level": "emergency",
-                    "safety_status": safety_heatmap["overall_safety"],
-                    "immediate_action": "close_to_traffic",
-                    "notification_sent": True
-                })
-                
-                # Step 4: Emergency maintenance planning
-                maintenance_plan = maintenance_agent.start(
-                    f"Plan emergency maintenance for {infra_id} "
-                    f"with safety: {safety_heatmap} and traffic: {traffic_data}"
-                )
-                monitoring_results["maintenance_required"].append(maintenance_plan)
-            
-            elif safety_heatmap.get("overall_safety") in ["warning", "caution"]:
-                # Regular maintenance planning
-                maintenance_plan = maintenance_agent.start(
-                    f"Plan maintenance for {infra_id} "
-                    f"with safety: {safety_heatmap} and traffic: {traffic_data}"
-                )
-                monitoring_results["maintenance_required"].append(maintenance_plan)
-            
-            # Calculate safety score contribution
-            safety_mapping = {
-                "safe": 1.0, "caution": 0.8, "warning": 0.6,
-                "danger": 0.3, "critical": 0.1
-            }
-            total_safety_score += safety_mapping.get(
-                safety_heatmap.get("overall_safety", "safe"), 0.5
-            )
-            
-        except Exception as e:
-            # Fallback: Manual inspection
-            monitoring_results["emergency_alerts"].append({
-                "infrastructure_id": infra_id,
-                "alert_level": "system_failure",
-                "error": str(e),
-                "action": "dispatch_manual_inspection_team"
-            })
-    
-    # Calculate network health score
-    if infrastructure_ids:
-        monitoring_results["network_health_score"] = (
-            total_safety_score / len(infrastructure_ids) * 100
-        )
-    
-    return monitoring_results
-
-
-# Multi-modal transportation patterns
-class MultiModalTransportPatterns:
-    """
-    Patterns for different transportation infrastructure types
-    Demonstrates cross-industry reuse for transportation sector
-    """
-    
-    @staticmethod
-    def adapt_for_bridge_monitoring() -> Agent:
-        """Adapt tunnel monitoring for bridge structures"""
-        return Agent(
-            name="BridgeMonitor",
-            instructions="""Monitor bridge structural health using sensors.
-            Track deck deflection, cable tension, and vibration patterns.
-            Detect fatigue cracks and corrosion.
-            Consider wind and seismic loads.
-            SLA: Real-time monitoring with 10-second updates."""
-        )
-    
-    @staticmethod
-    def adapt_for_railway_tracks() -> Agent:
-        """Adapt for railway track monitoring"""
-        return Agent(
-            name="RailwayMonitor",
-            instructions="""Monitor railway track geometry and condition.
-            Detect rail defects, gauge variations, and ballast degradation.
-            Analyze wheel-rail interaction forces.
-            Predict track maintenance needs.
-            SLA: Process track inspection data within 3 minutes per km."""
-        )
-    
-    @staticmethod
-    def adapt_for_airport_runways() -> Agent:
-        """Adapt for airport runway monitoring"""
-        return Agent(
-            name="RunwayMonitor",
-            instructions="""Monitor airport runway surface conditions.
-            Detect FOD (Foreign Object Debris) and surface irregularities.
-            Assess pavement condition and friction levels.
-            Monitor drainage and lighting systems.
-            SLA: Complete runway scan within 10 minutes."""
-        )
-
-
-# Intelligent traffic management integration
-class TrafficManagementPatterns:
-    """
-    Patterns for integrating with intelligent traffic systems
-    """
-    
-    @staticmethod
-    def optimize_traffic_flow(maintenance_window: Dict, traffic_patterns: Dict) -> Dict:
-        """Optimize traffic flow during maintenance"""
-        peak_hours = traffic_patterns.get("peak_hours", [7, 9, 17, 19])
-        maintenance_start = maintenance_window.get("start_time", "02:00")
-        
-        # Calculate optimal diversion routes
-        return {
-            "strategy": "dynamic_routing",
-            "diversion_routes": [
-                {"route_id": "ALT-1", "capacity": 0.6, "added_time": 15},
-                {"route_id": "ALT-2", "capacity": 0.3, "added_time": 20}
-            ],
-            "lane_management": {
-                "reversible_lanes": True,
-                "peak_direction": "inbound" if int(maintenance_start[:2]) < 12 else "outbound"
-            },
-            "public_transport": {
-                "increase_frequency": True,
-                "additional_services": 4
-            },
-            "estimated_delay": 10  # minutes
-        }
-    
-    @staticmethod
-    def incident_response_plan(incident_type: str, location: Dict) -> Dict:
-        """Generate incident response plan"""
-        response_times = {
-            "accident": 5,
-            "breakdown": 10,
-            "infrastructure_failure": 2,
-            "weather_hazard": 15
-        }
-        
-        return {
-            "incident_id": f"INC-{datetime.now().strftime('%Y%m%d%H%M')}",
-            "response_time": response_times.get(incident_type, 10),
-            "resources_dispatched": [
-                r for r in [
-                    "emergency_response_team",
-                    "traffic_control_unit",
-                    "maintenance_crew" if "infrastructure" in incident_type else None
-                ] if r is not None
-            ],
-            "traffic_management": {
-                "lane_closures": 2,
-                "speed_reduction": 30,  # km/h
-                "message_signs": ["Incident ahead", "Reduce speed", "Merge left"]
-            },
-            "estimated_clearance": 45  # minutes
-        }
-
-
-# Predictive maintenance patterns
-class PredictiveMaintenancePatterns:
-    """
-    ML-based predictive maintenance for transportation infrastructure
-    """
-    
-    @staticmethod
-    def predict_failure_probability(
-        displacement_trend: List[float],
-        environmental_factors: Dict
-    ) -> float:
-        """Predict probability of structural failure"""
-        # Simplified prediction model
-        trend_slope = (displacement_trend[-1] - displacement_trend[0]) / len(displacement_trend)
-        temperature_factor = abs(environmental_factors.get("temp_variation", 0)) / 50
-        moisture_factor = environmental_factors.get("humidity", 50) / 100
-        
-        base_probability = 0.1
-        trend_contribution = max(0, trend_slope * 10)
-        env_contribution = (temperature_factor + moisture_factor) * 0.2
-        
-        return min(1.0, base_probability + trend_contribution + env_contribution)
-    
-    @staticmethod
-    def optimize_maintenance_schedule(
-        infrastructure_conditions: List[Dict],
-        budget_constraint: float
-    ) -> List[Dict]:
-        """Optimize maintenance schedule across infrastructure network"""
-        # Sort by urgency (safety_factor * cost_efficiency)
-        prioritized = sorted(
-            infrastructure_conditions,
-            key=lambda x: (1 - x.get("safety_factor", 1)) * (1 / x.get("cost", 1)),
-            reverse=True
-        )
-        
-        scheduled = []
-        total_cost = 0
-        
-        for infra in prioritized:
-            if total_cost + infra.get("cost", 0) <= budget_constraint:
-                scheduled.append({
-                    "infrastructure_id": infra["id"],
-                    "scheduled_date": infra["recommended_date"],
-                    "priority": "high" if infra["safety_factor"] < 0.7 else "medium"
-                })
-                total_cost += infra.get("cost", 0)
-        
-        return scheduled
-
-
-# Example usage
-if __name__ == "__main__":
-    # Monitor tunnel network
-    tunnel_ids = [f"TUN-{i:03d}" for i in range(1, 6)]
-    traffic = {
-        "peak_volume": 3000,
-        "off_peak_volume": 800,
-        "peak_hours": [7, 9, 17, 19]
-    }
-    
-    result = infrastructure_monitoring_workflow(
-        tunnel_ids,
-        InfrastructureType.TUNNEL,
-        traffic
-    )
-    print("Infrastructure monitoring result:", result)
-    
-    # Multi-modal adaptation examples
-    bridge_monitor = MultiModalTransportPatterns.adapt_for_bridge_monitoring()
-    railway_monitor = MultiModalTransportPatterns.adapt_for_railway_tracks()
-    runway_monitor = MultiModalTransportPatterns.adapt_for_airport_runways()
-    
-    # Traffic management example
-    maintenance = {"start_time": "22:00", "duration": 8}
-    traffic_plan = TrafficManagementPatterns.optimize_traffic_flow(maintenance, traffic)
-    print("Traffic optimization plan:", traffic_plan)
-    
-    # Predictive maintenance example
-    displacement_history = [2.0, 2.3, 2.8, 3.5, 4.3]  # mm over months
-    environment = {"temp_variation": 30, "humidity": 75}
-    
-    failure_prob = PredictiveMaintenancePatterns.predict_failure_probability(
-        displacement_history, environment
-    )
-    print(f"Failure probability: {failure_prob:.2%}")
@@ -214,13 +214,8 @@ def example_subagent_delegation():
    assert hasattr(agent, "delegate"), "Agent should have delegate method"
    assert callable(agent.delegate), "delegate should be callable"
    
-    # Create a subagent using the public profile API
-    from praisonaiagents.agents.profiles import BUILTIN_PROFILES, get_profile
-    explorer_profile = get_profile("explorer") if "explorer" in BUILTIN_PROFILES else None
-    subagent = Agent(
-        name="subagent_explorer",
-        instructions=explorer_profile.system_prompt if explorer_profile else "You are an explorer assistant.",
-    )
+    # Check _create_subagent method
+    subagent = agent._create_subagent("explorer")
    print(f"Created subagent: {subagent.name}")
    print(f"Subagent instructions: {subagent.instructions[:50]}...")
    
@@ -6,18 +6,10 @@ It includes self-reflection for improving code quality and multiple specialized
 """

 from praisonaiagents import Agent, Task, AgentTeam
+from praisonaiagents.tools import CodeInterpreter

-def run_python_code(code: str) -> str:
-    """Execute Python code safely and return stdout or errors."""
-    import io
-    import contextlib
-    try:
-        buf = io.StringIO()
-        with contextlib.redirect_stdout(buf):
-            exec(code, {"__builtins__": __builtins__})
-        return buf.getvalue() or "Executed successfully (no output)"
-    except Exception as e:
-        return f"Execution error: {e}"
+# Initialize code execution tool
+code_interpreter = CodeInterpreter()

 # Custom code analysis tool
 def analyze_code_complexity(code: str) -> str:
@@ -40,7 +32,9 @@ code_writer = Agent(
    goal="Write clean, efficient, and well-documented Python code",
    backstory="You are a senior Python developer with 10+ years of experience in writing production-quality code.",
    instructions="Write Python code that follows PEP 8 standards, includes proper error handling, and has clear documentation.",
-    output="silent",
+    reflection=True,
+    
+    
 )

 code_reviewer = Agent(
@@ -58,7 +52,7 @@ code_executor = Agent(
    goal="Safely execute and test Python code",
    backstory="You are responsible for running code and reporting results or errors.",
    instructions="Execute the provided code safely and report the output or any errors encountered.",
-    tools=[run_python_code]
+    tools=[code_interpreter]
 )

 # Example tasks
@@ -93,7 +87,7 @@ def demonstrate_code_agent():
    workflow = AgentTeam(
        agents=[code_writer, code_reviewer, code_executor],
        tasks=[write_task, review_task, execute_task],
-        process="sequential", output="silent"
+        process="sequential", output="verbose"
    )
    
    # Run the workflow
@@ -112,8 +106,10 @@ general_code_agent = Agent(
 3. Explain code concepts
 4. Refactor code for better performance
 5. Add tests and documentation""",
-    tools=[run_python_code, analyze_code_complexity],
-    output="silent",
+    tools=[code_interpreter, analyze_code_complexity],
+    reflection=True,
+    
+    
 )

 if __name__ == "__main__":
@@ -41,25 +41,21 @@ def calculate_statistics(numbers: list) -> str:
    """Calculate basic statistics for a list of numbers."""
    if not numbers:
        return "No numbers provided"
-    try:
-        nums = [float(x) for x in numbers]
-    except (TypeError, ValueError):
-        return "Error: all values must be numbers"
-
-    n = len(nums)
-    mean = sum(nums) / n
-    sorted_nums = sorted(nums)
+    
+    n = len(numbers)
+    mean = sum(numbers) / n
+    sorted_nums = sorted(numbers)
    median = sorted_nums[n//2] if n % 2 else (sorted_nums[n//2-1] + sorted_nums[n//2]) / 2
-    variance = sum((x - mean)**2 for x in nums) / n
+    variance = sum((x - mean)**2 for x in numbers) / n
    std_dev = math.sqrt(variance)
-
+    
    return f"""Statistics:
 - Count: {n}
 - Mean: {mean:.4f}
 - Median: {median:.4f}
 - Standard Deviation: {std_dev:.4f}
- Min: {min(nums)}
- Max: {max(nums)}"""
+- Min: {min(numbers)}
+- Max: {max(numbers)}"""

 # Create specialized math agents
 calculation_agent = Agent(
@@ -70,7 +66,8 @@ calculation_agent = Agent(
    instructions="Perform the requested calculations step by step, showing your work clearly.",
    tools=[basic_calculator, solve_quadratic, calculate_statistics],
    reflection=True,
-    output="silent",
+    
+    
 )

 problem_solver_agent = Agent(
@@ -85,7 +82,8 @@ problem_solver_agent = Agent(
 4. Show each step of the solution
 5. Verify the answer makes sense""",
    reflection=True,
-    output="silent",
+    
+    
 )

 math_tutor_agent = Agent(
@@ -93,8 +91,7 @@ math_tutor_agent = Agent(
    role="Mathematics tutor and explainer",
    goal="Explain mathematical concepts clearly and help users understand math",
    backstory="You are a patient math tutor who can explain concepts at any level from elementary to advanced.",
-    instructions="Explain mathematical concepts using simple language, provide examples, and check for understanding.",
-    output="silent",
+    instructions="Explain mathematical concepts using simple language, provide examples, and check for understanding."
 )

 # Example workflow for solving complex problems
@@ -129,8 +126,7 @@ def solve_math_problem_workflow():
    workflow = AgentTeam(
        agents=[problem_solver_agent, calculation_agent, math_tutor_agent],
        tasks=[understand_task, solve_task, verify_task],
-        process="sequential",
-        output="silent",
+        process="sequential", output="verbose"
    )
    
    return workflow.start()
@@ -152,7 +148,8 @@ math_assistant = Agent(
 Always show your work step by step.""",
    tools=[basic_calculator, solve_quadratic, calculate_statistics],
    reflection=True,
-    output="silent",
+    
+    
 )

 if __name__ == "__main__":
@@ -67,11 +67,7 @@ def main():
    
    # 4. Subagent delegation
    print("\n4. Subagent Delegation:")
-    explorer_profile = get_profile("explorer") if "explorer" in BUILTIN_PROFILES else None
-    subagent = Agent(
-        name="subagent_explorer",
-        instructions=explorer_profile.system_prompt if explorer_profile else "You are an explorer assistant.",
-    )
+    subagent = agent._create_subagent("explorer")
    print(f"   Created subagent: {subagent.name}")
    print(f"   Profile: explorer (read-only)")
    
@@ -1,238 +0,0 @@
-#!/usr/bin/env python3
-"""
-Example: Bot In-Flight Run Control
-
-Demonstrates how to use SessionRunControl to provide better UX for bots
-during long-running agent operations. Shows busy feedback, pending message
-handling, and /stop command support.
-
-This example shows the "before" and "after" behavior described in issue #1914.
-"""
-
-import asyncio
-import time
-from typing import Optional
-
-# Mock agent for demonstration
-class MockAgent:
-    """Simple mock agent that simulates long-running tasks."""
-    
-    def __init__(self, name="assistant"):
-        self.name = name
-        self.chat_history = []
-        self._interrupt_controller = None
-    
-    def chat(self, prompt: str) -> str:
-        """Simulate a long-running chat response."""
-        print(f"[Agent] Starting to process: '{prompt[:50]}...'")
-        
-        # Simulate long processing with interrupt checking
-        for i in range(30):  # 3 second task (30 x 0.1s)
-            time.sleep(0.1)
-            
-            # Check for interruption
-            if (self._interrupt_controller and 
-                hasattr(self._interrupt_controller, 'is_set') and
-                self._interrupt_controller.is_set()):
-                reason = getattr(self._interrupt_controller, 'reason', 'unknown')
-                print(f"[Agent] ⚠️ Interrupted: {reason}")
-                raise InterruptedError(f"Task cancelled: {reason}")
-                
-        response = f"Completed analysis of: {prompt}"
-        print(f"[Agent] ✅ Finished: {response}")
-        return response
-
-
-async def demo_without_run_control():
-    """Demonstrate the old behavior (silent blocking)."""
-    print("\n" + "="*60)
-    print("🔴 BEFORE: Without Run Control (Silent Blocking)")
-    print("="*60)
-    
-    from praisonai.bots._session import BotSessionManager
-    
-    # Standard session manager (no run control)
-    session_mgr = BotSessionManager()
-    agent = MockAgent("research-agent")
-    user_id = "user123"
-    
-    print("\n1. User sends long task:")
-    print("   User: 'research quantum computing trends'")
-    
-    # Start first task (this will run for 3 seconds)
-    task1 = asyncio.create_task(
-        session_mgr.chat(agent, user_id, "research quantum computing trends")
-    )
-    
-    # Simulate user sending follow-up messages during execution
-    await asyncio.sleep(0.5)
-    print("\n2. User sends follow-up (this will block silently):")
-    print("   User: 'actually focus on business impact'")
-    
-    task2 = asyncio.create_task(
-        session_mgr.chat(agent, user_id, "actually focus on business impact")  
-    )
-    
-    await asyncio.sleep(0.5)
-    print("\n3. User tries to send /stop (this will also block):")
-    print("   User: '/stop'")
-    
-    task3 = asyncio.create_task(
-        session_mgr.chat(agent, user_id, "/stop")
-    )
-    
-    print("\n⏳ Waiting for all messages to complete...")
-    results = await asyncio.gather(task1, task2, task3)
-    
-    print("\n📊 Results (all processed sequentially with no feedback):")
-    for i, result in enumerate(results, 1):
-        print(f"   {i}. {result}")
-    
-    print("\n❌ Problems with this approach:")
-    print("   - No immediate feedback on follow-up messages")
-    print("   - No way to cancel long-running tasks") 
-    print("   - User doesn't know if bot is working or broken")
-
-
-async def demo_with_run_control():
-    """Demonstrate the new behavior with run control."""
-    print("\n" + "="*60)
-    print("🟢 AFTER: With Run Control (Better UX)")  
-    print("="*60)
-    
-    from praisonai.bots._run_control import SessionRunControl
-    from praisonai.bots._session import BotSessionManager
-    from praisonai.bots._commands import handle_stop_command
-    
-    # Session manager with run control
-    run_control = SessionRunControl(
-        busy_mode="queue",
-        busy_ack_template="⏳ {action} — will process after current task finishes"
-    )
-    session_mgr = BotSessionManager(run_control=run_control)
-    agent = MockAgent("research-agent")
-    user_id = "user123"
-    
-    print("\n1. User sends long task:")
-    print("   User: 'research quantum computing trends'")
-    
-    # Start first task
-    task1 = asyncio.create_task(
-        session_mgr.chat_with_run_control(agent, user_id, "research quantum computing trends")
-    )
-    
-    # Immediate follow-up messages get acknowledgments
-    await asyncio.sleep(0.5)
-    print("\n2. User sends follow-up (gets immediate feedback):")
-    print("   User: 'actually focus on business impact'")
-    
-    result2 = await session_mgr.chat_with_run_control(
-        agent, user_id, "actually focus on business impact"
-    )
-    print(f"   Bot: {result2['response']}")
-    print(f"   Metadata: {result2['metadata']}")
-    
-    await asyncio.sleep(0.5)
-    print("\n3. User decides to stop current task:")
-    print("   User: '/stop'")
-    
-    stop_response = await handle_stop_command(user_id, run_control)
-    print(f"   Bot: {stop_response}")
-    
-    # Wait for first task to complete (it should be cancelled)
-    try:
-        result1 = await task1
-        print(f"\n📊 Task 1 result: {result1['response']}")
-        print(f"   Metadata: {result1['metadata']}")
-    except asyncio.CancelledError:
-        print("\n📊 Task 1 was cancelled as expected")
-    
-    print("\n4. User sends fresh request after stopping:")
-    print("   User: 'what's the weather like?'")
-    
-    result4 = await session_mgr.chat_with_run_control(
-        agent, user_id, "what's the weather like?"
-    )
-    print(f"   Bot: {result4['response']}")
-    
-    print("\n✅ Benefits of this approach:")
-    print("   - Immediate feedback on all messages")
-    print("   - Can cancel long-running tasks with /stop")
-    print("   - Pending messages are queued and processed in order")
-    print("   - User always knows what's happening")
-
-
-async def demo_interrupt_mode():
-    """Demonstrate interrupt mode."""
-    print("\n" + "="*60)
-    print("⚡ BONUS: Interrupt Mode (Cancel and Restart)")
-    print("="*60)
-    
-    from praisonai.bots._run_control import SessionRunControl
-    from praisonai.bots._session import BotSessionManager
-    
-    # Use interrupt mode instead of queue mode
-    run_control = SessionRunControl(busy_mode="interrupt")
-    session_mgr = BotSessionManager(run_control=run_control)
-    agent = MockAgent("research-agent")
-    user_id = "user123"
-    
-    print("\n1. User starts long task:")
-    print("   User: 'research renewable energy for 30 pages'")
-    
-    task1 = asyncio.create_task(
-        session_mgr.chat_with_run_control(agent, user_id, "research renewable energy for 30 pages")
-    )
-    
-    await asyncio.sleep(0.5)
-    print("\n2. User changes mind (interrupts and restarts):")
-    print("   User: 'actually just summarize solar panel efficiency'")
-    
-    result2 = await session_mgr.chat_with_run_control(
-        agent, user_id, "actually just summarize solar panel efficiency"
-    )
-    print(f"   Bot: {result2['response']}")
-    
-    # Wait for original task (should be interrupted)
-    try:
-        result1 = await task1
-        print(f"\n📊 Original task result: {result1['response']}")
-        if result1['metadata'].get('interrupted'):
-            print("   ✅ Original task was properly interrupted")
-    except:
-        print("\n📊 Original task was cancelled")
-    
-    print("\n✅ Interrupt mode is great for:")
-    print("   - When users frequently change direction")
-    print("   - Interactive sessions where latest intent matters most")
-    print("   - Real-time collaboration scenarios")
-
-
-async def main():
-    """Run all demonstrations."""
-    print("🚀 Bot In-Flight Run Control Demo")
-    print("Solving the silent lock problem described in issue #1914")
-    
-    # Show the problem
-    await demo_without_run_control()
-    
-    # Show the solution  
-    await demo_with_run_control()
-    
-    # Show advanced feature
-    await demo_interrupt_mode()
-    
-    print("\n" + "="*60)
-    print("🎯 Summary")
-    print("="*60)
-    print("SessionRunControl solves the silent blocking problem by providing:")
-    print("1. Immediate feedback for mid-run messages")
-    print("2. /stop command support via InterruptController")
-    print("3. Configurable policies (queue/interrupt/steer)")
-    print("4. Run generation tracking to prevent race conditions")
-    print("5. Pending message merging and ordering")
-    print("\nThis makes bot interactions feel responsive and predictable!")
-
-
-if __name__ == "__main__":
-    asyncio.run(main())
@@ -1,201 +0,0 @@
-"""
-Prototype implementation of Gateway Server with backpressure and flow control.
-
-This demonstrates how the bounded queues and slow consumer detection would be
-implemented in the wrapper layer (praisonai package) using the configuration
-from the core SDK (praisonaiagents).
-"""
-
-from __future__ import annotations
-
-import asyncio
-import logging
-from dataclasses import dataclass, field
-from typing import Any, Dict, Optional
-
-# For the prototype, we'll define minimal config inline
-# from praisonaiagents.gateway import GatewayConfig
-
-@dataclass 
-class SessionConfig:
-    max_inbox: int = 256
-    
-@dataclass
-class GatewayConfig:
-    session_config: SessionConfig = field(default_factory=SessionConfig)
-    max_buffered_bytes: int = 1024 * 1024  # 1MB default
-
-logger = logging.getLogger(__name__)
-
-
-@dataclass
-class GatewaySession:
-    """Gateway session with bounded inbox for backpressure."""
-    
-    _session_id: str
-    _agent_id: str
-    _client_id: Optional[str] = None
-    _inbox: asyncio.Queue = field(default_factory=asyncio.Queue)
-    _max_inbox: int = 256
-    _is_executing: bool = False
-    
-    def setup_inbox(self, max_inbox: int = 256) -> None:
-        """Set up the inbox with proper bounds."""
-        self._max_inbox = max_inbox
-        if max_inbox > 0:
-            self._inbox = asyncio.Queue(maxsize=max_inbox)
-        else:
-            self._inbox = asyncio.Queue()  # Unbounded for backward compat if 0
-    
-    async def queue_message(self, message: str) -> bool:
-        """Queue a user message with backpressure handling.
-        
-        Returns:
-            True if message was queued, False if inbox is full.
-        """
-        if self._max_inbox > 0:
-            # Bounded queue - check if full
-            try:
-                self._inbox.put_nowait(message)
-                return True
-            except asyncio.QueueFull:
-                return False
-        else:
-            # Unbounded queue for backward compatibility
-            await self._inbox.put(message)
-            return True
-
-
-class GatewayServer:
-    """Gateway server with flow control."""
-    
-    def __init__(self, config: GatewayConfig):
-        self.config = config
-        self._sessions: Dict[str, GatewaySession] = {}
-        self._clients: Dict[str, Any] = {}  # client_id -> WebSocket
-        
-    def create_session(self, session_id: str, agent_id: str, client_id: str) -> GatewaySession:
-        """Create a new session with bounded inbox."""
-        session = GatewaySession(
-            _session_id=session_id,
-            _agent_id=agent_id,
-            _client_id=client_id,
-        )
-        # Set up inbox with configured bounds
-        session.setup_inbox(self.config.session_config.max_inbox)
-        self._sessions[session_id] = session
-        return session
-        
-    async def handle_message(self, session: GatewaySession, message: str, client_id: str) -> str:
-        """Handle incoming message with inbox overflow protection."""
-        # Try to queue the message
-        queued = await session.queue_message(message)
-        if not queued:
-            # Inbox is full, send error response
-            await self._send_to_client(
-                client_id,
-                {
-                    "type": "error",
-                    "code": "inbox_full",
-                    "message": "Message queue is full. Please wait for current messages to be processed."
-                }
-            )
-            return "Inbox full - message rejected."
-        
-        # Start processing if not already running
-        if not session._is_executing:
-            session._is_executing = True
-            asyncio.create_task(self._drain_session_queue(session, client_id))
-            return "Processing started."
-        
-        return "Message queued."
-    
-    async def _drain_session_queue(self, session: GatewaySession, client_id: str) -> None:
-        """Process all messages from the session's inbox queue."""
-        try:
-            while not session._inbox.empty():
-                msg = await session._inbox.get()
-                # Process message here (simplified for prototype)
-                logger.info(f"Processing message for session {session._session_id}: {msg}")
-                
-                # Simulate processing delay
-                await asyncio.sleep(0.1)
-                
-                # Send response to client (simplified)
-                await self._send_to_client(
-                    client_id,
-                    {
-                        "type": "response",
-                        "message": f"Processed: {msg}"
-                    }
-                )
-        except Exception as e:
-            logger.error(f"Error processing queue for session {session._session_id}: {e}")
-        finally:
-            session._is_executing = False
-    
-    async def _send_to_client(self, client_id: str, data: Dict[str, Any]) -> None:
-        """Send data to client with slow consumer detection."""
-        ws = self._clients.get(client_id)
-        if ws:
-            try:
-                # Check for slow consumer (buffered bytes)
-                if self.config.max_buffered_bytes > 0:
-                    # Try to get transport buffer size
-                    transport = getattr(ws, 'transport', None) or getattr(ws, '_transport', None)
-                    if transport and hasattr(transport, 'get_write_buffer_size'):
-                        buffered_bytes = transport.get_write_buffer_size()
-                        if buffered_bytes > self.config.max_buffered_bytes:
-                            # Check if this is a droppable event type
-                            event_type = data.get("type", "")
-                            droppable_types = {"presence", "typing", "status"}
-                            
-                            if event_type in droppable_types:
-                                # Drop the message for slow consumers
-                                logger.warning(f"Dropping {event_type} event for slow consumer {client_id} (buffered: {buffered_bytes} bytes)")
-                                return
-                            else:
-                                # Critical event - close the connection
-                                logger.warning(f"Closing slow consumer {client_id} (buffered: {buffered_bytes} bytes > {self.config.max_buffered_bytes})")
-                                await ws.close(code=1013, reason="slow consumer")
-                                return
-                
-                # Send the message
-                await ws.send_json(data)
-            except Exception as e:
-                logger.error(f"Error sending to client {client_id}: {e}", exc_info=True)
-
-
-# Example usage demonstrating the flow control
-async def demo():
-    """Demonstrate the flow control implementation."""
-    
-    # Create config with bounded inbox and slow consumer detection
-    config = GatewayConfig()
-    config.session_config.max_inbox = 10  # Small inbox for demo
-    config.max_buffered_bytes = 1024  # 1KB for demo
-    
-    # Create server and session
-    server = GatewayServer(config)
-    session = server.create_session("sess1", "agent1", "client1")
-    
-    # Simulate filling up the inbox
-    print(f"Inbox max size: {session._inbox.maxsize}")
-    
-    # Fill the inbox
-    for i in range(10):
-        success = await session.queue_message(f"Message {i}")
-        print(f"Queued message {i}: {success}")
-    
-    # Try to add one more - should fail
-    success = await session.queue_message("Overflow message")
-    print(f"Overflow message queued: {success}")
-    
-    # Process messages
-    while not session._inbox.empty():
-        msg = await session._inbox.get()
-        print(f"Processing: {msg}")
-
-
-if __name__ == "__main__":
-    asyncio.run(demo())
@@ -4,70 +4,43 @@ Basic handoff example demonstrating agent-to-agent delegation.
 This example shows how agents can hand off tasks to specialized agents.
 """

-from pydantic import BaseModel
-from typing import TypedDict, Optional
 from praisonaiagents import Agent

-# Define typed payload contracts for each handoff
-class BillingPayload(BaseModel):
-    account_id: str
-    invoice_amount: Optional[float] = None
-    issue_type: str = "billing"
-
-class RefundPayload(TypedDict):
-    transaction_id: str
-    amount: float
-    reason: str
-
-class TechnicalPayload(BaseModel):
-    error_code: Optional[str] = None
-    device_info: str
-    user_agent: str
-
-# Create specialized agents that expect typed payloads from the triage
-# (In a real scenario, the agents would validate and use the payload.)
+# Create specialized agents
 billing_agent = Agent(
    name="Billing Agent",
    role="Billing Specialist",
-    goal="Handle all billing-related inquiries and tasks using structured data",
-    backstory="I am an expert in billing systems, payment processing, and invoice management.",
-    input_payload_schema=BillingPayload  # hypothetical but consistent with design
+    goal="Handle all billing-related inquiries and tasks",
+    backstory="I am an expert in billing systems, payment processing, and invoice management."
 )

 refund_agent = Agent(
    name="Refund Agent",
    role="Refund Specialist",
-    goal="Process refund requests and handle refund-related issues with validated data",
-    backstory="I specialize in processing refunds, evaluating refund eligibility, and ensuring customer satisfaction.",
-    input_payload_schema=RefundPayload
+    goal="Process refund requests and handle refund-related issues",
+    backstory="I specialize in processing refunds, evaluating refund eligibility, and ensuring customer satisfaction."
 )

 technical_support_agent = Agent(
    name="Technical Support",
    role="Technical Support Specialist",
-    goal="Resolve technical issues and provide technical assistance with structured context",
-    backstory="I am skilled in troubleshooting technical problems and providing solutions.",
-    input_payload_schema=TechnicalPayload
+    goal="Resolve technical issues and provide technical assistance",
+    backstory="I am skilled in troubleshooting technical problems and providing solutions."
 )

-# Create a triage agent with typed handoffs
+# Create a triage agent with handoffs to specialized agents
 triage_agent = Agent(
    name="Triage Agent",
    role="Customer Service Triage",
-    goal="Understand customer needs and route them to the appropriate specialist with a validated payload",
+    goal="Understand customer needs and route them to the appropriate specialist",
    backstory="I analyze customer requests and direct them to the most suitable specialist for efficient resolution.",
    instructions="""Analyze the customer's request and determine which specialist can best help:
    - For billing questions, payment issues, or invoices, transfer to the Billing Agent
    - For refund requests or refund status inquiries, transfer to the Refund Agent  
    - For technical problems or product issues, transfer to Technical Support
    
-    Always explain why you're transferring the customer before doing so.
-    When handing off, populate the appropriate payload with extracted data.""",
-    handoffs=[
-        (billing_agent, BillingPayload),
-        (refund_agent, RefundPayload),
-        (technical_support_agent, TechnicalPayload)
-    ]
+    Always explain why you're transferring the customer before doing so.""",
+    handoffs=[billing_agent, refund_agent, technical_support_agent]
 )

 # Example usage
@@ -1,124 +0,0 @@
-"""
-Example: Using MuAPI with PraisonAI for image generation.
-
-MuAPI (https://muapi.ai) provides unified access to 400+ generative media
-models — Flux, Midjourney, GPT-4o Image, Google Imagen, Seedream, Veo3,
-Kling, Wan, and more — through a single API.
-
-This example creates a PraisonAI agent that generates images on demand
-using the MuAPI image generation API as a custom tool.
-
-Setup:
-    pip install praisonaiagents requests
-    export MUAPI_API_KEY=your_key_here
-    # Get an API key at https://muapi.ai/dashboard/api-keys
-"""
-
-import os
-import time
-
-import requests
-from praisonaiagents import Agent, Tool
-
-
-MUAPI_BASE_URL = "https://api.muapi.ai/api/v1"
-
-ALLOWED_MODELS = {
-    "flux-schnell", "flux-dev", "flux-kontext-dev", "flux-kontext-pro",
-    "midjourney", "gpt4o", "imagen4", "imagen4-fast", "seedream",
-    "hidream-fast", "hidream-dev", "reve", "ideogram", "hunyuan",
-}
-
-
-def generate_image(prompt: str, model: str = "flux-schnell") -> str:
-    """Generate an image using MuAPI and return the output URL.
-
-    Args:
-        prompt: Text description of the image to generate.
-        model: MuAPI model ID. Options include flux-schnell, flux-dev,
-               midjourney, gpt4o, imagen4, seedream, hidream-fast, reve.
-
-    Returns:
-        URL of the generated image.
-    """
-    if model not in ALLOWED_MODELS:
-        raise ValueError(
-            f"Unsupported model '{model}'. Choose from: {sorted(ALLOWED_MODELS)}"
-        )
-
-    api_key = os.environ.get("MUAPI_API_KEY", "")
-    if not api_key:
-        raise ValueError("MUAPI_API_KEY environment variable not set")
-
-    headers = {"x-api-key": api_key, "Content-Type": "application/json"}
-
-    # Submit the generation request
-    submit_resp = requests.post(
-        f"{MUAPI_BASE_URL}/{model}",
-        headers=headers,
-        json={"prompt": prompt},
-        timeout=30,
-    )
-    submit_resp.raise_for_status()
-    submit_data = submit_resp.json()
-    request_id = submit_data.get("request_id")
-    if not request_id:
-        raise RuntimeError(f"MuAPI did not return a request_id: {submit_data}")
-
-    # Poll until completion
-    for _ in range(120):
-        time.sleep(3)
-        poll_resp = requests.get(
-            f"{MUAPI_BASE_URL}/predictions/{request_id}/result",
-            headers={"x-api-key": api_key},
-            timeout=15,
-        )
-        poll_resp.raise_for_status()
-        data = poll_resp.json()
-        status = data.get("status")
-        if status == "completed":
-            outputs = data.get("outputs") or []
-            if not outputs:
-                raise RuntimeError(f"Generation completed but returned no outputs: {data}")
-            return outputs[0]
-        if status in ("failed", "cancelled"):
-            raise RuntimeError(f"Image generation {status}: {data.get('error', '')}")
-
-    raise TimeoutError("Image generation timed out after 6 minutes")
-
-
-if __name__ == "__main__":
-    image_tool = Tool(
-        name="generate_image",
-        description=(
-            "Generate an image from a text description using MuAPI's 400+ model library. "
-            "Supported models: flux-schnell (fast), flux-dev, midjourney, gpt4o, imagen4, "
-            "seedream, hidream-fast, reve. Returns a URL to the generated image."
-        ),
-        function=generate_image,
-    )
-
-    agent = Agent(
-        name="Image Generator",
-        instructions=(
-            "You are a creative image generation assistant. "
-            "When asked to create an image, use the generate_image tool. "
-            "Pick the best model for the task: flux-schnell for speed, "
-            "midjourney for artistic quality, gpt4o for photorealism."
-        ),
-        tools=[image_tool],
-    )
-
-    # Example: generate a product photo
-    response = agent.start(
-        "Generate a photorealistic product photo of a sleek black coffee mug "
-        "on a white marble table with soft morning light. Use the best model for photorealism."
-    )
-    print(response)
-
-    # Example: generate concept art
-    response = agent.start(
-        "Create a cyberpunk cityscape at night with neon reflections on wet streets. "
-        "Use midjourney for artistic quality."
-    )
-    print(response)
@@ -179,7 +179,7 @@ def main():
    
    context = memory.build_context_for_task(
        task_descr="Explain AI alignment challenges",
-        memory={"user_id": "quality_demo_user"},
+        user_id="quality_demo_user",
        max_items=3
    )
    
@@ -1,115 +0,0 @@
-#!/usr/bin/env python3
-"""
-JIRA Agent Watch Example
-
-This example demonstrates how to use the JIRA watch tools with PraisonAI agents.
-The agent can monitor JIRA issues and projects for changes, search for issues,
-and get detailed issue information.
-
-Setup:
-1. Install JIRA library: pip install jira
-2. Set environment variables:
-   - [email protected] (for cloud JIRA)
-   - JIRA_API_TOKEN=your_api_token
-   - Or use JIRA_USERNAME + JIRA_API_TOKEN for server JIRA
-
-Usage:
-    python jira_agent_example.py
-"""
-
-import os
-from praisonaiagents import Agent
-from praisonaiagents.tools import jira_tools
-
-
-def main():
-    """Demonstrate JIRA agent watch capabilities."""
-    
-    # Check if JIRA credentials are available
-    jira_email = os.getenv('JIRA_EMAIL')
-    jira_token = os.getenv('JIRA_API_TOKEN')
-    jira_username = os.getenv('JIRA_USERNAME')
-    
-    if not jira_token:
-        print("❌ JIRA_API_TOKEN environment variable not set")
-        print("Please set your JIRA API token:")
-        print("export JIRA_API_TOKEN='your_token_here'")
-        return
-        
-    if not (jira_email or jira_username):
-        print("❌ JIRA credentials incomplete")
-        print("Please set either:")
-        print("- JIRA_EMAIL (for cloud JIRA)")
-        print("- JIRA_USERNAME (for server JIRA)")
-        return
-
-    # Create agent with JIRA tools
-    agent = Agent(
-        name="JIRA Monitor Agent",
-        instructions="""
-        You are a JIRA monitoring agent. You can:
-        
-        1. Watch specific JIRA issues for changes
-        2. Monitor JIRA projects for new issues and updates  
-        3. Search for issues using JQL queries
-        4. Get detailed information about specific issues
-        
-        When using JIRA tools, always provide the full JIRA URL 
-        (e.g., https://yourcompany.atlassian.net).
-        
-        Be helpful and provide clear summaries of JIRA activity.
-        """,
-        tools=jira_tools(),  # Add all JIRA tools
-        llm="gpt-4o-mini"
-    )
-    
-    print("🎯 JIRA Agent Watch Example")
-    print("=" * 50)
-    print()
-    print("Available JIRA tools:")
-    print("- jira_watch_issue: Monitor a specific issue for changes")
-    print("- jira_watch_project: Monitor a project for new/updated issues")
-    print("- jira_get_issue_info: Get detailed info about an issue")  
-    print("- jira_search_issues: Search issues using JQL")
-    print()
-    
-    # Example interactions
-    example_prompts = [
-        "Get information about JIRA issue DEMO-1 from https://praisonai.atlassian.net",
-        "Search for open issues in project DEMO from https://praisonai.atlassian.net using JQL: 'project = DEMO AND status = Open'",
-        "Watch JIRA issue DEMO-1 from https://praisonai.atlassian.net for 2 minutes (check every 60 seconds, max 2 checks)",
-    ]
-    
-    print("📝 Example prompts you can try:")
-    for i, prompt in enumerate(example_prompts, 1):
-        print(f"{i}. {prompt}")
-    print()
-    
-    # Interactive mode
-    print("💬 Interactive JIRA Agent (type 'quit' to exit)")
-    print("-" * 50)
-    
-    while True:
-        try:
-            user_input = input("\n🧑 You: ").strip()
-            
-            if user_input.lower() in ['quit', 'exit', 'q']:
-                print("👋 Goodbye!")
-                break
-                
-            if not user_input:
-                continue
-                
-            print("🤖 Agent: ", end="", flush=True)
-            response = agent.start(user_input)
-            print(response)
-            
-        except KeyboardInterrupt:
-            print("\n👋 Goodbye!")
-            break
-        except Exception as e:
-            print(f"❌ Error: {e}")
-
-
-if __name__ == "__main__":
-    main()
@@ -6,6 +6,7 @@ to different next steps based on output content.
 """

 from praisonaiagents import AgentFlow, Task
+from praisonaiagents import AgentFlowManager

 # Create a workflow with branching
 workflow = AgentFlow(
@@ -35,9 +36,16 @@ workflow = AgentFlow(
 )

 if __name__ == "__main__":
-    result = workflow.run("", llm="gpt-4o-mini", verbose=True)
+    # Create manager and register workflow
+    manager = WorkflowManager()
+    manager.workflows["Decision Workflow"] = workflow
+    
+    # Execute - will branch to success_handler since 42 is positive
+    result = manager.execute(
+        "Decision Workflow",
+        default_llm="gpt-4o-mini"
+    )
+    
    print("Workflow completed!")
-    for step in result.get("steps", []):
-        name = step.get("step", step.get("name", "unknown"))
-        status = step.get("status", "completed")
-        print(f"  {name}: {status}")
+    for step_result in result["results"]:
+        print(f"  {step_result['step']}: {step_result['status']}")
@@ -6,8 +6,9 @@ functions that return StepResult with stop_workflow=True.
 """

 from praisonaiagents import (
-    AgentFlow, Task, WorkflowContext, StepResult
+    Workflow, Task, WorkflowContext, StepResult
 )
+from praisonaiagents import AgentFlowManager

 # Custom validator that can stop the workflow
 def validate_data(context: WorkflowContext) -> StepResult:
@@ -49,13 +50,20 @@ workflow = AgentFlow(
 )

 if __name__ == "__main__":
+    manager = WorkflowManager()
+    manager.workflows["Data Processing"] = workflow
+    
+    # Test with invalid data (will stop early)
    print("=== Testing with invalid data ===")
-    result = workflow.run("", llm="gpt-4o-mini", verbose=True)
-    for step in result.get("steps", []):
-        print(f"  {step.get('step', 'step')}: {step.get('output', step.get('status'))}")
-
+    result = manager.execute("Data Processing", default_llm="gpt-4o-mini")
+    
+    for step_result in result["results"]:
+        print(f"  {step_result['step']}: {step_result['output']}")
+    
+    # Test with valid data (will complete)
    print("\n=== Testing with valid data ===")
    workflow.variables["data"] = {"value": 42}
-    result = workflow.run("", llm="gpt-4o-mini", verbose=True)
-    for step in result.get("steps", []):
-        print(f"  {step.get('step', 'step')}: {step.get('status', 'completed')}")
+    result = manager.execute("Data Processing", default_llm="gpt-4o-mini")
+    
+    for step_result in result["results"]:
+        print(f"  {step_result['step']}: {step_result['status']}")
@@ -6,6 +6,7 @@ Each item is processed individually with the loop variable.
 """

 from praisonaiagents import AgentFlow, Task
+from praisonaiagents import AgentFlowManager

 # Create a workflow with loop
 workflow = AgentFlow(
@@ -29,12 +30,18 @@ workflow = AgentFlow(
 )

 if __name__ == "__main__":
-    result = workflow.run("", llm="gpt-4o-mini", verbose=True)
+    # Create manager and register workflow
+    manager = WorkflowManager()
+    manager.workflows["Process Items"] = workflow
+    
+    # Execute - will process each fruit in the list
+    result = manager.execute(
+        "Process Items",
+        default_llm="gpt-4o-mini"
+    )
+    
    print("Workflow completed!")
-    for step in result.get("steps", []):
-        name = step.get("step", step.get("name", "unknown"))
-        status = step.get("status", "completed")
-        print(f"  {name}: {status}")
-        out = step.get("output")
-        if isinstance(out, list):
-            print(f"    Loop processed {len(out)} items")
+    for step_result in result["results"]:
+        print(f"  {step_result['step']}: {step_result['status']}")
+        if isinstance(step_result.get("output"), list):
+            print(f"    Loop processed {len(step_result['output'])} items")
@@ -5,7 +5,7 @@ Demonstrates running multiple agents concurrently and
 combining their results with an aggregator agent.
 """

-from praisonaiagents import Agent, AgentFlow
+from praisonaiagents import Agent, Workflow
 from praisonaiagents.workflows import parallel

 # Create parallel research agents
@@ -5,7 +5,7 @@ Demonstrates decision-based routing where an agent classifier
 routes to specialized handler agents based on the request type.
 """

-from praisonaiagents import Agent, AgentFlow
+from praisonaiagents import Agent, Workflow
 from praisonaiagents.workflows import route

 # Create classifier agent
@@ -539,7 +539,7 @@ Repeat Phases 12 → 13 → 14 until the mergeability gate in Phase 13 is fully

 ### When the PR has merge conflicts

-If `mergeStateStatus=DIRTY`, the **Auto Merge Conflict → Claude** workflow (`.github/workflows/merge-conflict-claude.yml`) posts `@claude` automatically on push to `main` and on PR sync. You can still comment manually:
+If `mergeStateStatus=DIRTY`, ask Claude to rebase rather than resolving conflicts yourself on non-trivial logic:

 ```bash
 gh pr comment $PR --repo $REPO --body "@claude this PR has merge conflicts with \`main\` (PR #<X> was merged, touching the same regions). Please rebase onto latest \`main\`, re-run \`pytest <paths>\`, and force-push."
@@ -571,17 +571,6 @@ gh pr diff $PR --repo $REPO | less

 ### 15b. Merge

-**Primary:** Trigger or wait for **Claude PR merge gate** (`.github/workflows/claude-merge-gate.yml`):
-
-```bash
-gh workflow run claude-merge-gate.yml --repo $REPO -f pr_number=$PR
-gh pr view $PR --repo $REPO --json state,mergedAt,labels
-```
-
-The gate merges when `MERGE_GATE_VERDICT: APPROVE`, CI is green, and the PR lacks `no-auto-merge`. Do not merge while `claude-conflict-pending` or during active `@claude` review/fix.
-
-**Manual fallback** (opt-out or gate unavailable):
-
 ```bash
 gh pr merge $PR --repo $REPO --squash --delete-branch
 # --squash preserves linear main; --delete-branch keeps the repo clean
@@ -755,20 +755,6 @@ Developer Push ─────────────────────

 **Result**: `https://praison.ai/install.sh` is automatically updated **only when the source files change**!

-### Auto-deploy when install scripts change (PraisonAI → praison.ai)
-
-Pushing `install.sh` or `install.ps1` to PraisonAI **`main`** triggers [`.github/workflows/deploy-install-scripts.yml`](/Users/praison/praisonai-package/.github/workflows/deploy-install-scripts.yml), which dispatches the existing praison.ai **`CI`** workflow (`workflow_dispatch` on `master`). No manual website push or `make upgrade` is required.
-
-```mermaid
-flowchart LR
-  pushMain["PraisonAI push main\ninstall.sh|install.ps1"] --> deployWF["deploy-install-scripts.yml"]
-  deployWF --> dispatch["gh workflow run CI"]
-  dispatch --> siteCI["praison.ai azure-pipeline.yaml"]
-  siteCI --> live["https://praison.ai/install.sh"]
-```
-
-Manual re-deploy: `gh workflow run "Deploy install scripts to praison.ai" --repo MervinPraison/PraisonAI`
-
 ### How install.sh is Served

 The `web/install.sh` file is copied into the Docker image at `/var/www/html/web/install.sh` and served by nginx:
@@ -811,20 +797,20 @@ git push origin master

 **Task 2: Update install.sh and deploy**
 ```bash
-# 1. Update install.sh / install.ps1
+# 1. Update install.sh
 vim /Users/praison/praisonai-package/src/praisonai/scripts/install.sh

 # 2. Test install.sh
-bash /Users/praison/praisonai-package/src/praisonai/scripts/test-install-smoke.sh
+bash /Users/praison/praisonai-package/src/praisonai/scripts/install.sh --dry-run

-# 3. Commit and push to PraisonAI main — auto-triggers praison.ai deploy
-git add src/praisonai/scripts/install.sh src/praisonai/scripts/install.ps1
+# 3. Commit and push
+git add /Users/praison/praisonai-package/src/praisonai/scripts/install.sh
 git commit -m "fix: update install script"
 git push origin main
-# deploy-install-scripts.yml → praison.ai CI → https://praison.ai/install.sh

-# Manual re-deploy (optional):
-gh workflow run "Deploy install scripts to praison.ai" --repo MervinPraison/PraisonAI
+# 4. Deploy website (if needed)
+cd /Users/praison/Sites/localhost/praisonai
+make upgrade
 ```

 **Task 3: Rollback deployment**
@@ -137,15 +137,6 @@ Final rule: conclude only when evidence shows missing = 0.
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 PUBLISH WORKFLOW (reference)
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-GitHub Actions (recommended):
-  Manual: Actions → PyPI Release → Run workflow on main (bump defaults to patch)
-  Nightly: `.github/workflows/nightly-release-gate.yml` runs at 00:00 UTC (midnight) when:
-    - changes exist in src/praisonai or src/praisonai-agents since last v* tag
-    - Core Tests succeeded on current main HEAD
-    → dispatches pypi-release.yml with bump=patch, source=nightly
-    → still requires pypi environment approval before publish
-  Ad-hoc minor/major: manual PyPI Release only (explicit bump selection)
-
 Step 1 — Publish praisonaiagents (Core SDK):
  cd /Users/praison/praisonai-package/src/praisonai-agents
  praisonai publish pypi
@@ -170,4 +161,4 @@ Step 3 — Publish praisonai-tools (External, if needed):
  # Bump version in pyproject.toml, then:
  python3.13 -m build && uv run twine upload dist/*

-PraisonAI PRs: default merge via **Claude PR merge gate** (`.github/workflows/claude-merge-gate.yml`) when validation passes. Add label `no-auto-merge` to opt out. Manual `gh pr merge` only as fallback when the gate cannot run.
+PraisonAI PRs: NEVER merge automatically — user merges manually.
@@ -187,7 +187,7 @@ gh pr create \

 **Commit prefixes:** `fix:`, `feat:`, `security:`, `refactor:`, `docs:`, `test:`

-> **Merge:** Default is the **Claude PR merge gate** (`.github/workflows/claude-merge-gate.yml`). Do not merge manually unless the PR has `no-auto-merge` or you are explicitly told to use `gh pr merge`.
+> **Do NOT merge automatically** — user merges manually unless explicitly told otherwise.

 ### 5b. Verify PR state

@@ -129,9 +129,7 @@ python -m pytest tests/ -q \

 ## Step 6 — Approve or Request Changes

-**Primary (automated):** The **Claude PR merge gate** (`.github/workflows/claude-merge-gate.yml`) assesses readiness and merges when `MERGE_GATE_VERDICT: APPROVE`, CI is green, and the PR lacks `no-auto-merge`. It never posts `@claude` and does not use `gh pr review` (avoids re-triggering `claude.yml`).
-
-**Manual fallback — if the PR is good and gate is skipped or opted out:**
+**If the PR is good:**

 ```bash
 cd /Users/praison/praisonai-package && \
@@ -149,14 +147,6 @@ gh pr review <NUMBER> --request-changes --body "Requesting changes: <specific fe

 ## Step 7 — Merge

-**Primary:** Wait for merge gate or verify `mergedAt` after a gate run:
-
-```bash
-cd /Users/praison/praisonai-package && gh pr view <NUMBER> --json state,mergedAt,labels
-```
-
-**Manual fallback** (label `no-auto-merge` or gate disabled):
-
 ```bash
 cd /Users/praison/praisonai-package && \
 gh pr merge <NUMBER> --merge --subject "<PR title>"
@@ -267,21 +257,6 @@ find $(python -c "import site; print(site.getsitepackages()[0])") -name "<suspic

 ---

-## Fork PRs — contributor checklist and Claude push
-
-For PRs from external forks (`head.repo != MervinPraison/PraisonAI`), ask the contributor:
-
-1. **Allow edits from maintainers** on the PR (right sidebar).
-2. **Keep fork and branch** until the PR is merged or closed.
-
-Optional: private fork → add maintainer as collaborator; fork with Actions → only enable “Allow edits and access to secrets by maintainers” if they accept the risk.
-
-When `maintainerCanModify` is true, `claude.yml` pushes via `GH_TOKEN` to the fork remote and `merge-conflict-claude.yml` may auto-trigger `@claude` for `DIRTY` PRs (after FINAL architecture review).
-
-Manual unblock: `gh pr checkout <N> && git fetch origin main && git rebase origin/main && git push --force-with-lease`
-
---
-
 ## Quick Reference — Common Commands

 | Action | Command |
@@ -50,7 +50,7 @@ The workflow uses the agent configuration at `.github/praisonai-reviewer.yaml` a
 - Copilot NEVER triggers before CodeRabbit/Qodo complete.
 - Claude NEVER triggers before Copilot completes.
 - Duplicate `@copilot` comments are prevented by checking existing comments.
- Default merge: **Claude PR merge gate** (`.github/workflows/claude-merge-gate.yml`) after review chain + CI green. Opt out with label `no-auto-merge`. Manual merge only as fallback.
+- PraisonAI PRs: NEVER merge automatically — user merges manually.

 ## Security & Credentials

@@ -32,7 +32,7 @@ gh api repos/MervinPraison/PraisonAI/security-advisories/<GHSA_ID> --jq '{ghsa:
 3. For each advisory, validate the vulnerability in source code:
   - Read the file(s) mentioned in the advisory description
   - Confirm the vulnerability exists (or is already fixed)
-   - Classify using triage labels (see **Managing ongoing audits** below)
+   - Classify: `real` | `already-fixed` | `not-reproducible`
   - Reference the official repository for latest source: https://github.com/MervinPraison/PraisonAI

 ## PHASE 2 — VALIDATE & FIX
@@ -99,29 +99,9 @@ pip index versions praisonaiagents | head -1
 pip index versions praisonai | head -1
 ```

-13. Publish **praisonai-platform** (only when the advisory targets platform code):
-```bash
-cd /Users/praison/praisonai-package/src/praisonai-platform
-praisonai publish pypi
-# or: uv lock && uv build && uv publish --token $PYPI_TOKEN
-```
-Platform releases are **independent** of `praisonai` / `praisonaiagents` version numbers. A platform-only fix (e.g. `0.1.8`) does **not** require bumping the main SDK.
+## PHASE 4 — UPDATE ADVISORIES (only after packages are on PyPI)

-14. Publish **praisonai (npm / TypeScript)** (only when the advisory targets `src/praisonai-ts`):
-```bash
-cd /Users/praison/praisonai-package/src/praisonai-ts
-npm version patch && npm publish
-```
-
-15. Verify platform / npm when applicable:
-```bash
-pip index versions praisonai-platform | head -1
-npm view praisonai version
-```
-
-## PHASE 4 — UPDATE ADVISORIES (only after packages are on PyPI/npm)
-
-16. For each advisory, reopen to draft state and set patched versions + credits:
+13. For each advisory, reopen to draft state and set patched versions + credits:
 ```bash
 gh api repos/MervinPraison/PraisonAI/security-advisories/<GHSA_ID> \
  --method PATCH \
@@ -139,86 +119,34 @@ gh api repos/MervinPraison/PraisonAI/security-advisories/<GHSA_ID> \
 }
 EOF
 ```
- `<PACKAGE_NAME>`: `praisonaiagents` (core SDK), `praisonai` (Python wrapper), `praisonai-platform` (platform layer), or `praisonai` (npm ecosystem for TypeScript)
+- `<PACKAGE_NAME>`: `praisonaiagents` for core SDK, `praisonai` for wrapper
 - `<LAST_VULNERABLE_VERSION>`: version before the fix
 - `<FIRST_PATCHED_VERSION>`: the version just published with the fix
 - Always include credits for the reporter

-17. Publish each advisory:
+14. Publish each advisory:
 ```bash
 gh api repos/MervinPraison/PraisonAI/security-advisories/<GHSA_ID> \
  --method PATCH --field state=published
 ```

-18. Request CVE for each published advisory:
+15. Request CVE for each published advisory:
 ```bash
 gh api repos/MervinPraison/PraisonAI/security-advisories/<GHSA_ID>/cve --method POST
 ```
 GitHub reviews and assigns CVE IDs within 1-3 business days.

-19. Verify final state:
+16. Verify final state:
 ```bash
 gh api repos/MervinPraison/PraisonAI/security-advisories --jq '.[] | "\(.ghsa_id)  \(.state)  \(.cve_id // "pending")  \(.severity)  \([.credits[].login] | join(",") )  \(.summary | .[0:50])"'
 ```

-## MANAGING ONGOING AUDITS
-
-Continuous security reports are **expected** for an agent framework. Treat them as a product signal, not a crisis. Use a consistent process so triage stays manageable.
-
-### Triage labels
-
-After validating each advisory in source code, assign **one** label:
-
-| Label | Meaning | Action |
-|---|---|---|
-| `real` | Reproducible, valid security issue | Fix → publish → advisory |
-| `already-fixed` | Fixed in `main` but advisory still open | Set patched version to current release; publish advisory |
-| `duplicate` | Same root cause as another GHSA | Close or merge; link to primary advisory |
-| `not-reproducible` | Cannot reproduce with steps provided | Request more detail; hold in triage |
-| `theoretical` | Requires unlikely preconditions or no practical impact | Document reasoning; close as wontfix or low priority |
-| `documented-opt-out` | Behaviour is intentional with documented escape hatch | Close as wontfix; point to docs/env var |
-| `out-of-scope` | Wrong package, fork, or deployment the project does not ship | Close with explanation |
-
-### Triage load
-
- Not every report is valid — filter with the labels above before writing code.
- Batch related GHSAs (same file, same pattern) into one fix where possible.
- Prefer regression tests per GHSA so fixes are not re-opened by the next audit round.
-
-### Version churn and release boundaries
-
-| Package | Version line | Notes |
-|---|---|---|
-| `praisonaiagents` + `praisonai` | Aligned patch (e.g. `1.6.62` / `4.6.62`) | Main SDK + wrapper; bump together via `bump_and_release.py` |
-| `praisonai-platform` | Independent semver (e.g. `0.1.8`) | Platform-only fixes; users need **not** upgrade the main SDK |
-| `praisonai` (npm) | Independent semver (e.g. `1.7.2`) | TypeScript SDK; publish from `src/praisonai-ts` |
-
-Always set the advisory `patched_versions` to the **package that actually contains the fix**. Do not imply users must upgrade all packages when only one changed.
-
-### Researcher incentives
-
- Some reports are bounty-driven or low-quality — still acknowledge and triage promptly.
- Credit valid reporters in the advisory even for small fixes.
- Close invalid reports with a clear label and short rationale (no exploit discussion in public issues).
- Never fix “for the sake of it” — see Phase 2 validate checks.
-
-### When to close as wontfix
-
-Close (or publish advisory with no code change only if already mitigated) when:
-
- Documented opt-out exists and risk is accepted (e.g. localhost-only dev modes with explicit env vars).
- Issue is theoretical with no realistic attack path in supported deployments.
- Fix would remove existing features or materially bloat the SDK — escalate instead.
- Duplicate of an existing published advisory.
-
 ## REFERENCE — Package-to-file mapping

-| Package | Ecosystem | Canonical Path | Key files |
-|---|---|---|---|
-| praisonaiagents | pip | src/praisonai-agents/praisonaiagents/ | tools/, mcp/, memory/, agent/, sandbox/ |
-| praisonai | pip | src/praisonai/praisonai/ | cli/, capabilities/, mcp_server/, sandbox/, recipe/ |
-| praisonai-platform | pip | src/praisonai-platform/praisonai_platform/ | api/routes/, services/, auth |
-| praisonai | npm | src/praisonai-ts/src/ | agent/, mcp/, tools/, code-mode/ |
+| Package | Canonical Path | Key files |
+|---|---|---|
+| praisonaiagents | src/praisonai-agents/praisonaiagents/ | tools/python_tools.py, mcp/, memory/, agent/ |
+| praisonai | src/praisonai/praisonai/ | ui/sql_alchemy.py, capabilities/, mcp_server/, cli/ |

 ## RESOURCES & DOCUMENTATION

@@ -232,10 +160,8 @@ Close (or publish advisory with no code change only if already mitigated) when:
 - NEVER expose exploit details in public issues or PRs
 - ALWAYS credit the reporter in the advisory
 - ALWAYS set patched_versions before publishing advisory
- ALWAYS publish to PyPI/npm BEFORE publishing advisories (so Dependabot can point users to a fix)
- Python SDK: `praisonai publish pypi` from `src/praisonai-agents`, then `bump_and_release.py` for the wrapper
- Platform-only: `praisonai publish pypi` from `src/praisonai-platform` — no SDK bump required
- npm: `npm publish` from `src/praisonai-ts` when the TypeScript SDK is affected
+- ALWAYS publish to PyPI BEFORE publishing advisories (so Dependabot can point users to a fix) - to publish `praisonai publish pypi` from praisonai-agents folder, after that 
+`cd /Users/praison/praisonai-package/src/praisonai && python scripts/bump_and_release.py 4.5.125 --agents 1.5.125 --wait`. 4.5.125 is praisonai version number and 1.5.125 is praisonaiagents version number. Last 2 numbers will be same 4/1.xx.xxx
 - ALWAYS request CVE AFTER publishing advisory
 - Use generic commit messages: "refactor: harden X" not "fix: SQL injection in Y"
 - Fix on main, push directly — no feature branches for security patches
@@ -246,6 +172,7 @@ Close (or publish advisory with no code change only if already mitigated) when:

 ## REPORTING SECURITY ISSUES

-If you discover a security vulnerability, please report it responsibly. See [SECURITY.md](../../../../SECURITY.md) at the monorepo root for scope, supported packages, and the private advisory link.
-
-Maintainers and agents: use [.cursor/skills/security-audit/SKILL.md](../../.cursor/skills/security-audit/SKILL.md) for Cursor skill entry point.
+If you discover a security vulnerability, please report it responsibly:
+1. Open a draft security advisory at https://github.com/MervinPraison/PraisonAI/security/advisories/new
+2. Include detailed reproduction steps and impact assessment
+3. Allow time for the maintainers to address the issue before public disclosure
@@ -1,6 +1,6 @@
 # Documentation Parity Tracker (Python)

-> **Categories:** 66 | **Documented:** 65 | **Parity:** 98.5%
+> **Categories:** 60 | **Documented:** 57 | **Parity:** 95.0%

 This report compares **Python SDK feature categories** against **Python documentation** (docs/concepts, docs/features, etc.).

@@ -8,86 +8,108 @@ This report compares **Python SDK feature categories** against **Python document

 | Metric | Count |
 |--------|-------|
-| Feature Categories | 66 |
-| **Documented Categories** | **65** |
-| **Undocumented Categories** | **1** |
-| **Parity** | **98.5%** |
+| Feature Categories | 60 |
+| **Documented Categories** | **57** |
+| **Undocumented Categories** | **3** |
+| **Parity** | **95.0%** |

 ## Documented Categories

 | Category | Features | Docs | Lines |
 |----------|----------|------|-------|
 | ✅ AGUI | 1 | 2 | 420 |
-| ✅ Agent | 21 | 42 | 11340 |
-| ✅ Agent-to-Agent (A2A) | 1 | 7 | 2880 |
-| ✅ Agent-to-User (A2U) | 1 | 3 | 575 |
-| ✅ Approval | 1 | 3 | 1344 |
+| ✅ Agent | 19 | 30 | 8367 |
+| ✅ Agent-to-Agent (A2A) | 1 | 3 | 625 |
+| ✅ Approval | 1 | 2 | 889 |
 | ✅ Audio | 2 | 10 | 649 |
-| ✅ Auto Generation | 5 | 6 | 1637 |
-| ✅ Autonomy | 3 | 5 | 1919 |
-| ✅ Bots | 7 | 23 | 7714 |
-| ✅ Budget | 1 | 1 | 287 |
-| ✅ CLI | 5 | 108 | 26975 |
-| ✅ Chunking | 2 | 2 | 288 |
+| ✅ Auto Generation | 3 | 6 | 1386 |
+| ✅ Autonomy | 3 | 4 | 1146 |
+| ✅ Bots | 5 | 2 | 573 |
+| ✅ Chunking | 2 | 2 | 285 |
 | ✅ Citations | 2 | 1 | 202 |
-| ✅ Code Execution | 2 | 13 | 3554 |
-| ✅ Conditions | 1 | 3 | 1085 |
-| ✅ Configuration | 3 | 7 | 2160 |
-| ✅ Context Management | 16 | 33 | 10280 |
-| ✅ Database | 1 | 40 | 5373 |
-| ✅ Deep Research | 8 | 2 | 437 |
+| ✅ Code Execution | 3 | 13 | 3395 |
+| ✅ Conditions | 1 | 3 | 1086 |
+| ✅ Configuration | 2 | 7 | 1753 |
+| ✅ Context Management | 13 | 29 | 7576 |
+| ✅ Database | 1 | 31 | 3037 |
 | ✅ Display | 6 | 2 | 1175 |
 | ✅ Embeddings | 6 | 23 | 1966 |
-| ✅ Evaluation | 1 | 6 | 2539 |
-| ✅ Events | 1 | 2 | 764 |
-| ✅ Execution | 3 | 3 | 1099 |
-| ✅ Failover | 2 | 1 | 372 |
-| ✅ Files | 2 | 6 | 1632 |
-| ✅ Flow | 1 | 2 | 435 |
-| ✅ Gateway | 7 | 21 | 7477 |
-| ✅ Guardrails | 4 | 4 | 1619 |
-| ✅ Handoffs | 11 | 6 | 2499 |
-| ✅ Hooks | 2 | 8 | 3150 |
+| ✅ Evaluation | 1 | 6 | 2538 |
+| ✅ Events | 1 | 2 | 699 |
+| ✅ Execution | 3 | 2 | 515 |
+| ✅ Files | 2 | 5 | 863 |
+| ✅ Flow | 1 | 1 | 284 |
+| ✅ Gateway | 6 | 1 | 256 |
+| ✅ Guardrails | 4 | 4 | 2279 |
+| ✅ Handoffs | 10 | 4 | 1674 |
+| ✅ Hooks | 2 | 6 | 1606 |
 | ✅ Image | 1 | 11 | 1096 |
-| ✅ Knowledge | 4 | 24 | 4525 |
-| ✅ LLM | 3 | 10 | 3674 |
-| ✅ Loops | 4 | 3 | 774 |
-| ✅ MCP | 1 | 51 | 10939 |
-| ✅ Memory | 6 | 16 | 5879 |
-| ✅ OCR | 2 | 1 | 226 |
-| ✅ Observability | 2 | 23 | 2754 |
-| ✅ Optimizer | 1 | 2 | 653 |
-| ✅ Output | 3 | 4 | 860 |
-| ✅ Parallel Execution | 3 | 2 | 399 |
-| ✅ Planning | 6 | 5 | 1198 |
-| ✅ Plugins | 8 | 3 | 1502 |
-| ✅ Prompts | 2 | 7 | 1569 |
-| ✅ Providers | 1 | 52 | 6214 |
+| ✅ Knowledge | 3 | 12 | 3262 |
+| ✅ LLM | 2 | 2 | 1049 |
+| ✅ Loops | 4 | 1 | 250 |
+| ✅ MCP | 2 | 47 | 9754 |
+| ✅ Memory | 6 | 12 | 4474 |
+| ✅ Output | 3 | 4 | 868 |
+| ✅ Parallel Execution | 2 | 2 | 347 |
+| ✅ Planning | 6 | 5 | 1506 |
+| ✅ Plugins | 8 | 3 | 1494 |
+| ✅ Prompts | 2 | 4 | 796 |
+| ✅ Providers | 2 | 50 | 5396 |
 | ✅ Query | 1 | 2 | 616 |
-| ✅ RAG | 5 | 15 | 3193 |
+| ✅ RAG | 5 | 14 | 2910 |
 | ✅ Realtime | 2 | 5 | 584 |
-| ✅ Reflection | 3 | 2 | 477 |
-| ✅ Retrieval | 2 | 5 | 945 |
+| ✅ Reflection | 3 | 2 | 469 |
+| ✅ Retrieval | 2 | 5 | 941 |
 | ✅ Routing | 1 | 2 | 597 |
-| ✅ Sandbox | 5 | 6 | 2512 |
-| ✅ Security | 1 | 3 | 2563 |
-| ✅ Sessions | 4 | 12 | 3793 |
-| ✅ Skills | 6 | 10 | 3663 |
-| ✅ Tasks | 2 | 6 | 3064 |
-| ✅ Telemetry | 1 | 2 | 560 |
+| ✅ Sandbox | 5 | 3 | 1002 |
+| ✅ Security | 1 | 1 | 875 |
+| ✅ Sessions | 2 | 8 | 2128 |
+| ✅ Skills | 5 | 5 | 2002 |
+| ✅ Tasks | 2 | 5 | 2590 |
+| ✅ Telemetry | 1 | 2 | 541 |
 | ✅ Templates | 1 | 7 | 1411 |
-| ✅ Tools | 12 | 118 | 25895 |
+| ✅ Tools | 4 | 97 | 17827 |
 | ✅ Tracing | 3 | 2 | 139 |
-| ✅ Video | 2 | 6 | 510 |
+| ✅ Video | 2 | 5 | 460 |
 | ✅ Vision | 2 | 1 | 283 |
-| ✅ Web | 3 | 7 | 1537 |
-| ✅ Workflows | 4 | 14 | 5893 |
+| ✅ Web | 4 | 6 | 1267 |
+| ✅ Workflows | 4 | 9 | 3579 |
+
+## Stub Documentation (Need Content)
+
+These categories have documentation files but < 50 lines (stubs):
+
+| Category | Features | Docs | Lines |
+|----------|----------|------|-------|
+| ⚠️ OCR | 2 | 1 | 49 |

 ## Undocumented Categories (Need Documentation)

 | Category | Features |
 |----------|----------|
-| ❌ Vector Store | 1 |
+| ❌ Failover | 2 |
+| ❌ Optimizer | 1 |
+
+## Documentation Without Features
+
+These docs exist but don't match any implemented feature category:
+
+- ℹ️ AI SDK (1 docs, 112 lines)
+- ℹ️ Agent-to-User (A2U) (3 docs, 770 lines)
+- ℹ️ CLI (91 docs, 21386 lines)
+- ℹ️ Callbacks (1 docs, 985 lines)
+- ℹ️ Chat (3 docs, 546 lines)
+- ℹ️ Deep Research (2 docs, 437 lines)
+- ℹ️ Deployment (29 docs, 5931 lines)
+- ℹ️ Jobs (1 docs, 358 lines)
+- ℹ️ Middleware (1 docs, 60 lines)
+- ℹ️ Observability (21 docs, 1580 lines)
+- ℹ️ Process (1 docs, 506 lines)
+- ℹ️ Scheduler (2 docs, 858 lines)
+- ℹ️ Storage (4 docs, 722 lines)
+- ℹ️ Streaming (3 docs, 485 lines)
+- ℹ️ Token Management (3 docs, 1020 lines)
+- ℹ️ Vector Store (11 docs, 733 lines)

 ---

@@ -1,70 +1,35 @@
 # PraisonAI Agents - Benchmark Results

-**Generated:** 2026-06-17
-**Methodology:** 10 warmup runs + 1000 measured iterations (`performance_benchmark.py` config)
-**Config:** `Agent(name=..., instructions=..., model="gpt-4o-mini", tools=[get_weather], output="silent")`
+**Generated:** 2025-12-18 14:40:22
+**Iterations:** 100
+**Test:** Agent instantiation (without tools)

-## Agent Instantiation Time
+## Results

-| Framework | Avg Time (μs) | Min (μs) | Median (μs) | Relative |
-|-----------|---------------|----------|-------------|----------|
-| OpenAI Agents SDK | 5.05 | — | — | 1.00x (fastest) |
-| Agno | 5.27 | — | — | 1.04x |
-| **PraisonAI** | **13.62** | **12.58** | **12.96** | **2.69x** |
-| PraisonAI (LiteLLM) | 14.49 | 13.37 | 13.71 | 2.86x |
-
-### PraisonAI detail (1000 iter)
-
-| Variant | Avg (μs) | Min (μs) | Median (μs) | Std (μs) |
-|---------|----------|----------|-------------|----------|
-| With tools (`gpt-4o-mini`) | 13.62 | 12.58 | 12.96 | 2.73 |
-| With tools (`openai/gpt-4o-mini`) | 14.49 | 13.37 | 13.71 | 4.14 |
-| Minimal, no tools | 13.40 | — | — | — |
-
-## Historical baseline
-
-| Source | Config | Avg (μs) | Notes |
-|--------|--------|----------|-------|
-| `performance_benchmark.py` (Dec 2025) | tools + 1000 iter | **3.20** | Pre-init-bloat baseline |
-| `simple_benchmark.py` (Dec 2025) | no tools, 100 iter | **3.77** | Saved in prior `BENCHMARK_RESULTS.md` |
-| Current (Jun 2026) | tools + 1000 iter | **13.62** | After lazy-init optimisations (was ~31 μs pre-fix) |
+| Framework | Avg Time (μs) | Relative |
+|-----------|---------------|----------|
+| **PraisonAI** | **3.77** | **1.00x (fastest)** |
+| OpenAI Agents SDK | 5.26 | 1.39x |
+| Agno | 5.64 | 1.49x |
+| PraisonAI (LiteLLM) | 7.56 | 2.00x |
+| PydanticAI | 226.94 | 60.16x |
+| LangGraph | 4,558.71 | 1,209x |
+| CrewAI | 15,607.92 | 4,138x |

 ## Package Versions

 | Package | Version |
-|---------|---------|
-| PraisonAI | 1.6.60 |
-| Agno | 2.4.2 |
-| OpenAI Agents SDK | 0.7.0 |
+|---------|--------|
+| PraisonAI | 0.1.5 |
+| Agno | 2.3.14 |
+| PydanticAI | 1.35.0 |
+| OpenAI Agents SDK | 0.6.3 |
+| LangGraph | 1.0.5 |
+| CrewAI | 1.6.1 |

 ## How to Reproduce

-PraisonAI-only (fast; avoids CrewAI/LangGraph import hang):
-
 ```bash
 cd praisonai-agents
-PYTHONWARNINGS='ignore::DeprecationWarning' python -c "
-import time, statistics
-from typing import Literal
-from praisonaiagents import Agent
-
-def get_weather(city: Literal['nyc', 'sf']):
-    return 'cloudy' if city == 'nyc' else 'sunny'
-
-def bench(fn, n=1000, w=10):
-    for _ in range(w): fn()
-    times = []
-    for _ in range(n):
-        s = time.perf_counter(); fn(); times.append(time.perf_counter()-s)
-    print(f'avg={statistics.mean(times)*1e6:.2f} μs')
-
-bench(lambda: Agent(name='Test', instructions='Hi', model='gpt-4o-mini', tools=[get_weather], output='silent'))
-"
-```
-
-Full multi-framework benchmark:
-
-```bash
-cd praisonai-agents
-PYTHONWARNINGS='ignore::DeprecationWarning' python benchmarks/performance_benchmark.py --save
+python benchmarks/simple_benchmark.py
 ```
@@ -1,62 +0,0 @@
-# Agent Instantiation — Root Cause Analysis
-
-**Date:** 2026-06-17  
-**Version:** praisonaiagents 1.6.58 (dev)  
-**Methodology:** 10 warmup + 1000 measured iterations, `time.perf_counter()`
-
-## Summary
-
-Instantiation regressed from **~3.2 μs** (Dec 2025) to **~31 μs** (~10×). After fixes, steady-state is **~16 μs** (~2× historical). PraisonAI is competitive with Agno (~5 μs) on warmed benchmarks using `model=`.
-
-## Before vs After Fixes
-
-| Config | Historical | Pre-fix | Post-fix |
-|--------|------------|---------|----------|
-| Minimal (`name`, `output="silent"`) | 3.20 μs | 31.36 μs | **16.35 μs** |
-| `model="gpt-4o-mini"` | 3.77 μs | 32.19 μs | **16.86 μs** |
-| `llm="gpt-4o-mini"` (deprecated) | 3.77 μs | 46.48 μs | **~18 μs** (warn once) |
-| LiteLLM `openai/gpt-4o-mini` cold | 7.31 μs | **3.7 s** | **53.63 μs** |
-| LiteLLM warmed instantiate | — | 60 μs | **19.83 μs** |
-| Import `Agent` | — | 20.8 ms | **29.6 ms** |
-
-## Root Causes (validated)
-
-| ID | Cause | Status | Fix applied |
-|----|-------|--------|-------------|
-| H1 | Config loader cold start (+9 ms first init) | CONFIRMED | Fast path when no config defaults (`_defaults_has_any_values`) |
-| H2 | LoopGuard on every init (~2 μs) | CONFIRMED | Lazy `_ensure_loop_guard()` on first tool call |
-| H3 | Approval presets | REJECTED | No change needed |
-| H4 | `llm=` deprecation every init (+16 μs) | CONFIRMED | Warn once per process in `warn_deprecated_param` |
-| H5 | Init bloat / config resolution (~25 μs) | CONFIRMED | Partially addressed via loader fast path |
-| H6 | LiteLLM eager `LLM()` import (+3.7 s) | CONFIRMED | Deferred `_llm_init_params` + lazy `llm_instance` property |
-| H7 | `ExecutionConfig` deprecation every init | CONFIRMED | Fixed stack walk; internal SDK path cached |
-
-## cProfile — Post-fix (500 warmed `model=` inits)
-
-| Function | Time (ms) | Per call |
-|----------|-----------|----------|
-| `Agent.__init__` | 8 | ~16 μs |
-| `apply_config_defaults` | 1 | ~2 μs |
-| `ExecutionConfig.__post_init__` | 1 | ~2 μs (first call only walks stack) |
-| `_warnings.warn` | **0** | removed from hot path |
-
-## Fixes implemented
-
-1. **`praisonaiagents/config/feature_configs.py`** — Fix `ExecutionConfig.__post_init__` deprecation guard; cache internal SDK detection.
-2. **`praisonaiagents/utils/deprecation.py`** — Warn once per process for deprecated params.
-3. **`praisonaiagents/config/loader.py`** — Skip `apply_config_defaults` lookups when config has no defaults.
-4. **`praisonaiagents/agent/agent.py`** — Lazy LoopGuard, lazy LLM instance, deferred LiteLLM params.
-5. **`benchmarks/simple_benchmark.py`** — Use `model=`, add 10-iteration warmup.
-6. **`benchmarks/performance_benchmark.py`** — Use `model=` instead of deprecated `llm=`.
-
-## Remaining gap (~16 μs vs ~3 μs)
-
-The residual ~13 μs is primarily **`Agent.__init__` size** (~1200 lines of consolidated config resolution added since v0.1.5). Further optimisation requires a dedicated slim fast-path for `Agent(name=..., output="silent")` without full config object churn.
-
-## How to reproduce
-
-```bash
-cd praisonai-agents
-PYTHONWARNINGS='ignore::DeprecationWarning' python benchmarks/simple_benchmark.py
-PYTHONWARNINGS='ignore::DeprecationWarning' python benchmarks/performance_benchmark.py
-```
@@ -120,8 +120,8 @@ def run_benchmark(iterations: int = 100) -> List[ParamResult]:
        ('backstory', 'I am an AI assistant.', 'backstory="..."'),
        
        # LLM configuration
-        ('model', 'gpt-4o-mini', 'model="gpt-4o-mini"'),
-        ('model', 'openai/gpt-4o-mini', 'model="openai/gpt-4o-mini" (LiteLLM)'),
+        ('llm', 'gpt-4o-mini', 'llm="gpt-4o-mini"'),
+        ('llm', 'openai/gpt-4o-mini', 'llm="openai/gpt-4o-mini" (LiteLLM)'),
        
        # Tools
        ('tools', [sample_tool], 'tools=[sample_tool]'),
@@ -298,7 +298,7 @@ def save_results(results: List[ParamResult], filename: str = 'PARAM_IMPACT_RESUL
        f.write('### Best Practices\n\n')
        f.write('1. Use `output="silent"` for production workloads\n')
        f.write('2. Only enable `context=True` when needed for long conversations\n')
-        f.write('3. Use `model="gpt-4o-mini"` for fastest instantiation\n')
+        f.write('3. Use `llm="gpt-4o-mini"` (OpenAI SDK) for fastest instantiation\n')
        f.write('4. Avoid `memory=True` unless persistent memory is required\n')
        
        f.write('\n## How to Reproduce\n\n')
@@ -185,7 +185,7 @@ def instantiate_praisonai_agent():
    return PraisonAgent(
        name="Test Agent",
        instructions="Be concise, reply with one sentence.",
-        model="gpt-4o-mini",
+        llm="gpt-4o-mini",
        tools=tools,
        output="silent"
    )
@@ -196,7 +196,7 @@ def instantiate_praisonai_litellm_agent():
    return PraisonAgent(
        name="Test Agent",
        instructions="Be concise, reply with one sentence.",
-        model="openai/gpt-4o-mini",
+        llm="openai/gpt-4o-mini",
        tools=tools,
        output="silent"
    )
@@ -14,7 +14,6 @@ import argparse


 ITERATIONS = 100
-WARMUP = 10


 def sample_tool(city: Literal['nyc', 'sf']):
@@ -24,10 +23,8 @@ def sample_tool(city: Literal['nyc', 'sf']):
    return 'sunny'


-def measure_instantiation(create_fn, iterations=ITERATIONS, warmup=WARMUP):
+def measure_instantiation(create_fn, iterations=ITERATIONS):
    """Measure average instantiation time in microseconds."""
-    for _ in range(warmup):
-        create_fn()
    times = []
    for _ in range(iterations):
        start = time.perf_counter()
@@ -51,11 +48,11 @@ def run_benchmark():
    from praisonaiagents import Agent as PraisonAgent
    
    results['PraisonAI'] = measure_instantiation(
-        lambda: PraisonAgent(name='Test', model='gpt-4o-mini', output="silent")
+        lambda: PraisonAgent(name='Test', llm='gpt-4o-mini', output="silent")
    )
    
    results['PraisonAI (LiteLLM)'] = measure_instantiation(
-        lambda: PraisonAgent(name='Test', model='openai/gpt-4o-mini', output="silent")
+        lambda: PraisonAgent(name='Test', llm='openai/gpt-4o-mini', output="silent")
    )
    
    # Other frameworks for comparison
@@ -1,75 +0,0 @@
-#!/usr/bin/env python3
-"""Example of using LLMConfig with Agent for model fallback configuration."""
-
-from praisonaiagents import Agent
-from praisonaiagents.config import LLMConfig
-
-# Example 1: Pass LLMConfig via llm parameter
-def example_llm_param():
-    """Using LLMConfig via the llm parameter."""
-    config = LLMConfig(
-        model="gpt-4o",
-        fallback_models=["claude-3-5-sonnet", "gpt-4o-mini"],
-        base_url="https://api.example.com",
-        api_key="test-key"
-    )
-    
-    agent = Agent(
-        name="Assistant",
-        instructions="You are a helpful assistant",
-        llm=config  # Pass LLMConfig object directly
-    )
-    
-    print(f"Primary model: {agent.llm}")
-    print(f"Fallback models: {agent.fallback_models}")
-    print(f"Base URL: {agent.base_url}")
-    print(f"API Key: {agent.api_key}")
-
-# Example 2: Pass LLMConfig via model parameter
-def example_model_param():
-    """Using LLMConfig via the model parameter (preferred)."""
-    config = LLMConfig(
-        model="gpt-4o", 
-        fallback_models=["claude-3-5-sonnet", "gpt-4o-mini"],
-        base_url="https://api.example.com",
-        api_key="test-key"
-    )
-    
-    agent = Agent(
-        name="Assistant",
-        instructions="You are a helpful assistant",
-        model=config  # Pass LLMConfig via model parameter
-    )
-    
-    print(f"Primary model: {agent.llm}")
-    print(f"Fallback models: {agent.fallback_models}")
-    print(f"Base URL: {agent.base_url}")
-    print(f"API Key: {agent.api_key}")
-
-# Example 3: Traditional usage still works
-def example_traditional():
-    """Traditional usage without LLMConfig (no fallback support)."""
-    agent = Agent(
-        name="Assistant",
-        instructions="You are a helpful assistant",
-        model="gpt-4o",
-        base_url="https://api.example.com",
-        api_key="test-key"
-    )
-    
-    print(f"Primary model: {agent.llm}")
-    print(f"Fallback models: {agent.fallback_models}")  # Will be empty
-    print(f"Base URL: {agent.base_url}")
-    print(f"API Key: {agent.api_key}")
-
-if __name__ == "__main__":
-    print("=== Example 1: LLMConfig via llm parameter ===")
-    example_llm_param()
-    print()
-    
-    print("=== Example 2: LLMConfig via model parameter ===")
-    example_model_param()
-    print()
-    
-    print("=== Example 3: Traditional usage (no fallback) ===")
-    example_traditional()
@@ -114,9 +114,6 @@ def _get_lazy_cache():
 # ============================================================================

 _LAZY_IMPORTS = {
-    # Version information (lazy loaded to avoid file I/O on import)
-    '__version__': ('praisonaiagents._version', '__version__'),
-    
    # Tools (moved from eager imports for lazy loading)
    'Tools': ('praisonaiagents.tools.tools', 'Tools'),
    'BaseTool': ('praisonaiagents.tools.base', 'BaseTool'),
@@ -178,11 +175,6 @@ _LAZY_IMPORTS = {
    'loop': ('praisonaiagents.workflows', 'loop'),
    'repeat': ('praisonaiagents.workflows', 'repeat'),
    'when': ('praisonaiagents.workflows', 'when'),
-    'if_': ('praisonaiagents.workflows', 'if_'),
-    'include': ('praisonaiagents.workflows', 'include'),
-    'Include': ('praisonaiagents.workflows', 'Include'),
-    'MAX_NESTING_DEPTH': ('praisonaiagents.workflows', 'MAX_NESTING_DEPTH'),
-    'YAMLWorkflowParser': ('praisonaiagents.workflows', 'YAMLWorkflowParser'),
    
    # Conditions (Protocol-driven condition evaluation)
    'ConditionProtocol': ('praisonaiagents.conditions.protocols', 'ConditionProtocol'),
@@ -235,15 +227,8 @@ _LAZY_IMPORTS = {
    'CliBackendResult': ('praisonaiagents.cli_backend.protocols', 'CliBackendResult'),
    'CliBackendDelta': ('praisonaiagents.cli_backend.protocols', 'CliBackendDelta'),
    
-    # Runtime Selection (Auto harness selection and registry)
-    'AgentRuntimeProtocol': ('praisonaiagents.runtime.protocols', 'AgentRuntimeProtocol'),
-    'resolve_runtime': ('praisonaiagents.runtime.registry', 'resolve_runtime'),
-    'register_runtime': ('praisonaiagents.runtime.registry', 'register_runtime'),
-    'list_runtimes': ('praisonaiagents.runtime.registry', 'list_runtimes'),
-    
    # Agent classes
    'Agent': ('praisonaiagents.agent.agent', 'Agent'),
-    'RetryBackoffConfig': ('praisonaiagents.agent.retry_utils', 'RetryBackoffConfig'),
    'BudgetExceededError': ('praisonaiagents.errors', 'BudgetExceededError'),
    
    # Error hierarchy - structured exception handling
@@ -313,12 +298,6 @@ _LAZY_IMPORTS = {
    # Knowledge
    'Knowledge': ('praisonaiagents.knowledge.knowledge', 'Knowledge'),
    'Chunking': ('praisonaiagents.knowledge.chunking', 'Chunking'),
-    'KnowledgeStoreProtocol': ('praisonaiagents.knowledge.protocols', 'KnowledgeStoreProtocol'),
-    'ScopeRequiredError': ('praisonaiagents.knowledge.protocols', 'ScopeRequiredError'),
-    'BackendNotAvailableError': ('praisonaiagents.knowledge.protocols', 'BackendNotAvailableError'),
-    'FileTracker': ('praisonaiagents.knowledge.indexing', 'FileTracker'),
-    'IndexResult': ('praisonaiagents.knowledge.indexing', 'IndexResult'),
-    'CorpusStats': ('praisonaiagents.knowledge.indexing', 'CorpusStats'),
    
    # FastContext
    'FastContext': ('praisonaiagents.context.fast', 'FastContext'),
@@ -342,24 +321,9 @@ _LAZY_IMPORTS = {
    'SkillProperties': ('praisonaiagents.skills', 'SkillProperties'),
    'SkillMetadata': ('praisonaiagents.skills', 'SkillMetadata'),
    'SkillLoader': ('praisonaiagents.skills', 'SkillLoader'),
-    'validate': ('praisonaiagents.skills', 'validate'),
-    'validate_metadata': ('praisonaiagents.skills', 'validate_metadata'),
-    'discover_skills': ('praisonaiagents.skills', 'discover_skills'),
-    'load_skill': ('praisonaiagents.skills', 'load_skill'),
-    'EnforcementLevel': ('praisonaiagents.skills', 'EnforcementLevel'),
-    'SkillState': ('praisonaiagents.skills', 'SkillState'),
    
    # Memory
    'Memory': ('praisonaiagents.memory.memory', 'Memory'),
-    'AutoMemory': ('praisonaiagents.memory', 'AutoMemory'),
-    'ChromaMemory': ('praisonaiagents.memory.adapters.factories', 'create_chroma_memory_adapter'),
-    'register_memory_adapter': ('praisonaiagents.memory.adapters', 'register_memory_adapter'),
-    'register_memory_factory': ('praisonaiagents.memory.adapters', 'register_memory_factory'),
-    'get_memory_adapter': ('praisonaiagents.memory.adapters', 'get_memory_adapter'),
-    'list_memory_adapters': ('praisonaiagents.memory.adapters', 'list_memory_adapters'),
-    'add_memory_adapter': ('praisonaiagents.memory.adapters.registry', 'add_memory_adapter'),
-    'add_memory_factory': ('praisonaiagents.memory.adapters.registry', 'add_memory_factory'),
-    'has_memory_adapter': ('praisonaiagents.memory.adapters', 'has_memory_adapter'),
    
    # Planning
    'Plan': ('praisonaiagents.planning', 'Plan'),
@@ -509,7 +473,7 @@ _LAZY_IMPORTS = {
    'GatewayClientProtocol': ('praisonaiagents.gateway.protocols', 'GatewayClientProtocol'),
    'GatewayEvent': ('praisonaiagents.gateway.protocols', 'GatewayEvent'),
    'GatewayMessage': ('praisonaiagents.gateway.protocols', 'GatewayMessage'),
-    'GatewayEventType': ('praisonaiagents.gateway.protocols', 'EventType'),
+    'EventType': ('praisonaiagents.gateway.protocols', 'EventType'),
    'GatewayConfig': ('praisonaiagents.gateway.config', 'GatewayConfig'),
    'SessionConfig': ('praisonaiagents.gateway.config', 'SessionConfig'),
    
@@ -811,12 +775,8 @@ def warmup(include_litellm: bool = False, include_openai: bool = True) -> dict:
 # ============================================================================

 __all__ = [
-    # Version information
-    '__version__',
-    
    # Core classes - the essentials
    'Agent',
-    'RetryBackoffConfig',
    'AgentTeam',  # Primary class for multi-agent coordination (v1.0+)
    'AgentManager',  # Silent alias for AgentTeam
    'Agents',  # Deprecated alias for AgentTeam (emits warning)
@@ -111,3 +111,32 @@ def _get_plugins_list() -> list:

 PLUGINS_ENABLED = _get_plugins_enabled()
 PLUGINS_LIST = _get_plugins_list()
+
+
+def missing_dependency_error(
+    package_name: str,
+    extra_name: Optional[str] = None,
+    feature_name: Optional[str] = None
+) -> ImportError:
+    """
+    Create a standardized ImportError for missing optional dependencies.
+    
+    Args:
+        package_name: The name of the missing package (e.g., 'litellm')
+        extra_name: The pip extra to install (e.g., 'llm' for [llm])
+        feature_name: Human-readable feature name (e.g., 'LLM support')
+    
+    Returns:
+        ImportError with helpful install instructions
+    """
+    if extra_name:
+        install_cmd = f"pip install 'praisonaiagents[{extra_name}]'"
+    else:
+        install_cmd = f"pip install {package_name}"
+    
+    feature_desc = f" for {feature_name}" if feature_name else ""
+    
+    return ImportError(
+        f"{package_name} is required{feature_desc} but not installed. "
+        f"Please install with: {install_cmd}"
+    )
@@ -0,0 +1,49 @@
+"""
+Resolver helper functions for consolidated parameters.
+
+These helpers wrap the canonical resolver for specific parameter types,
+handling special cases like callable validators for guardrails.
+"""
+
+from typing import Any, Type
+
+from .config.param_resolver import resolve, ArrayMode
+from .config.presets import GUARDRAIL_PRESETS
+
+
+def resolve_guardrails(value: Any, config_class: Type) -> Any:
+    """
+    Resolve guardrails parameter.
+    
+    Handles special case of callable validators which have highest precedence.
+    
+    Args:
+        value: The guardrails parameter value
+        config_class: GuardrailConfig class
+        
+    Returns:
+        Resolved config, callable, or string
+    """
+    # Handle None
+    if value is None:
+        return None
+    
+    # Handle False (disabled)
+    if value is False:
+        return None
+    
+    # Handle callable (highest precedence after None check)
+    # This includes functions, lambdas, and callable objects
+    if callable(value) and not isinstance(value, type):
+        return value
+    
+    # Use canonical resolver for everything else
+    return resolve(
+        value=value,
+        param_name="guardrails",
+        config_class=config_class,
+        presets=GUARDRAIL_PRESETS,
+        array_mode=ArrayMode.PRESET_OVERRIDE,
+        string_mode="llm_prompt",
+        default=None,
+    )
@@ -1,51 +0,0 @@
-"""
-Version utilities for praisonaiagents package.
-
-This module provides version information by reading from pyproject.toml
-to avoid duplication and ensure the single source of truth.
-"""
-
-import re
-from pathlib import Path
-
-
-def get_version() -> str:
-    """
-    Get the version string from pyproject.toml.
-    
-    Returns:
-        str: Version string (e.g., "1.6.52")
-        
-    Raises:
-        RuntimeError: If version cannot be found or parsed
-    """
-    try:
-        # Get the package root (praisonaiagents directory)
-        package_root = Path(__file__).parent
-        # Go up to src/praisonai-agents directory
-        pyproject_path = package_root.parent / "pyproject.toml"
-        
-        if not pyproject_path.exists():
-            raise RuntimeError(f"pyproject.toml not found at {pyproject_path}")
-            
-        content = pyproject_path.read_text()
-        
-        # Look for version = "X.Y.Z" pattern
-        match = re.search(r'version\s*=\s*["\']([^"\']+)["\']', content)
-        if not match:
-            raise RuntimeError("Version not found in pyproject.toml")
-            
-        return match.group(1)
-        
-    except Exception as e:
-        # Fallback to "unknown" to avoid breaking imports
-        import warnings
-        warnings.warn(
-            f"Failed to read version from pyproject.toml: {e}. Using 'unknown'.",
-            RuntimeWarning
-        )
-        return "unknown"
-
-
-# Cache the version to avoid reading the file multiple times
-__version__ = get_version()
@@ -1,71 +0,0 @@
-"""
-Shared lazy-loaders for display utilities used across the Agent class and mixins.
-
-These helpers defer importing rich and the heavy ``..main`` display module until
-first use, which keeps import time low for silent mode and avoids circular
-imports during package initialisation. Results are cached in module globals and
-protected by per-resource locks using double-checked locking, so each import is
-resolved at most once across threads. Independent resources use separate locks so
-that a slow first import of one does not stall initialisation of the others.
-"""
-
-import threading
-
-# Lazy-loaded modules (populated on first use, each protected by its own lock)
-_console_lock = threading.Lock()
-_live_lock = threading.Lock()
-_main_module_lock = threading.Lock()
-_rich_console = None
-_rich_live = None
-_main_module = None
-
-
-def _get_console():
-    """Lazy load rich.console.Console (thread-safe)."""
-    global _rich_console
-    if _rich_console is None:
-        with _console_lock:
-            if _rich_console is None:
-                from rich.console import Console
-                _rich_console = Console
-    return _rich_console
-
-
-def _get_live():
-    """Lazy load rich.live.Live (thread-safe)."""
-    global _rich_live
-    if _rich_live is None:
-        with _live_lock:
-            if _rich_live is None:
-                from rich.live import Live
-                _rich_live = Live
-    return _rich_live
-
-
-def _get_display_functions():
-    """Lazy load display functions from main module (thread-safe)."""
-    global _main_module
-    if _main_module is None:
-        with _main_module_lock:
-            if _main_module is None:
-                from ..main import (
-                    display_error,
-                    display_instruction,
-                    display_interaction,
-                    display_generating,
-                    display_self_reflection,
-                    ReflectionOutput,
-                    adisplay_instruction,
-                    execute_sync_callback
-                )
-                _main_module = {
-                    'display_error': display_error,
-                    'display_instruction': display_instruction,
-                    'display_interaction': display_interaction,
-                    'display_generating': display_generating,
-                    'display_self_reflection': display_self_reflection,
-                    'ReflectionOutput': ReflectionOutput,
-                    'adisplay_instruction': adisplay_instruction,
-                    'execute_sync_callback': execute_sync_callback,
-                }
-    return _main_module
@@ -13,8 +13,31 @@ import logging
 from praisonaiagents._logging import get_logger
 from ..errors import BudgetExceededError

-# Shared lazy display helpers (cached, thread-safe; avoid circular imports)
-from ._lazy_display import _get_console, _get_live, _get_display_functions
+# Fallback helpers to avoid circular imports
+def _get_console():
+    from rich.console import Console
+    return Console
+
+def _get_live():
+    from rich.live import Live
+    return Live
+
+def _get_display_functions():
+    from ..main import (
+        display_error, display_instruction, display_interaction,
+        display_generating, display_self_reflection, ReflectionOutput,
+        adisplay_instruction, execute_sync_callback
+    )
+    return {
+        'display_error': display_error,
+        'display_instruction': display_instruction,
+        'display_interaction': display_interaction,
+        'display_generating': display_generating,
+        'display_self_reflection': display_self_reflection,
+        'ReflectionOutput': ReflectionOutput,
+        'adisplay_instruction': adisplay_instruction,
+        'execute_sync_callback': execute_sync_callback,
+    }

 logger = logging.getLogger(__name__)

@@ -64,41 +87,12 @@ Your Goal: {self.goal}"""
        
        # Add memory context if memory is enabled
        if self._memory_instance:
-            # Use cache-optimized context if model supports prompt caching and cache boundary is requested
-            use_cache_boundaries = hasattr(self, '_model_supports_prompt_caching') and self._model_supports_prompt_caching()
-            if use_cache_boundaries and hasattr(self._memory_instance, 'build_cache_optimized_context'):
-                try:
-                    # Get the goal/task description for context building
-                    task_desc = getattr(self, 'goal', '') or 'general assistance'
-                    cache_result = self._memory_instance.build_cache_optimized_context(
-                        task_descr=task_desc,
-                        include_cache_boundary=True
-                    )
-                    memory_context = cache_result.get('stable_prefix', '')
-                    cache_boundary = cache_result.get('cache_boundary', '')
-                    if memory_context:
-                        system_prompt += f"\n\n## Memory (Information you remember about the user)\n{memory_context}"
-                        if cache_boundary:
-                            system_prompt += cache_boundary
-                        # Display memory info to user if verbose
-                        if self.verbose:
-                            self._display_memory_info()
-                except (AttributeError, KeyError, TypeError) as e:
-                    # Fall back to standard memory context if cache optimization fails
-                    memory_context = self.get_memory_context()
-                    if memory_context:
-                        system_prompt += f"\n\n## Memory (Information you remember about the user)\n{memory_context}"
-                        # Display memory info to user if verbose
-                        if self.verbose:
-                            self._display_memory_info()
-            else:
-                # Standard memory context without cache boundaries
-                memory_context = self.get_memory_context()
-                if memory_context:
-                    system_prompt += f"\n\n## Memory (Information you remember about the user)\n{memory_context}"
-                    # Display memory info to user if verbose
-                    if self.verbose:
-                        self._display_memory_info()
+            memory_context = self.get_memory_context()
+            if memory_context:
+                system_prompt += f"\n\n## Memory (Information you remember about the user)\n{memory_context}"
+                # Display memory info to user if verbose
+                if self.verbose:
+                    self._display_memory_info()
            
            # Add learn context if learn is enabled (auto-inject when memory="learn")
            learn_context = self.get_learn_context()
@@ -112,60 +106,6 @@ Your Goal: {self.goal}"""
                system_prompt += f"\n\n## Available Skills\n{skills_prompt}"
                system_prompt += "\n\nWhen a skill is relevant to the task, read its SKILL.md file to get detailed instructions. If the skill has scripts in its scripts/ directory, you can execute them using the execute_code or run_script tool."
        
-        # Cache the base system prompt BEFORE adding session context
-        # Session context is per-turn and should not be cached
-        if cache_key:
-            self._cache_put(self._system_prompt_cache, cache_key, system_prompt)
-        
-        # Add session context (platform awareness) if available - AFTER caching
-        try:
-            from ..session.context import get_session_context
-            session_ctx = get_session_context()
-            
-            # Format session context into prompt if origin or targets are present
-            if session_ctx.origin or session_ctx.reachable_targets:
-                context_parts = []
-                
-                # Add origin information
-                if session_ctx.origin:
-                    origin = session_ctx.origin
-                    origin_str = f"You are replying on {origin.platform}"
-                    if origin.chat_type and origin.chat_type != "unknown":
-                        origin_str += f" ({origin.chat_type}"
-                        if origin.display_name:
-                            origin_str += f' "{origin.display_name}"'
-                        origin_str += ")"
-                    if origin.thread_id:
-                        origin_str += f" in thread {origin.thread_id}"
-                    context_parts.append(origin_str + ".")
-                
-                # Add reachable targets
-                if session_ctx.reachable_targets:
-                    target_descriptions = []
-                    for target in session_ctx.reachable_targets:
-                        desc = f"{target.name}"
-                        if target.kind == "home":
-                            desc += f" ({target.platform}, home channel)"
-                        elif target.kind == "alias":
-                            desc += f" ({target.platform}:{target.channel_id})"
-                        target_descriptions.append(desc)
-                    
-                    if target_descriptions:
-                        context_parts.append(
-                            f"Reachable delivery targets: {', '.join(target_descriptions)}."
-                        )
-                
-                if context_parts:
-                    system_prompt += "\n\n## Session Context\n" + "\n".join(context_parts)
-        except ImportError:
-            # Session context module not available, continue without it
-            pass
-        except Exception as e:
-            # Log unexpected errors but continue
-            import logging
-            logging.debug(f"Session context injection failed: {e}", exc_info=True)
-            pass
-        
        # Add tool usage instructions if tools are available
        # Use provided tools or fall back to self.tools
        tools_to_use = tools if tools is not None else self.tools
@@ -205,7 +145,10 @@ Your Goal: {self.goal}"""
        except ImportError:
            pass  # Trust module not available, skip security instructions
        
-        # Note: Caching is done BEFORE session context injection to avoid cross-user leakage
+        # Cache the generated system prompt (only if cache_key is set, i.e., memory not enabled)
+        # Use LRU eviction to prevent unbounded growth
+        if cache_key:
+            self._cache_put(self._system_prompt_cache, cache_key, system_prompt)
        return system_prompt

    def _build_response_format(self, schema_model):
@@ -509,17 +452,6 @@ Your Goal: {self.goal}"""
            else:
                cleaned_tools.append(tool)
        
-        # Sort tools deterministically for cache stability
-        def sort_formatted_tools(tools_list):
-            """Sort already-formatted tool schemas by function name."""
-            def sort_key(tool):
-                if isinstance(tool, dict) and tool.get('type') == 'function':
-                    return str(tool.get('function', {}).get('name') or '')
-                return ''
-            return sorted(tools_list, key=sort_key)
-        
-        cleaned_tools = sort_formatted_tools(cleaned_tools)
-        
        # Cache the formatted tools with LRU eviction, including tool search metadata
        self._cache_put(
            self._formatted_tools_cache,
@@ -614,6 +546,24 @@ Your Goal: {self.goal}"""
            # Fallback to str(response) is still fine, but now it's visible
        return str(response)

+    def _process_stream_response(self, messages, temperature, start_time, formatted_tools=None, reasoning_steps=False):
+        """Internal helper for streaming response processing with real-time events."""
+        if self._openai_client is None:
+            return None
+            
+        return self._openai_client.process_stream_response(
+            messages=messages,
+            model=self.llm,
+            temperature=temperature,
+            tools=formatted_tools,
+            start_time=start_time,
+            console=self.console,
+            display_fn=_get_display_functions()['display_generating'] if self.verbose else None,
+            reasoning_steps=reasoning_steps,
+            stream_callback=self.stream_emitter.emit,
+            emit_events=True
+        )
+
    def _get_compaction_policy(self):
        """
        Get the active compaction policy for this agent.
@@ -768,7 +718,7 @@ Your Goal: {self.goal}"""
        from ..context.policy import CompactionRoute
        return CompactionRoute.COMPACT_NEEDED, compacted_msgs

-    def _apply_tool_truncation(self, messages, compactor, policy, log_tag="tool-truncation"):
+    def _apply_tool_truncation(self, messages, compactor, policy):
        """Apply targeted tool output truncation."""
        from ..context.policy import CompactionRoute
        import logging
@@ -794,7 +744,7 @@ Your Goal: {self.goal}"""
        new_tokens = compactor.count_total_tokens(truncated_msgs)
        
        logging.info(
-            f"[{log_tag}] {self.name}: {original_tokens}→{new_tokens} tokens "
+            f"[tool-truncation] {self.name}: {original_tokens}→{new_tokens} tokens "
            f"(truncated large tool outputs)"
        )
        
@@ -888,25 +838,38 @@ Your Goal: {self.goal}"""
        return CompactionRoute.COMPACT_NEEDED, compacted_msgs

    async def _apply_tool_truncation_async(self, messages, compactor, policy):
-        """Async wrapper around _apply_tool_truncation (no awaitable work involved)."""
-        return self._apply_tool_truncation(messages, compactor, policy, log_tag="tool-truncation-async")
-
-    def _get_next_fallback_model(self, fallback_index):
-        """Get the next fallback model from the chain, if available.
+        """Async version of _apply_tool_truncation."""
+        from ..context.policy import CompactionRoute
+        import logging
        
-        Args:
-            fallback_index: Current index in the fallback chain (0-based)
+        # Create a copy to avoid modifying original
+        truncated_msgs = []
+        
+        for msg in messages:
+            if msg.get("role") == "tool" or msg.get("tool_call_id"):
+                content = msg.get("content", "")
+                if isinstance(content, str) and len(content) > 1000:
+                    # Truncate large tool outputs
+                    truncated_msg = msg.copy()
+                    head = content[:300]
+                    tail = content[-200:] if len(content) > 500 else ""
+                    truncated_msg["content"] = f"{head}\n...[truncated {len(content):,} chars for context budget]...\n{tail}"
+                    truncated_msgs.append(truncated_msg)
+                    continue
            
-        Returns:
-            str or None: Next model name if available, None otherwise
-        """
-        if not hasattr(self, 'fallback_models') or not self.fallback_models:
-            return None
-        if fallback_index >= len(self.fallback_models):
-            return None
-        return self.fallback_models[fallback_index]
+            truncated_msgs.append(msg)
+        
+        original_tokens = compactor.count_total_tokens(messages)
+        new_tokens = compactor.count_total_tokens(truncated_msgs)
+        
+        logging.info(
+            f"[tool-truncation-async] {self.name}: {original_tokens}→{new_tokens} tokens "
+            f"(truncated large tool outputs)"
+        )
+        
+        return CompactionRoute.TRUNCATE_TOOLS, truncated_msgs

-    def _chat_completion(self, messages, temperature=1.0, tools=None, stream=None, reasoning_steps=False, task_name=None, task_description=None, task_id=None, response_format=None, _retry_depth=0, _fallback_index=0):
+    def _chat_completion(self, messages, temperature=1.0, tools=None, stream=None, reasoning_steps=False, task_name=None, task_description=None, task_id=None, response_format=None, _retry_depth=0):
        start_time = time.time()

        # --- Proactive Context Budget Management (default-on) ---
@@ -948,29 +911,6 @@ Your Goal: {self.goal}"""
        # before_llm_input.messages; adopt that value for the actual LLM call.
        messages = before_llm_input.messages

-        # Pre-call budget guard (zero overhead when _max_budget is None).
-        # Estimate this call's minimum cost from the known input size plus the
-        # configured max_tokens output reservation and refuse to dispatch when
-        # it would breach the ceiling. This turns max_budget into a genuine hard
-        # cap instead of one that can be overshot by a whole LLM call.
-        if self._max_budget and self._on_budget_exceeded == "stop":
-            _est_min_cost = self._estimate_min_call_cost(
-                messages, getattr(self, 'max_tokens', None)
-            )
-            with self._cost_lock:
-                _projected_cost = self._total_cost + _est_min_cost
-                _current_cost = self._total_cost
-            if _projected_cost >= self._max_budget:
-                raise BudgetExceededError(
-                    f"Agent '{self.name}' would exceed budget before call: "
-                    f"${_current_cost:.4f} + est ${_est_min_cost:.4f} >= "
-                    f"${self._max_budget:.4f}",
-                    budget_type="cost",
-                    limit=self._max_budget,
-                    used=_current_cost,
-                    agent_id=self.name
-                )
-
        logging.debug(f"{self.name} sending messages to LLM: {messages}")
        
        # Emit LLM request trace event (zero overhead when not set)
@@ -993,7 +933,7 @@ Your Goal: {self.goal}"""
            try:
                # First attempt: try with streaming enabled for better user experience
                stream_callback = self.stream_emitter.emit if hasattr(self, 'stream_emitter') else None
-                final_response = self._chat_completion_with_retry(
+                final_response = self._execute_unified_chat_completion(
                    messages=messages,
                    temperature=temperature,
                    tools=formatted_tools,
@@ -1014,11 +954,7 @@ Your Goal: {self.goal}"""
                    stream = False  # Set for the main execution below
                else:
                    raise  # Re-raise if it's a different ValueError
-            except Exception as e:
-                from ..errors import LLMError
-                # Don't retry if it's an LLMError that has exhausted retries
-                if isinstance(e, LLMError):
-                    raise  # Re-raise LLMErrors immediately to avoid double retry
+            except Exception:
                # For any other exception, fall back to non-streaming
                logging.debug(f"{self.name}: Streaming attempt failed, falling back to non-streaming")
                stream = False  # Set for the main execution below
@@ -1029,7 +965,7 @@ Your Goal: {self.goal}"""
            # UNIFIED: Single protocol-driven dispatch path (fixes DRY violation)
            # All LLM providers now go through unified dispatcher for consistency and maintainability
            stream_callback = self.stream_emitter.emit if hasattr(self, 'stream_emitter') else None
-            final_response = self._chat_completion_with_retry(
+            final_response = self._execute_unified_chat_completion(
                messages=messages,
                temperature=temperature,
                tools=formatted_tools,
@@ -1176,8 +1112,7 @@ Your Goal: {self.goal}"""
                        return self._chat_completion(
                            truncated_messages, temperature, tools, stream, 
                            reasoning_steps, task_name, task_description, task_id, response_format, 
-                            _retry_depth=_retry_depth + 1,
-                            _fallback_index=_fallback_index
+                            _retry_depth=_retry_depth + 1
                        )
                except Exception as compression_error:
                    logging.error(f"[{self.name}] Context compression failed: {compression_error}")
@@ -1188,34 +1123,9 @@ Your Goal: {self.goal}"""
                # Don't retry without actual credential rotation
                
            elif classification.should_fallback_model:
-                # Try next model in the fallback chain
-                next_model = self._get_next_fallback_model(_fallback_index)
-                if next_model:
-                    current_model = self.llm if isinstance(self.llm, str) else str(self.llm)
-                    logging.info(f"[{self.name}] {current_model} unavailable — falling back to {next_model}")
-                    
-                    # Apply backoff if suggested
-                    if classification.backoff_seconds and classification.backoff_seconds > 0:
-                        time.sleep(classification.backoff_seconds)
-                    
-                    # Temporarily override the model and clear dispatcher cache for this call
-                    original_llm = self.llm
-                    original_dispatcher = getattr(self, '_unified_dispatcher', None)
-                    try:
-                        self.llm = next_model
-                        self._unified_dispatcher = None  # Force recreation with new model
-                        return self._chat_completion(
-                            messages, temperature, tools, stream,
-                            reasoning_steps, task_name, task_description, task_id, response_format,
-                            _retry_depth=_retry_depth + 1,
-                            _fallback_index=_fallback_index + 1
-                        )
-                    finally:
-                        self.llm = original_llm
-                        self._unified_dispatcher = original_dispatcher  # Restore original dispatcher
-                else:
-                    logging.warning(f"[{self.name}] {classification.user_message} (no more fallback models available)")
-                    # Continue to error handling without retry
+                # TODO: Implement model fallback when available
+                logging.warning(f"[{self.name}] {classification.user_message} (model fallback not yet implemented)")
+                # Don't retry without actual model fallback
                
            elif classification.is_retryable and classification.backoff_seconds > 0:
                if _retry_depth < 2:  # Limit retry attempts
@@ -1224,8 +1134,7 @@ Your Goal: {self.goal}"""
                    return self._chat_completion(
                        messages, temperature, tools, stream, 
                        reasoning_steps, task_name, task_description, task_id, response_format, 
-                        _retry_depth=_retry_depth + 1,
-                        _fallback_index=_fallback_index
+                        _retry_depth=_retry_depth + 1
                    )
            
            # Include remediation hints for unimplemented recovery actions
@@ -1233,10 +1142,7 @@ Your Goal: {self.goal}"""
            if classification.should_rotate_credential:
                user_message += " Credential rotation is not yet implemented."
            if classification.should_fallback_model:
-                if not self.fallback_models:
-                    user_message += " No fallback models configured."
-                elif _fallback_index >= len(self.fallback_models):
-                    user_message += f" All {len(self.fallback_models)} fallback models exhausted."
+                user_message += " Model fallback is not yet implemented."
            
            # Create LLMError with classification context
            error = LLMError(
@@ -1274,8 +1180,7 @@ Your Goal: {self.goal}"""
        response_format=None,
        stream_callback=None,
        emit_events=True,
-        _retry_depth=0,
-        _fallback_index=0
+        _retry_depth=0
    ):
        """
        Handle LLM errors in async context using structured classification.
@@ -1340,7 +1245,7 @@ Your Goal: {self.goal}"""
                        return await self._handle_async_llm_error(
                            retry_error, truncated_messages, temperature, tools, stream,
                            reasoning_steps, task_name, task_description, task_id, response_format,
-                            stream_callback, emit_events, _retry_depth + 1, _fallback_index
+                            stream_callback, emit_events, _retry_depth + 1
                        )
            except Exception as compression_error:
                logging.error(f"[{self.name}] Context compression failed: {compression_error}")
@@ -1351,39 +1256,9 @@ Your Goal: {self.goal}"""
            # Don't retry without actual credential rotation
            
        elif classification.should_fallback_model:
-            # Try next model in the fallback chain
-            next_model = self._get_next_fallback_model(_fallback_index)
-            if next_model:
-                current_model = self.llm if isinstance(self.llm, str) else str(self.llm)
-                logging.info(f"[{self.name}] {current_model} unavailable — falling back to {next_model}")
-                
-                # Apply backoff if suggested
-                if classification.backoff_seconds and classification.backoff_seconds > 0:
-                    await asyncio.sleep(classification.backoff_seconds)
-                
-                # Temporarily override the model and clear dispatcher cache for this call
-                original_llm = self.llm
-                original_dispatcher = getattr(self, '_unified_dispatcher', None)
-                try:
-                    self.llm = next_model
-                    self._unified_dispatcher = None  # Force recreation with new model
-                    return await self._execute_unified_achat_completion(
-                        messages, temperature, tools, stream,
-                        reasoning_steps, task_name, task_description, task_id, response_format,
-                        stream_callback, emit_events
-                    )
-                except Exception as retry_error:
-                    return await self._handle_async_llm_error(
-                        retry_error, messages, temperature, tools, stream,
-                        reasoning_steps, task_name, task_description, task_id, response_format,
-                        stream_callback, emit_events, _retry_depth + 1, _fallback_index + 1
-                    )
-                finally:
-                    self.llm = original_llm
-                    self._unified_dispatcher = original_dispatcher  # Restore original dispatcher
-            else:
-                logging.warning(f"[{self.name}] {classification.user_message} (no more fallback models available)")
-                # Continue to error handling without retry
+            # TODO: Implement model fallback when available
+            logging.warning(f"[{self.name}] {classification.user_message} (model fallback not yet implemented)")
+            # Don't retry without actual model fallback
            
        elif classification.is_retryable and classification.backoff_seconds > 0:
            if _retry_depth < 2:  # Limit retry attempts
@@ -1400,7 +1275,7 @@ Your Goal: {self.goal}"""
                    return await self._handle_async_llm_error(
                        retry_error, messages, temperature, tools, stream,
                        reasoning_steps, task_name, task_description, task_id, response_format,
-                        stream_callback, emit_events, _retry_depth + 1, _fallback_index
+                        stream_callback, emit_events, _retry_depth + 1
                    )
        
        # Include remediation hints for unimplemented recovery actions
@@ -1408,10 +1283,7 @@ Your Goal: {self.goal}"""
        if classification.should_rotate_credential:
            user_message += " Credential rotation is not yet implemented."
        if classification.should_fallback_model:
-            if not self.fallback_models:
-                user_message += " No fallback models configured."
-            elif _fallback_index >= len(self.fallback_models):
-                user_message += f" All {len(self.fallback_models)} fallback models exhausted."
+            user_message += " Model fallback is not yet implemented."
        
        # Create LLMError with classification context
        error = LLMError(
@@ -1455,7 +1327,7 @@ Your Goal: {self.goal}"""
        """
        Execute unified chat completion using composition instead of runtime class mutation.
        
-        This method provides a unified dispatch for all LLM providers using
+        This method provides the same functionality as UnifiedChatMixin but uses
        composition for safety and maintainability.
        """
        from ..llm import create_llm_dispatcher
@@ -1579,7 +1451,7 @@ Your Goal: {self.goal}"""
        """
        Execute unified async chat completion using composition instead of runtime class mutation.
        
-        This method provides a unified async dispatch for all LLM providers using
+        This method provides the same functionality as UnifiedChatMixin but uses
        composition for safety and maintainability.
        """
        from ..llm import create_llm_dispatcher
@@ -1649,6 +1521,151 @@ Your Goal: {self.goal}"""
                task_name, task_description, task_id, response_format, stream_callback, emit_events
            )

+    async def _finalize_unified_achat_response(
+        self,
+        response,
+        original_prompt,
+        temperature,
+        tools,
+        output_json,
+        output_pydantic,
+        reasoning_steps,
+        stream,
+        messages,
+        chat_history_length,
+        start_time,
+        task_name=None,
+        task_description=None,
+        task_id=None,
+    ):
+        """Post-process unified achat completion (parity with sync _chat_impl)."""
+        if not response:
+            self._truncate_chat_history(chat_history_length)
+            return None
+
+        if hasattr(response, "choices") and response.choices:
+            msg = response.choices[0].message
+            content = getattr(msg, "content", None)
+            response_text = content.strip() if content else ""
+            if (
+                reasoning_steps
+                and getattr(msg, "reasoning_content", None)
+            ):
+                response_text = msg.reasoning_content.strip()
+        else:
+            extracted = self._extract_llm_response_content(response)
+            response_text = (extracted or "").strip() if isinstance(extracted, str) else str(extracted or "")
+
+        if output_json or output_pydantic:
+            self._append_to_chat_history({"role": "assistant", "content": response_text})
+            self._persist_message("assistant", response_text)
+            try:
+                validated_response = self._apply_guardrail_with_retry(
+                    response_text, original_prompt, temperature, tools,
+                    task_name, task_description, task_id,
+                )
+                self._execute_callback_and_display(
+                    original_prompt, validated_response, time.time() - start_time,
+                    task_name, task_description, task_id,
+                )
+                return await self._atrigger_after_agent_hook(
+                    original_prompt, validated_response, start_time,
+                )
+            except Exception as e:
+                logging.error(f"Agent {self.name}: Guardrail validation failed for JSON output: {e}")
+                self._truncate_chat_history(chat_history_length)
+                return None
+
+        if not self.self_reflect:
+            self._append_to_chat_history({"role": "assistant", "content": response_text})
+            self._persist_message("assistant", response_text)
+            try:
+                validated_response = self._apply_guardrail_with_retry(
+                    response_text, original_prompt, temperature, tools,
+                    task_name, task_description, task_id,
+                )
+                self._execute_callback_and_display(
+                    original_prompt, validated_response, time.time() - start_time,
+                    task_name, task_description, task_id,
+                )
+                return await self._atrigger_after_agent_hook(
+                    original_prompt, validated_response, start_time,
+                )
+            except Exception as e:
+                logging.error(f"Agent {self.name}: Guardrail validation failed: {e}")
+                self._truncate_chat_history(chat_history_length)
+                return None
+
+        # Self-reflection (legacy OpenAI path)
+        reflection_count = 0
+        while True:
+            reflection_prompt = f"""
+Reflect on your previous response: '{response_text}'.
+{self.reflect_prompt if self.reflect_prompt else "Identify any flaws, improvements, or actions."}
+Provide a "satisfactory" status ('yes' or 'no').
+Output MUST be JSON with 'reflection' and 'satisfactory'.
+            """
+            reflection_messages = messages + [
+                {"role": "assistant", "content": response_text},
+                {"role": "user", "content": reflection_prompt},
+            ]
+            try:
+                if self._openai_client is None:
+                    self._append_to_chat_history({"role": "assistant", "content": response_text})
+                    return await self._atrigger_after_agent_hook(
+                        original_prompt, response_text, start_time,
+                    )
+                reflection_response = await self._openai_client.async_client.beta.chat.completions.parse(
+                    model=self.reflect_llm if self.reflect_llm else self.llm,
+                    messages=reflection_messages,
+                    temperature=temperature,
+                    response_format=_get_display_functions()['ReflectionOutput'],
+                )
+                reflection_output = reflection_response.choices[0].message.parsed
+                if reflection_output.satisfactory == "yes" and reflection_count >= self.min_reflect - 1:
+                    break
+                if reflection_count >= self.max_reflect - 1:
+                    break
+                regenerate_messages = reflection_messages + [
+                    {
+                        "role": "assistant",
+                        "content": (
+                            f"Self Reflection: {reflection_output.reflection} "
+                            f"Satisfactory?: {reflection_output.satisfactory}"
+                        ),
+                    },
+                    {"role": "user", "content": "Now regenerate your response using the reflection you made"},
+                ]
+                new_response = await self._openai_client.async_client.chat.completions.create(
+                    model=self.llm,
+                    messages=regenerate_messages,
+                    temperature=temperature,
+                )
+                response_text = new_response.choices[0].message.content
+                reflection_count += 1
+            except Exception as e:
+                logging.error("Reflection parsing failed.", exc_info=True)
+                reflection_count += 1
+                if reflection_count >= self.max_reflect:
+                    break
+
+        try:
+            validated_response = self._apply_guardrail_with_retry(
+                response_text, original_prompt, temperature, tools,
+                task_name, task_description, task_id,
+            )
+            self._execute_callback_and_display(
+                original_prompt, validated_response, time.time() - start_time,
+                task_name, task_description, task_id,
+            )
+            return await self._atrigger_after_agent_hook(
+                original_prompt, validated_response, start_time,
+            )
+        except Exception as e:
+            logging.error(f"Agent {self.name}: Guardrail validation failed for OpenAI client: {e}")
+            self._truncate_chat_history(chat_history_length)
+            return None
+
    def _execute_callback_and_display(self, prompt: str, response: str, generation_time: float, task_name=None, task_description=None, task_id=None):
        """Helper method to execute callbacks and display interaction.
        
@@ -1819,7 +1836,7 @@ Your Goal: {self.goal}"""
            logging.warning(f"Context management error (continuing without): {e}")
            return messages, None

-    def _truncate_tool_output(self, tool_name: str, output: str, tool_call_id: str | None = None) -> str:
+    def _truncate_tool_output(self, tool_name: str, output: str) -> str:
        """
        Truncate tool output according to configured budget.
        
@@ -1828,7 +1845,6 @@ Your Goal: {self.goal}"""
        Args:
            tool_name: Name of the tool
            output: Raw tool output
-            tool_call_id: Optional ID for the tool call
            
        Returns:
            Truncated output if over budget, otherwise original
@@ -1837,8 +1853,7 @@ Your Goal: {self.goal}"""
            return output
        
        try:
-            run_id = getattr(self, '_run_id', None)
-            return self.context_manager.truncate_tool_output(tool_name, output, tool_call_id, run_id)
+            return self.context_manager.truncate_tool_output(tool_name, output)
        except Exception as e:
            logging.warning(f"Tool output truncation error: {e}")
            return output
@@ -2481,9 +2496,8 @@ Output MUST be JSON with 'reflection' and 'satisfactory'.
                            # If not satisfactory and not at max reflections, continue with regeneration
                            logging.debug(f"{self.name} reflection count {reflection_count + 1}, continuing reflection process")
                            messages.append({"role": "user", "content": "Now regenerate your response using the reflection you made"})
-                            # For reflection, always use non-streaming to ensure compatibility with sync adapters
-                            # and to avoid streaming complexity during regeneration process
-                            use_stream = False
+                            # For custom LLMs during reflection, always use non-streaming to ensure complete responses
+                            use_stream = self.stream if not self._using_custom_llm else False
                            response = self._chat_completion(messages, temperature=temperature, tools=None, stream=use_stream, task_name=task_name, task_description=task_description, task_id=task_id)
                            content = response.choices[0].message.content
                            response_text = content.strip() if content else ""
@@ -2816,7 +2830,7 @@ Output MUST be JSON with 'reflection' and 'satisfactory'.
                            response_format = self._build_response_format(schema_model)
                        
                        # Use composition instead of runtime class mutation for safety
-                        response = await self._achat_completion_with_retry(
+                        response = await self._execute_unified_achat_completion(
                            messages=messages,
                            temperature=temperature,
                            tools=formatted_tools,
@@ -2988,7 +3002,7 @@ Output MUST be JSON with 'reflection' and 'satisfactory'.
                                        {"role": "user", "content": "Now regenerate your response using the reflection you made"}
                                    ]
                                    
-                                    new_response = await self._achat_completion_with_retry(
+                                    new_response = await self._execute_unified_achat_completion(
                                        messages=regenerate_messages,
                                        temperature=temperature,
                                        tools=formatted_tools,
@@ -3871,156 +3885,3 @@ Output MUST be JSON with 'reflection' and 'satisfactory'.
                raise
            logging.debug(f"[compaction] skipped (non-fatal): {_ce}")
            return False
-
-    def _chat_completion_with_retry(self, messages, temperature=1.0, tools=None, stream=None, reasoning_steps=False, task_name=None, task_description=None, task_id=None, response_format=None, stream_callback=None, emit_events=True):
-        """
-        Wrapper for _execute_unified_chat_completion that adds jittered exponential backoff retry logic.
-        
-        This method wraps the unified chat completion call and adds retry capability for 
-        transient failures like rate limits, network errors, and service outages.
-        """
-        retry_config = getattr(self, '_retry_config', None)
-        if not retry_config:
-            return self._execute_unified_chat_completion(messages, temperature, tools, stream, reasoning_steps, 
-                                       task_name, task_description, task_id, response_format,
-                                       stream_callback=stream_callback, emit_events=emit_events)
-        
-        from .retry_utils import jittered_backoff
-        from ..hooks import HookEvent, OnRetryInput
-        import time
-        
-        max_attempts = retry_config.max_retries + 1
-        
-        for attempt in range(max_attempts):
-            try:
-                # Call the underlying unified chat completion directly to avoid infinite recursion
-                return self._execute_unified_chat_completion(messages, temperature, tools, stream, reasoning_steps, 
-                                           task_name, task_description, task_id, response_format,
-                                           stream_callback=stream_callback, emit_events=emit_events)
-            
-            except Exception as e:
-                from ..errors import LLMError
-                
-                # Only retry LLMErrors that are marked as retryable
-                if not isinstance(e, LLMError) or not e.is_retryable:
-                    raise  # Re-raise non-retryable errors immediately
-                
-                # If this is the last attempt, re-raise the error
-                if attempt >= max_attempts - 1:
-                    raise
-                
-                # Calculate delay for this retry attempt
-                delay = jittered_backoff(
-                    attempt,
-                    base_delay=retry_config.base_delay,
-                    max_delay=retry_config.max_delay,
-                    jitter_ratio=retry_config.jitter_ratio,
-                )
-                
-                # Fire OnRetry hook with delay information
-                retry_input = OnRetryInput(
-                    session_id=getattr(self, '_session_id', 'default'),
-                    cwd=os.getcwd(),
-                    event_name=HookEvent.ON_RETRY,
-                    timestamp=str(time.time()),
-                    agent_name=self.name,
-                    retry_count=attempt + 1,
-                    max_retries=retry_config.max_retries,
-                    error_message=str(e),
-                    operation="llm_request",
-                    delay_seconds=delay,
-                    attempt=attempt
-                )
-                self._hook_runner.execute_sync(HookEvent.ON_RETRY, retry_input)
-                
-                # Log retry attempt (buffered to avoid spam during transient failures)
-                logger.debug(f"[{self.name}] Retry {attempt + 1}/{max_attempts} after {delay:.1f}s: {str(e)[:100]}")
-                
-                # Sleep with interrupt awareness - make interruption terminal
-                interrupt_fn = getattr(self, '_is_interrupted', lambda: False)
-                sleep_start = time.time()
-                while time.time() - sleep_start < delay:
-                    if interrupt_fn():
-                        # Interruption is terminal - stop retrying
-                        raise RuntimeError("Agent interrupted during retry backoff")
-                    time.sleep(min(0.2, delay - (time.time() - sleep_start)))
-        
-        # This should never be reached, but just in case
-        raise RuntimeError("Retry loop completed without returning or raising an exception")
-
-    async def _achat_completion_with_retry(self, messages, temperature=1.0, tools=None, stream=None, reasoning_steps=False, task_name=None, task_description=None, task_id=None, response_format=None, stream_callback=None, emit_events=True):
-        """
-        Async wrapper for _execute_unified_achat_completion that adds jittered exponential backoff retry logic.
-        
-        This method wraps the async chat completion call and adds retry capability for 
-        transient failures like rate limits, network errors, and service outages.
-        """
-        retry_config = getattr(self, '_retry_config', None)
-        if not retry_config:
-            return await self._execute_unified_achat_completion(
-                messages, temperature, tools, stream, reasoning_steps, 
-                task_name, task_description, task_id, response_format,
-                stream_callback=stream_callback, emit_events=emit_events
-            )
-        
-        from .retry_utils import jittered_backoff, interruptible_sleep
-        from ..hooks import HookEvent, OnRetryInput
-        import time
-        import asyncio
-        
-        max_attempts = retry_config.max_retries + 1
-        
-        for attempt in range(max_attempts):
-            try:
-                # Call the underlying unified chat completion directly to avoid infinite recursion
-                return await self._execute_unified_achat_completion(
-                    messages, temperature, tools, stream, reasoning_steps,
-                    task_name, task_description, task_id, response_format,
-                    stream_callback=stream_callback, emit_events=emit_events
-                )
-            
-            except Exception as e:
-                from ..errors import LLMError
-                
-                # Only retry LLMErrors that are marked as retryable
-                if not isinstance(e, LLMError) or not e.is_retryable:
-                    raise  # Re-raise non-retryable errors immediately
-                
-                # If this is the last attempt, re-raise the error
-                if attempt >= max_attempts - 1:
-                    raise
-                
-                # Calculate delay for this retry attempt
-                delay = jittered_backoff(
-                    attempt,
-                    base_delay=retry_config.base_delay,
-                    max_delay=retry_config.max_delay,
-                    jitter_ratio=retry_config.jitter_ratio,
-                )
-                
-                # Fire OnRetry hook with delay information
-                retry_input = OnRetryInput(
-                    session_id=getattr(self, '_session_id', 'default'),
-                    cwd=os.getcwd(),
-                    event_name=HookEvent.ON_RETRY,
-                    timestamp=str(time.time()),
-                    agent_name=self.name,
-                    retry_count=attempt + 1,
-                    max_retries=retry_config.max_retries,
-                    error_message=str(e),
-                    operation="async_llm_request",
-                    delay_seconds=delay,
-                    attempt=attempt
-                )
-                await self._hook_runner.execute_async(HookEvent.ON_RETRY, retry_input)
-                
-                # Log retry attempt
-                logger.debug(f"[{self.name}] Async retry {attempt + 1}/{max_attempts} after {delay:.1f}s: {str(e)[:100]}")
-                
-                # Async sleep with interrupt awareness using the helper
-                interrupt_fn = getattr(self, '_is_interrupted', lambda: False)
-                if not await interruptible_sleep(delay, interrupt_fn=interrupt_fn):
-                    raise RuntimeError("Agent interrupted during retry backoff")
-        
-        # This should never be reached, but just in case
-        raise RuntimeError("Async retry loop completed without returning or raising an exception")
@@ -252,28 +252,67 @@ Follow best practices and coding standards."""
    
    def _execute_python(self, code: str, **kwargs) -> Dict[str, Any]:
        """Execute Python code."""
+        import subprocess
        import time
-        from praisonaiagents.tools.python_tools import execute_code
-
+        import tempfile
+        import os
+        
        start_time = time.time()
-        mode = "sandbox" if self._code_config.sandbox else "direct"
-        result = execute_code(
-            code,
-            timeout=self._code_config.timeout,
-            max_output_size=self._code_config.max_output_length,
-            sandbox_mode=mode,
-        )
-        execution_time = time.time() - start_time
-
-        if result.get("success"):
+        
+        # Write code to temp file
+        with tempfile.NamedTemporaryFile(mode='w', suffix='.py', delete=False) as f:
+            f.write(code)
+            temp_file = f.name
+        
+        try:
+            # Execute in subprocess (basic sandboxing)
+            env = os.environ.copy()
+            env.update(self._code_config.environment)
+            
+            result = subprocess.run(
+                ["python", temp_file],
+                capture_output=True,
+                text=True,
+                timeout=self._code_config.timeout,
+                cwd=self._code_config.working_directory,
+                env=env
+            )
+            
+            execution_time = time.time() - start_time
+            
+            stdout = result.stdout[:self._code_config.max_output_length]
+            stderr = result.stderr[:self._code_config.max_output_length]
+            
            self._log(f"Execution completed in {execution_time:.2f}s", "green")
-
-        return {
-            "stdout": (result.get("stdout") or "")[: self._code_config.max_output_length],
-            "stderr": (result.get("stderr") or "")[: self._code_config.max_output_length],
-            "return_code": 0 if result.get("success") else 1,
-            "execution_time": execution_time,
-        }
+            
+            return {
+                "stdout": stdout,
+                "stderr": stderr,
+                "return_code": result.returncode,
+                "execution_time": execution_time
+            }
+            
+        except subprocess.TimeoutExpired:
+            return {
+                "stdout": "",
+                "stderr": f"Execution timed out after {self._code_config.timeout}s",
+                "return_code": -1,
+                "execution_time": self._code_config.timeout
+            }
+        except Exception as e:
+            return {
+                "stdout": "",
+                "stderr": str(e),
+                "return_code": -1,
+                "execution_time": time.time() - start_time
+            }
+        finally:
+            # Clean up temp file
+            try:
+                os.unlink(temp_file)
+            except Exception as e:
+                # Temp file cleanup failed - not critical
+                get_logger(__name__).debug(f"Failed to clean up temp file {temp_file}: {e}")
    
    async def aexecute(self, code: str, language: str = "python", **kwargs) -> Dict[str, Any]:
        """Execute code asynchronously.
@@ -15,8 +15,31 @@ import asyncio
 import threading


-# Shared lazy display helpers (cached, thread-safe; avoid circular imports)
-from ._lazy_display import _get_console, _get_live, _get_display_functions
+# Fallback helpers to avoid circular imports
+def _get_console():
+    from rich.console import Console
+    return Console
+
+def _get_live():
+    from rich.live import Live
+    return Live
+
+def _get_display_functions():
+    from ..main import (
+        display_error, display_instruction, display_interaction,
+        display_generating, display_self_reflection, ReflectionOutput,
+        adisplay_instruction, execute_sync_callback
+    )
+    return {
+        'display_error': display_error,
+        'display_instruction': display_instruction,
+        'display_interaction': display_interaction,
+        'display_generating': display_generating,
+        'display_self_reflection': display_self_reflection,
+        'ReflectionOutput': ReflectionOutput,
+        'adisplay_instruction': adisplay_instruction,
+        'execute_sync_callback': execute_sync_callback,
+    }

 logger = logging.getLogger(__name__)

@@ -35,6 +58,10 @@ class ExecutionMixin:
        """Synchronous sleep - safe for sync contexts only."""
        time.sleep(duration)

+    async def _safe_sleep_async(self, duration: float) -> None:
+        """Async sleep - use this in async contexts."""
+        await asyncio.sleep(duration)
+
    def generate_task(self) -> 'Task':
        """Generate a Task object from the agent's instructions"""
        from ..task.task import Task
@@ -914,113 +941,6 @@ Write the complete compiled report:"""
            self.verbose = original_verbose
            self.markdown = original_markdown

-    def learn_skill(self, request: str, **kwargs: Any) -> Union[str, Generator[str, None, None], None]:
-        """Author a grounded, reusable skill from sources you describe.
-
-        Point the agent at real source material — a directory of code, API docs,
-        PDFs/manuals, configs, pasted notes, or "what we just did in this chat" —
-        and it gathers them with the tools it already has (file read, fetch, PDF
-        read) and authors **one grounded SKILL.md** via the ``skill_manage`` tool.
-
-        House-style rules keep self-written skills trustworthy: only flags, paths,
-        and APIs that appear verbatim in the source are used (never invented), the
-        SKILL.md is kept tight (~100-200 lines), and long material is moved to
-        supporting files referenced by relative path. The result is a permanent,
-        reusable skill, immediately invocable like any other skill.
-
-        This is the **skill authoring** path and is intentionally distinct from
-        the memory ``learn=`` constructor param (conversation→memory extraction)
-        and the automatic ``self_improve`` loop. It is a prompt + entry point on
-        top of existing tools, not a new distillation engine: it ensures
-        ``skill_manage`` is available, then runs a single turn with :meth:`start`.
-
-        Args:
-            request: Natural-language description of the sources to learn from and
-                the skill to produce, e.g.
-                ``"Read ./my-repo and the docs in ./docs/*.pdf and make a
-                'deploy-flow' skill"``.
-            **kwargs: Additional arguments forwarded to :meth:`start`.
-
-        Returns:
-            The agent's response (the authored skill summary), as returned by
-            :meth:`start`.
-
-        Example::
-
-            agent = Agent()
-            agent.learn_skill("Read ./my-repo and make a 'deploy-flow' skill")
-        """
-        from praisonaiagents.skills import build_learn_prompt
-
-        self._ensure_skill_management_tools()
-        return self.start(build_learn_prompt(request), **kwargs)
-
-    async def alearn_skill(self, request: str, **kwargs: Any) -> Union[str, Generator[str, None, None], None]:
-        """Async variant of :meth:`learn_skill` (parity with ``start``/``astart``).
-
-        Ensures ``skill_manage`` is available, then awaits a single turn via
-        :meth:`astart` so async callers never block the event loop.
-
-        Args:
-            request: See :meth:`learn_skill`.
-            **kwargs: Additional arguments forwarded to :meth:`astart`.
-
-        Returns:
-            The agent's response (the authored skill summary).
-        """
-        from praisonaiagents.skills import build_learn_prompt
-
-        self._ensure_skill_management_tools()
-        return await self.astart(build_learn_prompt(request), **kwargs)
-
-    def learn(self, request: str, **kwargs: Any) -> Union[str, Generator[str, None, None], None]:
-        """Backward-compatible alias for :meth:`learn_skill`.
-
-        Retained so existing ``agent.learn("...")`` calls keep working. Prefer
-        :meth:`learn_skill` to avoid confusion with the memory ``learn=`` param.
-        """
-        return self.learn_skill(request, **kwargs)
-
-    async def alearn(self, request: str, **kwargs: Any) -> Union[str, Generator[str, None, None], None]:
-        """Backward-compatible async alias for :meth:`alearn_skill`."""
-        return await self.alearn_skill(request, **kwargs)
-
-    def _ensure_skill_management_tools(self) -> None:
-        """Ensure the agent has the ``skill_manage`` tool for authoring skills.
-
-        Adds the skill-management tool (and lightweight file readers) to
-        ``self.tools`` if they are not already present, so :meth:`learn` can
-        author a skill without the caller wiring tools manually. Idempotent.
-        """
-        if not hasattr(self, 'tools') or self.tools is None:
-            self.tools = []
-
-        def _tool_id(tool: Any) -> str:
-            # Mirror tool_execution.py: handle callables, dict/JSON specs, and
-            # anything else via a stable string fallback so dedup never lets a
-            # dict-form spec slip past the callable form (or vice versa).
-            name = getattr(tool, '__name__', None)
-            if isinstance(tool, dict):
-                name = name or tool.get('name') or (tool.get('function') or {}).get('name')
-            return name or str(tool)
-
-        existing_names = {_tool_id(t) for t in self.tools}
-
-        try:
-            from praisonaiagents.tools.skill_tools import (
-                skill_manage,
-                read_skill_file,
-                list_skill_scripts,
-            )
-        except Exception as exc:  # pragma: no cover - defensive
-            logger.debug(f"Could not import skill tools for learn_skill(): {exc}")
-            return
-
-        for tool in (skill_manage, read_skill_file, list_skill_scripts):
-            if _tool_id(tool) not in existing_names:
-                self.tools.append(tool)
-                existing_names.add(_tool_id(tool))
-
    def execute(self, task: Any, context: Optional[Any] = None) -> Optional[str]:
        """Execute a task synchronously - backward compatibility method"""
        if hasattr(task, 'description'):
@@ -405,59 +405,6 @@ class Handoff:
            # In intersect mode, empty list means no tools allowed - this is the security boundary
            return effective_tools
    
-    @staticmethod
-    def _extract_model_ref(agent: 'Agent') -> str:
-        """Extract a model reference string from an agent configuration."""
-        llm = getattr(agent, 'llm', None) or getattr(agent, 'model', None)
-        if isinstance(llm, str):
-            return llm
-        if llm is not None:
-            for attr in ('model', 'model_name', 'model_ref'):
-                value = getattr(llm, attr, None)
-                if isinstance(value, str) and value:
-                    return value
-        return 'gpt-4o-mini'
-
-    def _execute_with_runtime_resolution(
-        self, 
-        source_agent: 'Agent', 
-        prompt: str, 
-        effective_tools: Optional[List[Any]], 
-        context: Dict[str, Any]
-    ) -> str:
-        """Execute handoff through the target agent's full chat pipeline."""
-        target_model_ref = self._extract_model_ref(self.agent)
-        logger.debug(
-            "Handoff to %s via agent.chat (model=%s, depth=%s)",
-            getattr(self.agent, 'name', 'unknown'),
-            target_model_ref,
-            _get_handoff_depth(),
-        )
-        return self.agent.chat(prompt, tools=effective_tools)
-    
-    async def _execute_with_runtime_resolution_async(
-        self, 
-        source_agent: 'Agent', 
-        prompt: str, 
-        effective_tools: Optional[List[Any]], 
-        context: Dict[str, Any]
-    ) -> str:
-        """Execute handoff through the target agent's full async chat pipeline."""
-        target_model_ref = self._extract_model_ref(self.agent)
-        logger.debug(
-            "Async handoff to %s via agent chat (model=%s, depth=%s)",
-            getattr(self.agent, 'name', 'unknown'),
-            target_model_ref,
-            _get_handoff_depth(),
-        )
-        async_chat = getattr(self.agent, 'achat', None)
-        if callable(async_chat) and inspect.iscoroutinefunction(async_chat):
-            return await async_chat(prompt, tools=effective_tools)
-        loop = asyncio.get_running_loop()
-        return await loop.run_in_executor(
-            None, lambda: self.agent.chat(prompt, tools=effective_tools)
-        )
-    
    def _check_safety(self, source_agent: 'Agent') -> None:
        """
        Check safety constraints before handoff.
@@ -622,10 +569,8 @@ class Handoff:
            # Compute effective tools based on tool policy
            effective_tools = self._compute_effective_tools(source_agent)
            
-            # Resolve runtime at turn-time instead of using construction-time pin
-            response = self._execute_with_runtime_resolution(
-                source_agent, full_prompt, effective_tools, kwargs
-            )
+            # Execute with tool boundary enforcement
+            response = self.agent.chat(full_prompt, tools=effective_tools)
            
            result = HandoffResult(
                success=True,
@@ -730,10 +675,13 @@ class Handoff:
                # Compute effective tools based on tool policy
                effective_tools = self._compute_effective_tools(source_agent)
                
-                # Resolve runtime at turn-time for async execution
-                response = await self._execute_with_runtime_resolution_async(
-                    source_agent, full_prompt, effective_tools, kwargs
-                )
+                # Execute with tool boundary enforcement - check for async chat method
+                if hasattr(self.agent, 'achat'):
+                    response = await self.agent.achat(full_prompt, tools=effective_tools)
+                else:
+                    # Run sync chat in executor with tool constraint
+                    loop = asyncio.get_event_loop()
+                    response = await loop.run_in_executor(None, lambda: self.agent.chat(full_prompt, tools=effective_tools))
                
                result = HandoffResult(
                    success=True,
@@ -869,10 +817,7 @@ class Handoff:
                    # Compute effective tools based on tool policy
                    effective_tools = self._compute_effective_tools(source_agent)
                    
-                    # Resolve runtime at turn-time for tool function execution
-                    response = self._execute_with_runtime_resolution(
-                        source_agent, prompt, effective_tools, kwargs
-                    )
+                    response = self.agent.chat(prompt, tools=effective_tools)
                    
                    result = HandoffResult(
                        success=True,
@@ -10,8 +10,31 @@ import os
 import logging
 from typing import Optional

-# Shared lazy display helpers (cached, thread-safe; avoid circular imports)
-from ._lazy_display import _get_console, _get_live, _get_display_functions
+# Fallback helpers to avoid circular imports
+def _get_console():
+    from rich.console import Console
+    return Console
+
+def _get_live():
+    from rich.live import Live
+    return Live
+
+def _get_display_functions():
+    from ..main import (
+        display_error, display_instruction, display_interaction,
+        display_generating, display_self_reflection, ReflectionOutput,
+        adisplay_instruction, execute_sync_callback
+    )
+    return {
+        'display_error': display_error,
+        'display_instruction': display_instruction,
+        'display_interaction': display_interaction,
+        'display_generating': display_generating,
+        'display_self_reflection': display_self_reflection,
+        'ReflectionOutput': ReflectionOutput,
+        'adisplay_instruction': adisplay_instruction,
+        'execute_sync_callback': execute_sync_callback,
+    }

 logger = logging.getLogger(__name__)

@@ -294,9 +317,8 @@ class MemoryMixin:
            return
        
        try:
-            if self._session_store is None:
-                from ..session import get_default_session_store
-                self._session_store = get_default_session_store()
+            from ..session import get_default_session_store
+            self._session_store = get_default_session_store()
            
            # Restore chat history from previous session
            history = self._session_store.get_chat_history(self._session_id)
@@ -626,18 +648,10 @@ class MemoryMixin:
        
        try:
            import os
-            from praisonaiagents.tools.path_safety import resolve_within_root
-
-            project_root = os.environ.get("PRAISONAI_PROJECT_ROOT") or os.getcwd()
-            file_path = resolve_within_root(self._output_file, project_root)
-            if file_path is None:
-                logging.warning(
-                    "Output file %r is outside project root %r; skipping save",
-                    self._output_file,
-                    project_root,
-                )
-                print(f"⚠️ Output path outside project root: {self._output_file}")
-                return False
+            
+            # Expand user home directory and resolve path
+            file_path = os.path.expanduser(self._output_file)
+            file_path = os.path.abspath(file_path)
            
            # Create parent directories if they don't exist
            parent_dir = os.path.dirname(file_path)
@@ -1,101 +0,0 @@
-"""
-Retry utilities with jittered exponential backoff for the PraisonAI SDK.
-
-Provides interrupt-aware retry mechanisms for LLM API calls and tool execution
-with configurable backoff strategies, jitter, and buffered status reporting.
-"""
-
-import asyncio
-import random
-import time
-from dataclasses import dataclass
-from typing import Callable, Optional
-
-
-@dataclass
-class RetryBackoffConfig:
-    """Configuration for jittered exponential backoff retry behavior."""
-    base_delay: float = 5.0      # Base delay in seconds
-    max_delay: float = 120.0     # Maximum delay in seconds  
-    jitter_ratio: float = 0.5    # Jitter as fraction of delay added on top of base (0.0-1.0)
-    max_retries: int = 3         # Maximum retry attempts
-    
-    def __post_init__(self):
-        """Validate configuration parameters."""
-        if self.base_delay <= 0:
-            raise ValueError("base_delay must be > 0")
-        if self.max_delay < self.base_delay:
-            raise ValueError("max_delay must be >= base_delay")
-        if not (0 <= self.jitter_ratio <= 1):
-            raise ValueError("jitter_ratio must be between 0 and 1")
-        if self.max_retries < 0:
-            raise ValueError("max_retries must be >= 0")
-
-
-def jittered_backoff(
-    attempt: int,
-    *,
-    base_delay: float = 5.0,
-    max_delay: float = 120.0,
-    jitter_ratio: float = 0.5,
-) -> float:
-    """
-    Calculate delay for jittered exponential backoff.
-    
-    Args:
-        attempt: Current attempt number (0-based)
-        base_delay: Base delay in seconds
-        max_delay: Maximum delay cap in seconds
-        jitter_ratio: Jitter ratio (0.0-1.0), fraction of delay added as positive jitter
-    
-    Returns:
-        Delay in seconds with jitter applied
-    
-    Example:
-        >>> # Attempt 0: ~5s, Attempt 1: ~10s, Attempt 2: ~20s
-        >>> delay = jittered_backoff(1, base_delay=5.0, max_delay=120.0, jitter_ratio=0.5)
-    """
-    # Exponential backoff: base * 2^attempt
-    delay = min(base_delay * (2 ** max(0, attempt)), max_delay)
-    
-    # Apply jitter: delay + uniform(0, jitter_ratio * delay) for additive positive jitter
-    if jitter_ratio > 0:
-        jitter_range = delay * jitter_ratio
-        jitter = random.uniform(0, jitter_range)  # Positive additive jitter only
-        delay = max(0.1, min(delay + jitter, max_delay))  # Clamp again after jitter
-    
-    return delay
-
-
-async def interruptible_sleep(
-    seconds: float,
-    check_interval: float = 0.2,
-    interrupt_fn: Optional[Callable[[], bool]] = None,
-) -> bool:
-    """
-    Sleep with periodic interruption checks.
-    
-    Args:
-        seconds: Total sleep duration in seconds
-        check_interval: How often to check for interruption (seconds)
-        interrupt_fn: Function that returns True if sleep should be interrupted
-    
-    Returns:
-        True if completed full sleep, False if interrupted
-    
-    Example:
-        >>> interrupted = await interruptible_sleep(30.0, interrupt_fn=lambda: agent.is_stopped())
-    """
-    if interrupt_fn is None:
-        interrupt_fn = lambda: False
-    
-    elapsed = 0.0
-    while elapsed < seconds:
-        if interrupt_fn():
-            return False  # Interrupted
-        
-        sleep_time = min(check_interval, seconds - elapsed)
-        await asyncio.sleep(sleep_time)
-        elapsed += sleep_time
-    
-    return True  # Completed full sleep
@@ -5,10 +5,15 @@ Provides safe code execution capabilities to agents via the sandbox framework.
 """

 import logging
-from typing import Optional, Dict, Any, Union, TYPE_CHECKING
+from typing import Optional, Dict, Any, Union

-if TYPE_CHECKING:
-    from ..sandbox import SandboxConfig, SandboxManager, SandboxResult
+from ..sandbox import (
+    SandboxConfig, 
+    SandboxManager, 
+    SandboxResult,
+    check_code_safety,
+    format_warnings,
+)

 logger = logging.getLogger(__name__)

@@ -21,33 +26,27 @@ class SandboxMixin:
    
    def __init__(self, *args, **kwargs):
        # Extract sandbox config from kwargs
-        sandbox = kwargs.pop('sandbox', None)
+        self.sandbox_config: Optional[SandboxConfig] = kwargs.pop('sandbox', None)
        
        # Convert boolean True to default config
-        if sandbox is True:
-            from ..sandbox import SandboxConfig
-            sandbox = SandboxConfig.subprocess()
-        elif sandbox is False:
-            sandbox = None
-        
-        self.sandbox_config: Optional['SandboxConfig'] = sandbox
+        if self.sandbox_config is True:
+            self.sandbox_config = SandboxConfig.subprocess()
        
        super().__init__(*args, **kwargs)
        
-        self._sandbox_manager: Optional['SandboxManager'] = None
+        self._sandbox_manager: Optional[SandboxManager] = None
    
    @property
    def has_sandbox(self) -> bool:
        """Check if sandbox is configured for this agent."""
        return self.sandbox_config is not None
    
-    def get_sandbox_manager(self) -> Optional['SandboxManager']:
+    def get_sandbox_manager(self) -> Optional[SandboxManager]:
        """Get or create sandbox manager."""
        if not self.has_sandbox:
            return None
        
        if self._sandbox_manager is None:
-            from ..sandbox import SandboxManager
            self._sandbox_manager = SandboxManager(self.sandbox_config)
        
        return self._sandbox_manager
@@ -58,7 +57,7 @@ class SandboxMixin:
        language: str = "python",
        check_security: bool = True,
        **kwargs
-    ) -> 'SandboxResult':
+    ) -> SandboxResult:
        """Execute code safely in configured sandbox.
        
        Args:
@@ -80,7 +79,6 @@ class SandboxMixin:
        
        # Security pre-check if enabled
        if check_security:
-            from ..sandbox import check_code_safety, format_warnings
            warnings = check_code_safety(code, language)
            if warnings:
                warning_text = format_warnings(warnings)
@@ -99,7 +97,7 @@ class SandboxMixin:
        language: str = "python", 
        check_security: bool = True,
        **kwargs
-    ) -> 'SandboxResult':
+    ) -> SandboxResult:
        """Synchronous wrapper for execute_code.
        
        Args:
@@ -119,7 +117,7 @@ class SandboxMixin:
        command: Union[str, list],
        check_security: bool = True,
        **kwargs
-    ) -> 'SandboxResult':
+    ) -> SandboxResult:
        """Run a shell command in the sandbox.
        
        Args:
@@ -141,7 +139,6 @@ class SandboxMixin:
        
        # Security pre-check
        if check_security:
-            from ..sandbox import check_code_safety, format_warnings
            warnings = check_code_safety(command_str, "bash")
            if warnings:
                warning_text = format_warnings(warnings)
@@ -1,162 +0,0 @@
-"""Autonomous skill self-improvement loop.
-
-Closes the loop on the existing skill subsystem: after a task finishes, an
-opt-in *guarded* review pass runs a short, isolated turn restricted to the
-``skill_manage`` tool and asks the agent whether the session revealed a
-reusable technique worth persisting as a skill.
-
-Design guarantees (see issue #2231):
- Off by default — only runs when ``self_improve`` is enabled.
- Never runs with the full toolset — the review turn is restricted to the
-  single skill-mutation tool.
- Re-entrancy guarded — the review turn itself cannot trigger another review.
- Best-effort — any failure is swallowed so it never breaks the main task.
-
-This logic lives in its own mixin so ``agent.py`` stays lean.
-"""
-
-import logging
-
-logger = logging.getLogger(__name__)
-
-
-class SkillReviewMixin:
-    """Mixin providing the guarded skill-review pass for ``Agent``."""
-
-    def _skill_review_policy(self):
-        """Return the active review policy, building the default lazily."""
-        policy = getattr(self, "_self_improve_policy", None)
-        if policy is not None:
-            return policy
-        from ..skills import DefaultSkillReviewPolicy
-        policy = DefaultSkillReviewPolicy()
-        self._self_improve_policy = policy
-        return policy
-
-    def _build_skill_review_tool(self):
-        """Return the restricted toolset for the review turn (skill_manage only)."""
-        try:
-            from ..tools.skill_tools import create_skill_tools
-            skill_tools = create_skill_tools()
-            return [skill_tools.skill_manage]
-        except Exception as e:  # pragma: no cover - defensive
-            logger.debug("Could not build skill review tool: %s", e, exc_info=True)
-            return []
-
-    def _skill_review_log_context(self):
-        """Return (agent_name, session_id) for swallowed-failure warnings."""
-        return (
-            getattr(self, "name", None),
-            getattr(self, "_session_id", None),
-        )
-
-    def _prepare_skill_review(self, prompt, response, tools_used=None):
-        """Gate + build the review turn. Returns (review_prompt, review_tools) or None.
-
-        Shared by the sync and async paths so they stay in lock-step. Applies
-        the opt-in switch, the re-entrancy guard, the policy decision, and the
-        restricted-toolset build.
-        """
-        # Opt-in only.
-        if not getattr(self, "_self_improve", False):
-            return None
-        # Re-entrancy guard: never review a review.
-        if getattr(self, "_in_skill_review", False):
-            return None
-
-        trajectory = {
-            "prompt": prompt if isinstance(prompt, str) else str(prompt),
-            "response": response or "",
-            "tools_used": list(tools_used or []),
-        }
-
-        try:
-            policy = self._skill_review_policy()
-            if not policy.should_review(trajectory):
-                return None
-            review_prompt = policy.review_prompt(trajectory)
-        except Exception as e:
-            name, session_id = self._skill_review_log_context()
-            logger.warning(
-                "Skill review policy failed for agent=%s session_id=%s: %s",
-                name, session_id, e, exc_info=True,
-            )
-            return None
-
-        review_tools = self._build_skill_review_tool()
-        if not review_tools:
-            logger.debug("Skill review skipped: skill_manage tool unavailable")
-            return None
-
-        return review_prompt, review_tools
-
-    def _snapshot_chat_history_len(self):
-        """Return the current chat_history length, or None if unavailable."""
-        history = getattr(self, "chat_history", None)
-        return len(history) if isinstance(history, list) else None
-
-    def _restore_chat_history(self, snapshot_len):
-        """Trim chat_history back to ``snapshot_len`` so the review turn is isolated.
-
-        The review turn is internal: any user/assistant messages it appended to
-        the shared, persistent chat_history must not leak into subsequent turns
-        of a reused agent (chatbot, REPL, agentic loop).
-        """
-        if snapshot_len is None:
-            return
-        history = getattr(self, "chat_history", None)
-        if isinstance(history, list) and len(history) > snapshot_len:
-            del history[snapshot_len:]
-
-    def _run_skill_review(self, prompt, response, tools_used=None):
-        """Run the guarded skill-review pass.
-
-        Args:
-            prompt: The original user prompt for the finished task.
-            response: The final response produced for the task.
-            tools_used: Names of tools used during the task (if known).
-        """
-        prepared = self._prepare_skill_review(prompt, response, tools_used)
-        if prepared is None:
-            return
-        review_prompt, review_tools = prepared
-
-        self._in_skill_review = True
-        history_len = self._snapshot_chat_history_len()
-        try:
-            if getattr(self, "verbose", False):
-                logger.info("Agent %s running guarded skill-review pass", self.name)
-            # Restricted turn: only skill_manage is exposed via the tools= arg.
-            self.chat(review_prompt, tools=review_tools)
-        except Exception as e:
-            name, session_id = self._skill_review_log_context()
-            logger.warning(
-                "Skill review pass failed for agent=%s session_id=%s: %s",
-                name, session_id, e, exc_info=True,
-            )
-        finally:
-            self._restore_chat_history(history_len)
-            self._in_skill_review = False
-
-    async def _arun_skill_review(self, prompt, response, tools_used=None):
-        """Async variant of :meth:`_run_skill_review`."""
-        prepared = self._prepare_skill_review(prompt, response, tools_used)
-        if prepared is None:
-            return
-        review_prompt, review_tools = prepared
-
-        self._in_skill_review = True
-        history_len = self._snapshot_chat_history_len()
-        try:
-            if getattr(self, "verbose", False):
-                logger.info("Agent %s running guarded skill-review pass", self.name)
-            await self.achat(review_prompt, tools=review_tools)
-        except Exception as e:
-            name, session_id = self._skill_review_log_context()
-            logger.warning(
-                "Skill review pass failed for agent=%s session_id=%s: %s",
-                name, session_id, e, exc_info=True,
-            )
-        finally:
-            self._restore_chat_history(history_len)
-            self._in_skill_review = False
@@ -7,10 +7,21 @@ Reuses existing telemetry/token_collector for token tracking.

 import logging
 from praisonaiagents._logging import get_logger
-from typing import Optional, List, Dict, Any
+from typing import Optional, List, Dict, Any, Callable
+from dataclasses import dataclass, field

 logger = get_logger(__name__)

+@dataclass
+class SummarizationConfig:
+    """Configuration for auto-summarization."""
+    enabled: bool = False
+    threshold: float = 0.8  # 80% of context window
+    context_window: int = 128000  # Default for GPT-4
+    summary_model: Optional[str] = None  # Use smaller model for summaries
+    preserve_system: bool = True  # Keep system message
+    preserve_recent: int = 2  # Keep last N message pairs
+
 class SummarizationManager:
    """
    Manages automatic conversation summarization.
@@ -168,3 +179,45 @@ SUMMARY:"""
        """Number of times summarization has been performed."""
        return self._summarization_count

+async def summarize_conversation(
+    messages: List[Dict[str, str]],
+    llm_call: Callable,
+    manager: SummarizationManager,
+) -> List[Dict[str, str]]:
+    """
+    Summarize conversation using LLM.
+    
+    Args:
+        messages: Current message history
+        llm_call: Async function to call LLM
+        manager: SummarizationManager instance
+        
+    Returns:
+        New message history with summary
+    """
+    prompt = manager.generate_summary_prompt(messages)
+    
+    # Call LLM to generate summary
+    summary = await llm_call(prompt)
+    
+    # Prepare new history
+    new_history = manager.prepare_summarized_history(messages, summary)
+    
+    # Reset token count (will be recalculated on next message)
+    manager.reset_tokens()
+    
+    return new_history
+
+def summarize_conversation_sync(
+    messages: List[Dict[str, str]],
+    llm_call: Callable,
+    manager: SummarizationManager,
+) -> List[Dict[str, str]]:
+    """
+    Synchronous version of summarize_conversation.
+    """
+    prompt = manager.generate_summary_prompt(messages)
+    summary = llm_call(prompt)
+    new_history = manager.prepare_summarized_history(messages, summary)
+    manager.reset_tokens()
+    return new_history
@@ -1,300 +0,0 @@
-"""
-Integration tests for handoff runtime resolution.
-
-This test suite validates that handoffs now use turn-time runtime resolution
-instead of construction-time pins, addressing issue #1938.
-"""
-
-import pytest
-import time
-from unittest.mock import Mock, patch, MagicMock, AsyncMock
-import asyncio
-
-from .handoff import Handoff, HandoffConfig, ContextPolicy, HandoffResult
-
-
-class MockAgent:
-    """Mock agent for testing handoff scenarios."""
-    
-    def __init__(self, name, model="gpt-4o", agent_id=None, session_id="test"):
-        self.name = name
-        self.llm = model
-        self.model = model  
-        self.agent_id = agent_id or name
-        self._session_id = session_id
-        self.chat_history = []
-    
-    def chat(self, prompt, **kwargs):
-        """Mock chat method that returns a simple response."""
-        response = f"Response from {self.name} using {self.model}: {prompt[:50]}..."
-        self.chat_history.append({"role": "user", "content": prompt})
-        self.chat_history.append({"role": "assistant", "content": response})
-        return response
-    
-    async def achat(self, prompt, **kwargs):
-        """Mock async chat method."""
-        return self.chat(prompt, **kwargs)
-
-
-class TestHandoffRuntimeResolution:
-    """Test handoff runtime resolution functionality."""
-    
-    def test_handoff_uses_turn_time_resolution(self):
-        """Test that handoff uses turn-time runtime resolution instead of construction-time."""
-        # Create agents with different models
-        source_agent = MockAgent("SourceAgent", "gpt-3.5-turbo")
-        target_agent = MockAgent("TargetAgent", "gpt-4o")
-        
-        # Create handoff
-        handoff = Handoff(
-            agent=target_agent,
-            config=HandoffConfig(context_policy=ContextPolicy.NONE)
-        )
-        
-        # Mock the runtime resolution to verify it's called
-        with patch.object(handoff, '_execute_with_runtime_resolution') as mock_execute:
-            mock_execute.return_value = "Resolved response"
-            
-            # Execute programmatic handoff
-            result = handoff.execute_programmatic(
-                source_agent=source_agent,
-                prompt="Test handoff",
-                context={}
-            )
-            
-            # Verify runtime resolution was called instead of direct agent.chat
-            mock_execute.assert_called_once()
-            args = mock_execute.call_args[0]
-            assert args[0] == source_agent  # source_agent
-            assert "Test handoff" in args[1]  # full_prompt with context
-            assert args[3] == {}  # context kwargs
-    
-    @pytest.mark.asyncio
-    async def test_async_handoff_uses_turn_time_resolution(self):
-        """Test that async handoff uses turn-time runtime resolution."""
-        source_agent = MockAgent("SourceAgent", "gpt-3.5-turbo") 
-        target_agent = MockAgent("TargetAgent", "claude-3-sonnet")
-        
-        handoff = Handoff(
-            agent=target_agent,
-            config=HandoffConfig(context_policy=ContextPolicy.NONE)
-        )
-        
-        # Mock the async runtime resolution
-        with patch.object(handoff, '_execute_with_runtime_resolution_async') as mock_execute_async:
-            mock_execute_async.return_value = "Async resolved response"
-            
-            # Execute async handoff
-            result = await handoff.execute_async(
-                source_agent=source_agent,
-                prompt="Test async handoff", 
-                context={}
-            )
-            
-            # Verify async runtime resolution was called
-            mock_execute_async.assert_called_once()
-            args = mock_execute_async.call_args[0]
-            assert args[0] == source_agent
-            assert "Test async handoff" in args[1]
-    
-    def test_handoff_tool_function_uses_runtime_resolution(self):
-        """Test that handoff tool function uses runtime resolution."""
-        source_agent = MockAgent("SourceAgent", "gpt-4")
-        source_agent.chat_history = [{"role": "user", "content": "Complete this task"}]
-        target_agent = MockAgent("TargetAgent", "gpt-4o")
-        
-        handoff = Handoff(
-            agent=target_agent,
-            config=HandoffConfig(context_policy=ContextPolicy.SUMMARY)
-        )
-        
-        # Get tool function
-        tool_func = handoff.to_tool_function(source_agent)
-        
-        # Mock the runtime resolution
-        with patch.object(handoff, '_execute_with_runtime_resolution') as mock_execute:
-            mock_execute.return_value = "Tool resolved response"
-            
-            # Execute tool function
-            result = tool_func(task="Complete this task")
-            
-            # Verify runtime resolution was called
-            mock_execute.assert_called_once()
-            args = mock_execute.call_args[0]
-            assert args[0] == source_agent
-    
-    @patch.object(MockAgent, 'chat')
-    def test_runtime_resolution_uses_agent_chat(self, mock_chat):
-        """Handoff must execute via agent.chat, not a bare LLM runtime."""
-        mock_chat.return_value = "Agent chat response"
-        source_agent = MockAgent("SourceAgent", "gpt-3.5-turbo")
-        target_agent = MockAgent("TargetAgent", "gpt-4o")
-        
-        handoff = Handoff(agent=target_agent)
-        
-        response = handoff._execute_with_runtime_resolution(
-            source_agent=source_agent,
-            prompt="Test prompt",
-            effective_tools=[],
-            context={}
-        )
-        
-        mock_chat.assert_called_once_with("Test prompt", tools=[])
-        assert response == "Agent chat response"
-    
-    @pytest.mark.asyncio
-    async def test_async_runtime_resolution_uses_agent_achat(self):
-        """Async handoff must execute via agent.achat, not a bare LLM runtime."""
-        source_agent = MockAgent("SourceAgent", "gpt-4")
-        target_agent = MockAgent("TargetAgent", "claude-3-sonnet")
-        
-        with patch.object(target_agent, 'achat', new_callable=AsyncMock) as mock_achat:
-            mock_achat.return_value = "Async agent chat response"
-            handoff = Handoff(agent=target_agent)
-            
-            response = await handoff._execute_with_runtime_resolution_async(
-                source_agent=source_agent,
-                prompt="Test async prompt",
-                effective_tools=[],
-                context={}
-            )
-            
-            mock_achat.assert_called_once_with("Test async prompt", tools=[])
-            assert response == "Async agent chat response"
-    
-    def test_runtime_resolution_uses_agent_chat_even_when_resolve_available(self):
-        """Regression: bare resolve_runtime must not bypass agent instructions."""
-        source_agent = MockAgent("SourceAgent", "gpt-4")
-        target_agent = MockAgent("TargetAgent", "gpt-4o")
-        
-        handoff = Handoff(agent=target_agent)
-        
-        with patch('praisonaiagents.runtime.resolve.resolve_runtime') as mock_resolve:
-            mock_runtime = Mock()
-            mock_runtime.execute.return_value = "Bare runtime response"
-            mock_resolve.return_value = mock_runtime
-            
-            with patch.object(target_agent, 'chat', wraps=target_agent.chat) as mock_chat:
-                response = handoff._execute_with_runtime_resolution(
-                    source_agent=source_agent,
-                    prompt="Test fallback",
-                    effective_tools=[],
-                    context={}
-                )
-                
-                mock_resolve.assert_not_called()
-                mock_runtime.execute.assert_not_called()
-                mock_chat.assert_called_once()
-                assert "Response from TargetAgent using gpt-4o" in response
-    
-    @pytest.mark.asyncio
-    async def test_async_runtime_resolution_uses_agent_chat(self):
-        """Async handoff should route through agent chat, not bare runtime."""
-        source_agent = MockAgent("SourceAgent", "gpt-4")
-        target_agent = MockAgent("TargetAgent", "claude-3-haiku")
-        
-        handoff = Handoff(agent=target_agent)
-        
-        with patch('praisonaiagents.runtime.resolve.resolve_runtime') as mock_resolve:
-            mock_resolve.side_effect = Exception("Should not be called")
-            
-            response = await handoff._execute_with_runtime_resolution_async(
-                source_agent=source_agent,
-                prompt="Test async fallback",
-                effective_tools=[], 
-                context={}
-            )
-            
-            mock_resolve.assert_not_called()
-            assert "Response from TargetAgent using claude-3-haiku" in response
-    
-    def test_extract_model_ref_from_llm_object(self):
-        """Model extraction should handle LLM objects, not pass them to resolvers."""
-        source_agent = MockAgent("SourceAgent", "gpt-4")
-        target_agent = MockAgent("TargetAgent", "gpt-4o")
-        llm_object = Mock()
-        llm_object.model = "claude-3-opus"
-        target_agent.llm = llm_object
-        
-        handoff = Handoff(agent=target_agent)
-        assert handoff._extract_model_ref(target_agent) == "claude-3-opus"
-
-
-class TestConstructionVsTurnTimeRegression:
-    """
-    Regression tests for construction-time vs turn-time resolution.
-    
-    These tests specifically validate that the fix for issue #1938 works correctly.
-    """
-    
-    def test_model_change_after_handoff_creation(self):
-        """
-        Test the core regression case: model change after handoff creation.
-        
-        Before fix: handoff would use construction-time model
-        After fix: handoff should use turn-time model
-        """
-        source_agent = MockAgent("SourceAgent", "gpt-3.5-turbo")
-        target_agent = MockAgent("TargetAgent", "gpt-4")
-        
-        # Create handoff with initial target model
-        handoff = Handoff(agent=target_agent)
-        
-        # Target agent changes model (router decision, user preference, etc.)
-        target_agent.llm = "gpt-4o"
-        target_agent.model = "gpt-4o"
-        
-        with patch.object(target_agent, 'chat', wraps=target_agent.chat) as mock_chat:
-            handoff._execute_with_runtime_resolution(
-                source_agent=source_agent,
-                prompt="Test with new model",
-                effective_tools=[],
-                context={}
-            )
-            
-            mock_chat.assert_called_once()
-            assert handoff._extract_model_ref(target_agent) == "gpt-4o"
-    
-    def test_sub_agent_tool_respects_current_model(self):
-        """Test that as_tool() handoffs route through the target agent chat path."""
-        parent_agent = MockAgent("ParentAgent", "gpt-4")
-        parent_agent.chat_history = [{"role": "user", "content": "Complete subtask"}]
-        sub_agent = MockAgent("SubAgent", "claude-3-sonnet")
-        
-        handoff = Handoff(agent=sub_agent)
-        sub_agent.llm = "claude-3-opus"
-        sub_agent.model = "claude-3-opus"
-        
-        with patch.object(sub_agent, 'chat', wraps=sub_agent.chat) as mock_chat:
-            tool_func = handoff.to_tool_function(parent_agent)
-            result = tool_func(task="Complete subtask")
-            
-            mock_chat.assert_called_once()
-            assert "claude-3-opus" in result
-            assert handoff._extract_model_ref(sub_agent) == "claude-3-opus"
-            
-    def test_multiple_handoffs_different_models(self):
-        """Test multiple handoffs with different models work correctly."""
-        source_agent = MockAgent("SourceAgent", "gpt-3.5-turbo")
-        target_a = MockAgent("TargetA", "gpt-4o") 
-        target_b = MockAgent("TargetB", "claude-3-sonnet")
-        
-        handoff_a = Handoff(agent=target_a)
-        handoff_b = Handoff(agent=target_b)
-        
-        response_a = handoff_a._execute_with_runtime_resolution(
-            source_agent, "Task A", [], {}
-        )
-        response_b = handoff_b._execute_with_runtime_resolution(
-            source_agent, "Task B", [], {}
-        )
-        
-        assert "gpt-4o" in response_a
-        assert "claude-3-sonnet" in response_b
-        assert handoff_a._extract_model_ref(target_a) == "gpt-4o"
-        assert handoff_b._extract_model_ref(target_b) == "claude-3-sonnet"
-
-
-# Run tests 
-if __name__ == "__main__":
-    pytest.main([__file__, "-v"])
@@ -14,7 +14,6 @@ import asyncio
 import inspect
 import contextvars
 import concurrent.futures
-import random
 from typing import List, Optional, Any, Dict, Union, TYPE_CHECKING
 from ..errors import ToolExecutionError
 from ..tools.trust import wrap_if_external
@@ -25,61 +24,8 @@ if TYPE_CHECKING:
    pass


-class BackoffPolicy:
-    """Exponential backoff policy for tool retries."""
-    
-    @staticmethod
-    def delay(attempt: int, initial_delay: float, backoff_factor: float, jitter: float, max_delay: float = 60.0) -> float:
-        """Calculate delay for a retry attempt.
-        
-        Args:
-            attempt: Attempt number (1-based)
-            initial_delay: Initial delay in seconds
-            backoff_factor: Exponential backoff multiplier
-            jitter: Fraction of base delay to add as random jitter
-            max_delay: Maximum delay to cap exponential growth
-            
-        Returns:
-            Delay in seconds
-        """
-        base = initial_delay * (backoff_factor ** (attempt - 1))
-        # Cap the base delay to prevent excessively long waits
-        base = min(base, max_delay)
-        jitter_amount = random.uniform(0, jitter * base)
-        return base + jitter_amount
-
-
 class ToolExecutionMixin:
    """Mixin providing toolexecution methods for the Agent class."""
-    
-    def _register_artifact_tools(self):
-        """Register artifact retrieval tools when artifacts are first created."""
-        try:
-            from ..tools import artifact_tools
-            
-            # Set the store reference for the tools
-            artifact_tools.set_artifact_store(self._artifact_store)
-            
-            # Add the retrieval tools
-            tools_to_add = [
-                artifact_tools.artifact_head,
-                artifact_tools.artifact_tail,
-                artifact_tools.artifact_grep,
-                artifact_tools.artifact_chunk,
-                artifact_tools.artifact_load,
-                artifact_tools.artifact_list,
-            ]
-            
-            # Only add if not already present
-            existing_tool_names = {getattr(t, '__name__', str(t)) for t in self.tools}
-            for tool in tools_to_add:
-                tool_name = getattr(tool, '__name__', str(tool))
-                if tool_name not in existing_tool_names:
-                    self.tools.append(tool)
-            
-            logging.debug("Registered artifact retrieval tools")
-        except Exception as e:
-            logging.warning(f"Failed to register artifact tools: {e}")

    def _get_existing_stream_emitter(self):
        """Return an already-initialized stream emitter without creating one."""
@@ -285,10 +231,9 @@ class ToolExecutionMixin:
                    arguments.update(res.output.modified_data)

            # Loop guard check - prevent tool execution loops with graduated response
-            if hasattr(self, '_ensure_loop_guard'):
-                loop_guard = self._ensure_loop_guard()
+            if hasattr(self, '_loop_guard') and self._loop_guard:
                from ..escalation.loop_guard import GuardAction
-                decision = loop_guard.check(function_name, arguments, is_pre_execution=True)
+                decision = self._loop_guard.check(function_name, arguments, is_pre_execution=True)
                
                if decision.action == GuardAction.WARN:
                    # Inject warning into tool result so LLM sees guidance
@@ -300,6 +245,7 @@ class ToolExecutionMixin:
                    blocked_result = {"error": f"[loop-guard] {decision.message}", "loop_blocked": True}
                elif decision.action == GuardAction.HALT:
                    # Halt execution with exception
+                    from ..errors import ToolExecutionError
                    raise ToolExecutionError(
                        f"[loop-guard] {decision.message}",
                        tool_name=function_name,
@@ -327,177 +273,32 @@ class ToolExecutionMixin:
            if blocked_result is not None:
                result = blocked_result
            else:
-                # Apply tool retry logic with exponential backoff
-                execution_config = getattr(self, '_execution_config', None)
-                if execution_config is None:
-                    # Fall back to reading individual config attributes for backward compatibility
-                    max_retry_limit = getattr(self, 'max_retry_limit', 2)
-                    retry_initial_delay = 1.0
-                    retry_backoff_factor = 2.0
-                    retry_jitter = 0.1
-                else:
-                    max_retry_limit = execution_config.max_retry_limit
-                    retry_initial_delay = execution_config.retry_initial_delay
-                    retry_backoff_factor = execution_config.retry_backoff_factor
-                    retry_jitter = execution_config.retry_jitter
-                
-                result = None
-                last_exception = None
-                
-                # max_retry_limit is the number of retries (not total attempts)
-                # So total attempts = 1 (initial) + max_retry_limit (retries)
-                for attempt in range(1, max_retry_limit + 2):
-                    try:
-                        # P8/G11: Apply tool timeout if configured
-                        tool_timeout = getattr(self, '_tool_timeout', None)
-                        if tool_timeout and tool_timeout > 0:
-                            # Use copy_context to preserve injection context in executor thread
-                            ctx = contextvars.copy_context()
-                            
-                            def execute_with_context():
-                                with with_injection_context(state):
-                                    return self._execute_tool_with_circuit_breaker(function_name, arguments)
-                            
-                            # Use reusable executor to prevent resource leaks
-                            if not hasattr(self, '_tool_executor'):
-                                self._tool_executor = concurrent.futures.ThreadPoolExecutor(
-                                    max_workers=2, thread_name_prefix=f"tool-{self.name}"
-                                )
-                            
-                            future = self._tool_executor.submit(ctx.run, execute_with_context)
-                            try:
-                                result = future.result(timeout=tool_timeout)
-                            except concurrent.futures.TimeoutError:
-                                future.cancel()
-                                logging.warning(f"Tool {function_name} timed out after {tool_timeout}s")
-                                result = {"error": f"Tool timed out after {tool_timeout}s", "timeout": True}
-                        else:
-                            with with_injection_context(state):
-                                result = self._execute_tool_with_circuit_breaker(function_name, arguments)
-                        
-                        # Check if the result indicates a retryable error
-                        if isinstance(result, dict) and result.get("error"):
-                            # Check if this is a circuit breaker error (always retryable)
-                            if result.get("circuit_open"):
-                                raise ToolExecutionError(
-                                    result["error"],
-                                    tool_name=function_name,
-                                    agent_id=self.name,
-                                    is_retryable=True,
-                                )
-                            # Check if this is a timeout error (retryable)
-                            elif result.get("timeout"):
-                                raise ToolExecutionError(
-                                    result["error"],
-                                    tool_name=function_name,
-                                    agent_id=self.name,
-                                    is_retryable=True,
-                                )
-                            # For other error dicts: approval/permission denials are legitimate
-                            # non-retryable outcomes; everything else represents a tool failure
-                            # that should engage the outer retry/backoff loop.
-                            elif result.get("approval_denied") or result.get("permission_denied") or result.get("approval_error"):
-                                break
-                            else:
-                                # Avoid compounding with the inner retry loop in
-                                # _execute_tool_with_circuit_breaker: error types it already
-                                # retries (e.g. timeout/rate_limit/connection_error) are
-                                # exhausted by the time they reach here, so escalating them as
-                                # retryable would re-run the entire inner loop from scratch on
-                                # every outer attempt. Only escalate error types the inner loop
-                                # does NOT retry (e.g. "unknown") so they get one outer pass.
-                                error_type = self._classify_error_type(result, None)
-                                inner_policy = self._get_tool_retry_policy(function_name)
-                                is_retryable = error_type not in inner_policy.retry_on
-                                raise ToolExecutionError(
-                                    result.get("error", f"Tool '{function_name}' failed"),
-                                    tool_name=function_name,
-                                    agent_id=self.name,
-                                    is_retryable=is_retryable,
-                                )
-                        else:
-                            # Success path - return the result
-                            break
-                            
-                    except ToolExecutionError as e:
-                        last_exception = e
-                        # Only retry if the error is marked as retryable and we have retries left
-                        # attempt starts at 1, so (attempt - 1) gives us the retry count
-                        if not e.is_retryable or (attempt - 1) >= max_retry_limit:
-                            raise e
-                        
-                        # Calculate delay for exponential backoff
-                        delay = BackoffPolicy.delay(attempt, retry_initial_delay, retry_backoff_factor, retry_jitter)
-                        logging.warning(
-                            f"Tool {function_name} failed (attempt {attempt}/{max_retry_limit + 1}): {e}. "
-                            f"Retrying in {delay:.2f}s..."
-                        )
-                        time.sleep(delay)
-                        
-                    except Exception as e:
-                        # Wrap unexpected exceptions in ToolExecutionError
-                        # Most tool errors are considered retryable unless they're programming errors
-                        is_retryable = not isinstance(e, (ValueError, TypeError, AttributeError))
-                        tool_error = ToolExecutionError(
-                            f"Tool '{function_name}' failed: {e}",
-                            tool_name=function_name,
-                            agent_id=self.name,
-                            is_retryable=is_retryable,
-                        )
-                        last_exception = tool_error
-                        
-                        # attempt starts at 1, so (attempt - 1) gives us the retry count
-                        if not is_retryable or (attempt - 1) >= max_retry_limit:
-                            raise tool_error from e
-                        
-                        # Calculate delay for exponential backoff
-                        delay = BackoffPolicy.delay(attempt, retry_initial_delay, retry_backoff_factor, retry_jitter)
-                        logging.warning(
-                            f"Tool {function_name} failed (attempt {attempt}/{max_retry_limit + 1}): {e}. "
-                            f"Retrying in {delay:.2f}s..."
-                        )
-                        time.sleep(delay)
-            
-            # Apply runtime-scoped middleware normalization BEFORE hooks fire
-            # Plugin harnesses can register middleware to normalize vendor-specific results
-            runtime_id = getattr(self, '_runtime_id', 'praisonai')  # Default to native runtime
-            normalized_result = None  # Track normalized result for hooks
-            if runtime_id != 'praisonai':  # Skip for native runtime to avoid allocation
-                try:
-                    from ..runtime import get_middleware, MiddlewareContext
-                    middleware = get_middleware(runtime_id)
+                # P8/G11: Apply tool timeout if configured
+                tool_timeout = getattr(self, '_tool_timeout', None)
+                if tool_timeout and tool_timeout > 0:
+                    # Use copy_context to preserve injection context in executor thread
+                    ctx = contextvars.copy_context()
                    
-                    # Only allocate context and normalize if not using default pass-through
-                    if middleware.runtime_id != 'praisonai':
-                        mw_ctx = MiddlewareContext(
-                            tool_name=function_name,
-                            runtime_id=runtime_id,
-                            agent_id=self.name,
-                            session_id=getattr(self, '_session_id', None),
-                            execution_time_ms=(_time.time() - _tool_start_time) * 1000,
-                            metadata={'original_result_type': type(result).__name__}
+                    def execute_with_context():
+                        with with_injection_context(state):
+                            return self._execute_tool_with_circuit_breaker(function_name, arguments)
+                    
+                    # Use reusable executor to prevent resource leaks
+                    if not hasattr(self, '_tool_executor'):
+                        self._tool_executor = concurrent.futures.ThreadPoolExecutor(
+                            max_workers=2, thread_name_prefix=f"tool-{self.name}"
                        )
-                        
-                        normalized_result = middleware.normalize(result, function_name, mw_ctx)
-                        
-                        # Handle error cases by propagating error message as result
-                        if not normalized_result.success and normalized_result.error_message:
-                            # For failed tools, include error context in the result
-                            result = f"Tool Error: {normalized_result.error_message}"
-                        else:
-                            # For successful tools, use the normalized content
-                            result = normalized_result.content
-                        
-                        # Store normalized result for hooks to access full context
-                        self._last_normalized_result = normalized_result
-                        
-                        logger.debug(f"Applied runtime middleware for {runtime_id}: {function_name}")
-                except ImportError:
-                    # Runtime middleware not available - continue without normalization
-                    logger.debug("Runtime middleware not available, skipping normalization")
-                except Exception as e:
-                    # Don't let middleware failures break tool execution
-                    logger.warning(f"Runtime middleware failed for {runtime_id}: {e}")
+                    
+                    future = self._tool_executor.submit(ctx.run, execute_with_context)
+                    try:
+                        result = future.result(timeout=tool_timeout)
+                    except concurrent.futures.TimeoutError:
+                        future.cancel()
+                        logging.warning(f"Tool {function_name} timed out after {tool_timeout}s")
+                        result = {"error": f"Tool timed out after {tool_timeout}s", "timeout": True}
+                else:
+                    with with_injection_context(state):
+                        result = self._execute_tool_with_circuit_breaker(function_name, arguments)
            
            # Apply prompt injection protection for external tools
            # Zero-cost for trusted tools, wraps external content in security markers
@@ -509,78 +310,28 @@ class ToolExecutionMixin:
                try:
                    result_str = str(result)
                    
-                    # Get configured limit
-                    limit = getattr(self, 'tool_output_limit', 16000)
-                    
-                    # Check if we need to spill to artifact store
-                    if len(result_str) > limit:
-                        # Try to use artifact store if available
-                        artifact_ref = None
-                        if hasattr(self, '_artifact_store') and self._artifact_store is not None:
-                            try:
-                                from ..context.artifacts import ArtifactMetadata
-                                
-                                # Create metadata for this artifact
-                                metadata = ArtifactMetadata(
-                                    agent_id=self.name,
-                                    run_id=getattr(self, '_current_run_id', 'unknown'),
-                                    tool_name=function_name,
-                                    turn_id=getattr(self, '_turn_counter', 0),
-                                )
-                                
-                                # Store the full output
-                                artifact_ref = self._artifact_store.store(result_str, metadata)
-                                logging.debug(f"Stored {function_name} output ({len(result_str)} bytes) as artifact {artifact_ref.artifact_id}")
-                                
-                                # Register artifact retrieval tools if not already registered
-                                if not hasattr(self, '_artifact_tools_registered'):
-                                    self._register_artifact_tools()
-                                    self._artifact_tools_registered = True
-                            except Exception as e:
-                                logging.debug(f"Failed to store artifact: {e}")
-                        
-                        # Generate truncated preview
-                        tail_size = min(limit // 5, 2000)
-                        head = result_str[:limit - tail_size]
-                        tail = result_str[-tail_size:] if tail_size > 0 else ""
-                        
-                        # If we stored an artifact, include reference in the output
-                        if artifact_ref:
-                            truncated = (
-                                f"{head}\n"
-                                f"...[{len(result_str):,} chars total, showing first/last portions]...\n"
-                                f"{tail}\n\n"
-                                f"{artifact_ref.to_inline()}"
-                            )
-                        else:
-                            # Fallback to simple truncation
-                            truncated = f"{head}\n...[{len(result_str):,} chars, showing first/last portions]...\n{tail}"
+                    if self.context_manager:
+                        # Use context-aware truncation with configured budget
+                        truncated = self._truncate_tool_output(function_name, result_str)
                    else:
-                        truncated = result_str
-                    
-                    if self.context_manager and hasattr(self, '_truncate_tool_output'):
-                        # Use context-aware truncation if available, but preserve artifact reference
-                        if 'artifact_ref' in locals() and artifact_ref:
-                            # Extract the artifact reference from the truncated string
-                            artifact_inline = artifact_ref.to_inline()
-                            # Remove the artifact reference before context truncation
-                            truncated_without_ref = truncated.replace(artifact_inline, "").rstrip()
-                            # Apply context truncation
-                            truncated_without_ref = self._truncate_tool_output(function_name, truncated_without_ref, tool_call_id)
-                            # Re-append the artifact reference
-                            truncated = f"{truncated_without_ref}\n\n{artifact_inline}"
+                        # Apply default limit even without context management
+                        # This prevents runaway tool outputs from causing overflow
+                        limit = getattr(self, 'tool_output_limit', 16000)
+                        if len(result_str) > limit:
+                            # Use smart truncation format that judge recognizes as OK
+                            tail_size = min(limit // 5, 2000)
+                            head = result_str[:limit - tail_size]
+                            tail = result_str[-tail_size:] if tail_size > 0 else ""
+                            truncated = f"{head}\n...[{len(result_str):,} chars, showing first/last portions]...\n{tail}"
                        else:
-                            truncated = self._truncate_tool_output(function_name, truncated, tool_call_id)
+                            truncated = result_str
                    
                    if len(truncated) < len(result_str):
                        logging.debug(f"Truncated {function_name} output from {len(result_str)} to {len(truncated)} chars")
                        # For dicts, truncate large string fields (e.g., raw_content from search)
                        if isinstance(result, dict):
                            max_field_chars = getattr(self, 'tool_output_limit', 16000) if not self.context_manager else None
-                            result = self._truncate_dict_fields(result, function_name, max_field_chars, tool_call_id)
-                            # Add artifact reference to dict result if available
-                            if 'artifact_ref' in locals() and artifact_ref:
-                                result["_artifact_ref"] = artifact_ref.to_dict()
+                            result = self._truncate_dict_fields(result, function_name, max_field_chars)
                        else:
                            result = truncated
                except Exception as e:
@@ -622,15 +373,6 @@ class ToolExecutionMixin:
            
            # Trigger AFTER_TOOL hook
            from ..hooks import HookEvent, AfterToolInput
-            
-            # Extract error information from normalized result if available
-            tool_error = None
-            if hasattr(self, '_last_normalized_result') and self._last_normalized_result:
-                if not self._last_normalized_result.success and self._last_normalized_result.error_message:
-                    tool_error = self._last_normalized_result.error_message
-                # Clean up temporary attribute
-                delattr(self, '_last_normalized_result')
-            
            after_tool_input = AfterToolInput(
                session_id=getattr(self, '_session_id', 'default'),
                cwd=os.getcwd(),
@@ -640,7 +382,6 @@ class ToolExecutionMixin:
                tool_name=function_name,
                tool_input=arguments,
                tool_output=result,
-                tool_error=tool_error,
                execution_time_ms=(_time.time() - _tool_start_time) * 1000
            )
            self._hook_runner.execute_sync(HookEvent.AFTER_TOOL, after_tool_input, target=function_name)
@@ -653,12 +394,11 @@ class ToolExecutionMixin:
                    self._doom_loop_tracker.mark_progress(f"tool:{function_name}")
            
            # Record tool execution in loop guard
-            if hasattr(self, '_ensure_loop_guard'):
-                loop_guard = self._ensure_loop_guard()
+            if hasattr(self, '_loop_guard') and self._loop_guard:
                is_success = result is not None and not (isinstance(result, dict) and result.get('error'))
-                loop_guard.record(function_name, arguments, is_success)
+                self._loop_guard.record(function_name, arguments, is_success)
                # Handle warning injection for WARN decisions
-                decision = loop_guard.check(function_name, arguments, is_pre_execution=False) 
+                decision = self._loop_guard.check(function_name, arguments, is_pre_execution=False) 
                if decision.action.value == "warn":
                    if isinstance(result, str):
                        result = f"{result}\n\n[loop-guard] {decision.message}"
@@ -677,29 +417,25 @@ class ToolExecutionMixin:
            
            return result
        except Exception as e:
-            # Emit tool call end with error for exceptions that escape the retry loop
+            # Emit tool call end with error
            _duration_ms = (_time.time() - _tool_start_time) * 1000
            _trace_emitter.tool_call_end(self.name, function_name, None, _duration_ms, str(e))
            
-            # Wrap the exception if it's not already a ToolExecutionError
-            if not isinstance(e, ToolExecutionError):
-                is_retryable = not isinstance(e, (ValueError, TypeError, AttributeError))
-                raise ToolExecutionError(
-                    f"Tool '{function_name}' failed: {e}",
-                    tool_name=function_name,
-                    agent_id=self.name,
-                    is_retryable=is_retryable,
-                ) from e
-            raise  # Re-raise if already wrapped
+            # Preserve existing ToolExecutionError unchanged to maintain loop guard HALT behavior
+            if isinstance(e, ToolExecutionError):
+                raise
+            
+            # Gap 3a fix: Wrap exceptions in ToolExecutionError for better observability
+            is_retryable = not isinstance(e, (ValueError, TypeError, AttributeError))
+            raise ToolExecutionError(
+                f"Tool '{function_name}' failed: {e}",
+                tool_name=function_name,
+                agent_id=self.name,
+                is_retryable=is_retryable,
+            ) from e

    def _trigger_after_agent_hook(self, prompt, response, start_time, tools_used=None):
        """Trigger AFTER_AGENT hook and return response."""
-        # During a guarded skill-review turn, skip the entire after-agent
-        # side-effect pipeline (hooks, auto-memory, auto-learning, nudge,
-        # skill-review). The review turn is internal and must not re-fire
-        # these effects or recurse into another review.
-        if getattr(self, "_in_skill_review", False):
-            return response
        from ..hooks import HookEvent, AfterAgentInput
        after_agent_input = AfterAgentInput(
            session_id=getattr(self, '_session_id', 'default'),
@@ -735,31 +471,10 @@ class ToolExecutionMixin:
            # Log learning nudge failures for debugging
            logger.warning("Learning nudge generation failed: %s", e, exc_info=True)

-        # Autonomous skill self-improvement loop (opt-in via self_improve=True).
-        # Runs a guarded review pass restricted to skill_manage. No-op when
-        # disabled or already inside a review (re-entrancy guarded).
-        try:
-            run_review = getattr(self, "_run_skill_review", None)
-            if run_review is not None:
-                run_review(prompt, response, tools_used)
-        except Exception as e:
-            logger.warning(
-                "Skill self-improvement review failed for agent=%s session_id=%s; "
-                "set self_improve=False to disable this best-effort pass. error=%s",
-                getattr(self, "name", None),
-                getattr(self, "_session_id", None),
-                e,
-                exc_info=True,
-            )
-
        return response

    async def _atrigger_after_agent_hook(self, prompt, response, start_time, tools_used=None):
        """Async version: Trigger AFTER_AGENT hook and return response."""
-        # During a guarded skill-review turn, skip the entire after-agent
-        # side-effect pipeline (see sync variant for rationale).
-        if getattr(self, "_in_skill_review", False):
-            return response
        from ..hooks import HookEvent, AfterAgentInput
        after_agent_input = AfterAgentInput(
            session_id=getattr(self, '_session_id', 'default'),
@@ -792,21 +507,6 @@ class ToolExecutionMixin:
            # Log learning nudge failures for debugging
            logger.warning("Learning nudge generation failed: %s", e, exc_info=True)

-        # Autonomous skill self-improvement loop (opt-in via self_improve=True).
-        try:
-            arun_review = getattr(self, "_arun_skill_review", None)
-            if arun_review is not None:
-                await arun_review(prompt, response, tools_used)
-        except Exception as e:
-            logger.warning(
-                "Skill self-improvement review failed for agent=%s session_id=%s; "
-                "set self_improve=False to disable this best-effort pass. error=%s",
-                getattr(self, "name", None),
-                getattr(self, "_session_id", None),
-                e,
-                exc_info=True,
-            )
-
        return response

    def _calculate_llm_cost(self, prompt_tokens: int, completion_tokens: int, response: any = None) -> float:
@@ -831,57 +531,7 @@ class ToolExecutionMixin:
            response=response,
        )

-    DEFAULT_OUTPUT_RESERVATION_TOKENS = 1024
-
-    def _estimate_min_call_cost(self, messages, max_tokens=None) -> float:
-        """Estimate the minimum cost of an upcoming LLM call before dispatch.
-
-        Uses the known input size (messages) plus an output-token reservation
-        so the budget can be enforced as a hard pre-call ceiling. When the agent
-        has no ``max_tokens`` configured, a conservative default reservation
-        (``DEFAULT_OUTPUT_RESERVATION_TOKENS``) is used so a small prompt with a
-        large provider-default response cannot trivially bypass the guard and
-        then overshoot the cap. Input tokens are approximated as ~4 characters
-        per token, matching the rest of the codebase.
-
-        Note: This is intentionally a *minimum* estimate. It counts message
-        content only and does not include tool-schema tokens, so a tool-heavy
-        request may still cost more than estimated; the reactive post-call
-        accounting remains the backstop for that case.
-
-        Args:
-            messages: The messages that will be sent to the LLM.
-            max_tokens: Optional output-token reservation. Falls back to
-                ``DEFAULT_OUTPUT_RESERVATION_TOKENS`` when not provided.
-
-        Returns:
-            Estimated minimum cost in USD for the call.
-        """
-        prompt_chars = 0
-        if messages:
-            for _msg in messages:
-                _content = _msg.get("content") if isinstance(_msg, dict) else None
-                if isinstance(_content, str):
-                    prompt_chars += len(_content)
-                elif isinstance(_content, list):
-                    for _part in _content:
-                        if isinstance(_part, dict):
-                            _text = _part.get("text")
-                            if isinstance(_text, str):
-                                prompt_chars += len(_text)
-                elif _content is not None:
-                    prompt_chars += len(str(_content))
-
-        prompt_tokens = prompt_chars // 4
-        if max_tokens:
-            completion_tokens = int(max_tokens)
-        elif prompt_chars > 0:
-            completion_tokens = self.DEFAULT_OUTPUT_RESERVATION_TOKENS
-        else:
-            completion_tokens = 0
-        return self._calculate_llm_cost(prompt_tokens, completion_tokens)
-
-    def _truncate_dict_fields(self, data: dict, tool_name: str, max_field_chars: int = None, tool_call_id: str = None) -> dict:
+    def _truncate_dict_fields(self, data: dict, tool_name: str, max_field_chars: int = None) -> dict:
        """Truncate large string fields in a dict to prevent context overflow."""
        if max_field_chars is None:
            # Use tool budget from context manager (default 5000 tokens * 4 chars/token = 20000 chars)
@@ -896,30 +546,13 @@ class ToolExecutionMixin:
                tail_limit = int(max_field_chars * 0.15)
                head = value[:head_limit]
                tail = value[-tail_limit:] if tail_limit > 0 else ""
-                truncated = f"{head}\n...[{len(value):,} chars, showing first/last portions]...\n{tail}"
-                
-                # Store full field value for later retrieval
-                try:
-                    from ..runtime.tool_output_store import get_tool_output_store
-                    from uuid import uuid4
-                    store = get_tool_output_store(getattr(self, '_run_id', None))
-                    # Add unique suffix to prevent collisions with repeated keys
-                    unique_suffix = uuid4().hex[:8]
-                    field_call_id = f"{tool_call_id}_{key}_{unique_suffix}" if tool_call_id else f"{tool_name}_{key}_{unique_suffix}"
-                    metadata = store.store(f"{tool_name}.{key}", value, call_id=field_call_id)
-                    if metadata:
-                        truncated = store.format_reference(metadata, truncated)
-                        logging.debug(f"Stored full {tool_name}.{key} field ({len(value)} bytes) at {metadata.get('path')}")
-                except Exception as e:
-                    logging.debug(f"Failed to store dict field: {e}")
-                
-                result[key] = truncated
+                result[key] = f"{head}\n...[{len(value):,} chars, showing first/last portions]...\n{tail}"
                logging.debug(f"Smart truncated field '{key}' from {len(value)} to ~{max_field_chars} chars")
            elif isinstance(value, dict):
-                result[key] = self._truncate_dict_fields(value, tool_name, max_field_chars, tool_call_id)
+                result[key] = self._truncate_dict_fields(value, tool_name, max_field_chars)
            elif isinstance(value, list):
                result[key] = [
-                    self._truncate_dict_fields(item, tool_name, max_field_chars, tool_call_id) if isinstance(item, dict)
+                    self._truncate_dict_fields(item, tool_name, max_field_chars) if isinstance(item, dict)
                    else (self._smart_truncate_str(item, max_field_chars) if isinstance(item, str) and len(item) > max_field_chars else item)
                    for item in value
                ]
@@ -951,22 +584,12 @@ class ToolExecutionMixin:
        from ..approval import get_approval_registry
        from ..approval.protocols import ApprovalRequest, ApprovalDecision
        from ..approval.registry import DEFAULT_DANGEROUS_TOOLS
-        from ..tools import get_registry as get_tool_registry
        
        backend = getattr(self, '_approval_backend', None)
        approve_all = getattr(self, '_approve_all_tools', False)
        
-        # Check tool trust level if available
-        tool_registry = get_tool_registry()
-        trust_level = tool_registry.get_trust_level(tool_name)
-        
        if backend is not None:
-            # Check if tool needs approval based on multiple criteria
-            needs_approval = (
-                approve_all 
-                or tool_name in DEFAULT_DANGEROUS_TOOLS
-                or (trust_level == "external")  # External tools need approval
-            )
+            needs_approval = approve_all or tool_name in DEFAULT_DANGEROUS_TOOLS
            if needs_approval:
                request = ApprovalRequest(
                    tool_name=tool_name,
@@ -0,0 +1,152 @@
+"""
+Unified Chat Mixin for Agent class.
+
+This module consolidates the dual execution paths (custom LLM vs OpenAI)
+into a single async-first protocol-driven implementation, eliminating
+the DRY violation and maintenance burden described in Issue #1304.
+"""
+
+import asyncio
+import logging
+import time
+from typing import Any, Dict, List, Optional, Union
+
+from ..llm import create_llm_dispatcher, UnifiedLLMDispatcher
+
+
+class UnifiedChatMixin:
+    """
+    Unified chat implementation that replaces dual execution paths.
+    
+    This mixin consolidates:
+    - Path A: LLM.get_response() / LLM.get_response_async() 
+    - Path B: OpenAIClient chat completion
+    
+    Into a single async-first protocol-driven dispatch using UnifiedLLMDispatcher.
+    """
+    
+    def _get_unified_dispatcher(self) -> UnifiedLLMDispatcher:
+        """
+        Get or create the unified LLM dispatcher for this agent.
+        
+        This method implements the provider selection logic that was
+        previously scattered in _chat_completion() method.
+        """
+        if hasattr(self, '_unified_dispatcher') and self._unified_dispatcher is not None:
+            return self._unified_dispatcher
+        
+        # Determine which adapter to use based on existing agent configuration
+        if self._using_custom_llm and hasattr(self, 'llm_instance'):
+            # Use LiteLLM adapter for custom LLM instances
+            dispatcher = create_llm_dispatcher(llm_instance=self.llm_instance)
+        else:
+            # Use OpenAI adapter for direct OpenAI access
+            if not hasattr(self, '_openai_client') or self._openai_client is None:
+                # Initialize OpenAI client if not present
+                from ..llm import get_openai_client
+                self._openai_client = get_openai_client(api_key=getattr(self, 'api_key', None))
+            dispatcher = create_llm_dispatcher(openai_client=self._openai_client, model=self.llm)
+        
+        # Cache the dispatcher
+        self._unified_dispatcher = dispatcher
+        return dispatcher
+    
+    def _unified_chat_completion(
+        self, 
+        messages: List[Dict[str, Any]], 
+        temperature: float = 1.0, 
+        tools: Optional[List[Dict[str, Any]]] = None, 
+        stream: bool = True,
+        reasoning_steps: bool = False,
+        task_name: Optional[str] = None,
+        task_description: Optional[str] = None,
+        task_id: Optional[str] = None,
+        response_format: Optional[Dict[str, Any]] = None
+    ) -> Any:
+        """
+        Unified sync chat completion method.
+        
+        This replaces the existing _chat_completion() method's dual dispatch logic
+        with a single path through UnifiedLLMDispatcher.
+        """
+        start_time = time.time()
+        
+        # Get unified dispatcher (handles provider selection)
+        dispatcher = self._get_unified_dispatcher()
+        
+        # Use the same message processing logic but route through unified dispatcher
+        formatted_tools = self._format_tools_for_completion(tools)
+        
+        try:
+            # Single dispatch path for all providers
+            final_response = dispatcher.chat_completion(
+                messages=messages,
+                tools=formatted_tools,
+                temperature=temperature,
+                max_tokens=getattr(self, 'max_tokens', None),
+                stream=stream,
+                response_format=response_format,
+                **{
+                    'reasoning_steps': reasoning_steps,
+                    'task_name': task_name,
+                    'task_description': task_description,
+                    'task_id': task_id,
+                }
+            )
+            
+            return final_response
+            
+        except Exception as e:
+            logging.error(f"Unified chat completion failed: {e}")
+            # Apply the same error handling as the original implementation
+            raise
+    
+    async def _unified_achat_completion(
+        self, 
+        messages: List[Dict[str, Any]], 
+        temperature: float = 1.0, 
+        tools: Optional[List[Dict[str, Any]]] = None, 
+        stream: bool = True,
+        reasoning_steps: bool = False,
+        task_name: Optional[str] = None,
+        task_description: Optional[str] = None,
+        task_id: Optional[str] = None,
+        response_format: Optional[Dict[str, Any]] = None
+    ) -> Any:
+        """
+        Unified async chat completion method.
+        
+        This provides native async implementation, eliminating the need
+        for separate async/sync implementations in the LLM layer.
+        """
+        start_time = time.time()
+        
+        # Get unified dispatcher (handles provider selection)
+        dispatcher = self._get_unified_dispatcher()
+        
+        # Use the same message processing logic but route through unified dispatcher
+        formatted_tools = self._format_tools_for_completion(tools)
+        
+        try:
+            # Single async dispatch path for all providers
+            final_response = await dispatcher.achat_completion(
+                messages=messages,
+                tools=formatted_tools,
+                temperature=temperature,
+                max_tokens=getattr(self, 'max_tokens', None),
+                stream=stream,
+                response_format=response_format,
+                **{
+                    'reasoning_steps': reasoning_steps,
+                    'task_name': task_name,
+                    'task_description': task_description,
+                    'task_id': task_id,
+                }
+            )
+            
+            return final_response
+            
+        except Exception as e:
+            logging.error(f"Unified async chat completion failed: {e}")
+            # Apply the same error handling as the original implementation
+            raise
@@ -34,7 +34,7 @@ class UnifiedExecutionMixin:
    This replaces the duplicated logic between chat/achat and execute_tool/execute_tool_async
    with a single async-first implementation plus sync bridge.
    """
-    
+
    async def _unified_chat_impl(
        self,
        prompt: str,
@@ -78,38 +78,12 @@ class UnifiedExecutionMixin:
        _trace_emitter.agent_start(self.name, {"role": self.role, "goal": self.goal})
        
        # Reset loop guard for new chat turn (always-on protection)
-        if hasattr(self, '_ensure_loop_guard'):
-            self._ensure_loop_guard().reset_turn()
+        if hasattr(self, '_loop_guard') and self._loop_guard:
+            self._loop_guard.reset_turn()
        
        try:
-            # Runtime resolution - check for model-scoped runtime before legacy CLI backend
-            runtime_instance = None
-            if getattr(self, '_runtime_config', None) is not None:
-                runtime_instance = await self._resolve_turn_runtime()
-            
-            # Model-scoped runtime routing - delegate entire turn to resolved runtime
-            if runtime_instance is not None:
-                return await self._chat_via_runtime(
-                    runtime_instance=runtime_instance,
-                    prompt=prompt,
-                    temperature=temperature,
-                    tools=tools,
-                    output_json=output_json,
-                    output_pydantic=output_pydantic,
-                    reasoning_steps=reasoning_steps,
-                    stream=stream,
-                    task_name=task_name,
-                    task_description=task_description,
-                    task_id=task_id,
-                    config=config,
-                    force_retrieval=force_retrieval,
-                    skip_retrieval=skip_retrieval,
-                    attachments=attachments,
-                    tool_choice=tool_choice
-                )
-            elif hasattr(self, '_cli_backend') and self._cli_backend is not None:
-                # CLI Backend routing - delegate entire turn if configured
-                # Note: Turn-time resolution placeholder is in place for future backend switching
+            # CLI Backend routing - delegate entire turn if configured
+            if hasattr(self, '_cli_backend') and self._cli_backend is not None:
                return await self._chat_via_cli_backend(
                    prompt=prompt,
                    temperature=temperature,
@@ -40,21 +40,6 @@ class TaskStatus(Enum):
 # Set up logger
 logger = get_logger(__name__)

-
-def _launch_auth_token() -> Optional[str]:
-    return os.environ.get("PRAISONAI_LAUNCH_AUTH_TOKEN")
-
-
-def _authorise_launch_request(request) -> bool:
-    token = _launch_auth_token()
-    if not token:
-        return True
-    auth = request.headers.get("Authorization", "")
-    if auth.startswith("Bearer ") and auth[7:] == token:
-        return True
-    return request.headers.get("X-Auth-Token") == token
-
-
 # Agent server registry for thread-safe server management


@@ -260,20 +245,7 @@ Expected Output: {task.expected_output}."""
    # Add memory context if available
    if task.memory:
        try:
-            # Use cache-optimized context if available for better prompt caching
-            if hasattr(task.memory, 'build_cache_optimized_context'):
-                try:
-                    cache_result = task.memory.build_cache_optimized_context(
-                        task_descr=task.description,
-                        include_cache_boundary=False  # Don't include boundary for agent task context
-                    )
-                    memory_context = cache_result.get('stable_prefix', '')
-                except Exception as e:
-                    # Fall back to standard context building
-                    logger.debug(f"Cache-optimized context failed for task '{task.description or task.id}', falling back to standard: {e}")
-                    memory_context = task.memory.build_context_for_task(task.description)
-            else:
-                memory_context = task.memory.build_context_for_task(task.description)
+            memory_context = task.memory.build_context_for_task(task.description)
            if memory_context:
                # Log detailed memory context for debugging
                logger.debug(f"Memory context for task '{task.description}': {memory_context}")
@@ -680,7 +652,7 @@ class AgentTeam(SpawnAnnounceProtocol):
            _stream = _output_config.stream
        else:
            _verbose = 0
-            _stream = False
+            _stream = True
        
        # ─────────────────────────────────────────────────────────────────────
        # Resolve EXECUTION param using canonical resolver
@@ -1056,65 +1028,6 @@ class AgentTeam(SpawnAnnounceProtocol):
        task_result = _process_task_result(self, context, agent_output)
        return task_result.task_output

-    def _apply_task_guardrail(self, task, task_id, task_output):
-        """Apply guardrail validation to task output.
-        
-        Returns:
-            tuple: (task_output, should_retry) where should_retry is True if task should be retried
-            
-        Raises:
-            Exception: If guardrail validation fails after max retries
-        """
-        if not task._guardrail_fn:
-            return task_output, False
-            
-        try:
-            guardrail_result = task._process_guardrail(task_output)
-            if not guardrail_result.success:
-                if task.retry_count >= task.max_retries:
-                    raise Exception(
-                        f"Task failed guardrail validation after {task.max_retries} retries. "
-                        f"Last error: {guardrail_result.error}"
-                    )
-                
-                task.retry_count += 1
-                task.status = "in progress"  # Keep task in progress for retry
-                logger.warning(f"Task {task_id}: Guardrail validation failed (retry {task.retry_count}/{task.max_retries}): {guardrail_result.error}")
-                return task_output, True  # Signal retry needed
-            
-            # If guardrail passed and returned a modified result
-            if guardrail_result.result is not None:
-                if isinstance(guardrail_result.result, str):
-                    # Update the task output with the modified result
-                    task_output.raw = guardrail_result.result
-                    # Clear structured fields to avoid stale cache
-                    if hasattr(task_output, 'json_dict'):
-                        task_output.json_dict = None
-                    if hasattr(task_output, 'pydantic'):
-                        task_output.pydantic = None
-                    task.result = task_output
-                elif hasattr(guardrail_result.result, 'raw'):
-                    # Replace with the new task output
-                    task_output = guardrail_result.result
-                    task.result = task_output
-            
-            logger.info(f"Task {task_id}: Guardrail validation passed")
-            return task_output, False  # No retry needed
-            
-        except Exception as e:
-            logger.error(f"Task {task_id}: Error in guardrail processing: {e}")
-            # Handle guardrail failure with retry logic
-            if task.retry_count >= task.max_retries:
-                raise Exception(
-                    f"Task failed due to guardrail processing error after {task.max_retries} retries. "
-                    f"Last error: {e}"
-                ) from e
-            task.retry_count += 1
-            task.status = "in progress"
-            logger.warning(f"Task {task_id}: Guardrail processing error (retry {task.retry_count}/{task.max_retries}): {e}")
-            return task_output, True  # Signal retry needed
-
-
    async def arun_task(self, task_id):
        """Async version of run_task method"""
        if task_id not in self.tasks:
@@ -1133,11 +1046,53 @@ class AgentTeam(SpawnAnnounceProtocol):
            if task.status in ["not started", "in progress"]:
                task_output = await self.aexecute_task(task_id)
                if task_output and self.completion_checker(task, task_output.raw):
-                    # Apply guardrail validation using shared helper
-                    task_output, should_retry = self._apply_task_guardrail(task, task_id, task_output)
-                    if should_retry:
-                        retries += 1
-                        continue
+                    # Run guardrail validation BEFORE marking task complete
+                    if task._guardrail_fn:
+                        try:
+                            guardrail_result = task._process_guardrail(task_output)
+                            if not guardrail_result.success:
+                                if task.retry_count >= task.max_retries:
+                                    raise Exception(
+                                        f"Task failed guardrail validation after {task.max_retries} retries. "
+                                        f"Last error: {guardrail_result.error}"
+                                    )
+                                
+                                task.retry_count += 1
+                                task.status = "in progress"  # Keep task in progress for retry
+                                logger.warning(f"Task {task_id}: Guardrail validation failed (retry {task.retry_count}/{task.max_retries}): {guardrail_result.error}")
+                                retries += 1
+                                continue  # Actually retry the task
+                            
+                            # If guardrail passed and returned a modified result
+                            if guardrail_result.result is not None:
+                                if isinstance(guardrail_result.result, str):
+                                    # Update the task output with the modified result
+                                    task_output.raw = guardrail_result.result
+                                    # Clear structured fields to avoid stale cache
+                                    if hasattr(task_output, 'json_dict'):
+                                        task_output.json_dict = None
+                                    if hasattr(task_output, 'pydantic'):
+                                        task_output.pydantic = None
+                                    task.result = task_output
+                                elif hasattr(guardrail_result.result, 'raw'):
+                                    # Replace with the new task output
+                                    task_output = guardrail_result.result
+                                    task.result = task_output
+                            
+                            logger.info(f"Task {task_id}: Guardrail validation passed")
+                        except Exception as e:
+                            logger.error(f"Task {task_id}: Error in guardrail processing: {e}")
+                            # Handle guardrail failure with retry logic
+                            if task.retry_count >= task.max_retries:
+                                raise Exception(
+                                    f"Task failed due to guardrail processing error after {task.max_retries} retries. "
+                                    f"Last error: {e}"
+                                ) from e
+                            task.retry_count += 1
+                            task.status = "in progress"
+                            logger.warning(f"Task {task_id}: Guardrail processing error (retry {task.retry_count}/{task.max_retries}): {e}")
+                            retries += 1
+                            continue  # Retry the task
                    
                    task.status = "completed"
                    # Run execute_callback for memory operations
@@ -1151,6 +1106,20 @@ class AgentTeam(SpawnAnnounceProtocol):
                            raise
                        if hasattr(task, 'fail_on_memory_error') and task.fail_on_memory_error:
                            raise
+                    
+                    # Run task callback if exists
+                    if task.callback:
+                        try:
+                            if asyncio.iscoroutinefunction(task.callback):
+                                if run_coroutine_safely:
+                                    run_coroutine_safely(task.callback(task_output))
+                                else:
+                                    logger.warning("run_coroutine_safely not available, skipping async callback")
+                            else:
+                                task.callback(task_output)
+                        except Exception as e:
+                            logger.error(f"Error executing task callback for task {task_id}: {e}")
+                            logger.exception(e)
                            
                    self.save_output_to_file(task, task_output)
                    if self.verbose >= 1:
@@ -1365,11 +1334,52 @@ class AgentTeam(SpawnAnnounceProtocol):
            if task.status in ["not started", "in progress"]:
                task_output = self.execute_task(task_id)
                if task_output and self.completion_checker(task, task_output.raw):
-                    # Apply guardrail validation using shared helper
-                    task_output, should_retry = self._apply_task_guardrail(task, task_id, task_output)
-                    if should_retry:
-                        retries += 1
-                        continue
+                    # Add guardrail validation (matches arun_task logic)
+                    if task._guardrail_fn:
+                        try:
+                            guardrail_result = task._process_guardrail(task_output)
+                            if not guardrail_result.success:
+                                if task.retry_count >= task.max_retries:
+                                    raise Exception(
+                                        f"Task failed guardrail validation after {task.max_retries} retries. "
+                                        f"Last error: {guardrail_result.error}"
+                                    )
+                                task.retry_count += 1
+                                task.status = "in progress"  # Keep task in progress for retry
+                                logger.warning(f"Task {task_id}: Guardrail validation failed (retry {task.retry_count}/{task.max_retries}): {guardrail_result.error}")
+                                retries += 1
+                                continue  # Retry the task
+                            
+                            # If guardrail passed and returned a modified result
+                            if guardrail_result.result is not None:
+                                if isinstance(guardrail_result.result, str):
+                                    # Update the task output with the modified result
+                                    task_output.raw = guardrail_result.result
+                                    # Clear structured fields to avoid stale cache
+                                    if hasattr(task_output, 'json_dict'):
+                                        task_output.json_dict = None
+                                    if hasattr(task_output, 'pydantic'):
+                                        task_output.pydantic = None
+                                    task.result = task_output
+                                elif hasattr(guardrail_result.result, 'raw'):
+                                    # Replace with the new task output
+                                    task_output = guardrail_result.result
+                                    task.result = task_output
+                            
+                            logger.info(f"Task {task_id}: Guardrail validation passed")
+                        except Exception as e:
+                            logger.error(f"Task {task_id}: Error in guardrail processing: {e}")
+                            # Handle guardrail failure with retry logic
+                            if task.retry_count >= task.max_retries:
+                                raise Exception(
+                                    f"Task failed due to guardrail processing error after {task.max_retries} retries. "
+                                    f"Last error: {e}"
+                                ) from e
+                            task.retry_count += 1
+                            task.status = "in progress"
+                            logger.warning(f"Task {task_id}: Guardrail processing error (retry {task.retry_count}/{task.max_retries}): {e}")
+                            retries += 1
+                            continue  # Retry the task
                    
                    task.status = "completed"
                    # Run execute_callback for memory operations
@@ -1379,6 +1389,20 @@ class AgentTeam(SpawnAnnounceProtocol):
                    except Exception as e:
                        logger.error(f"Error executing memory callback for task {task_id}: {e}")
                        logger.exception(e)
+                    
+                    # Run task callback if exists
+                    if task.callback:
+                        try:
+                            if asyncio.iscoroutinefunction(task.callback):
+                                if run_coroutine_safely:
+                                    run_coroutine_safely(task.callback(task_output))
+                                else:
+                                    logger.warning("run_coroutine_safely not available, skipping async callback")
+                            else:
+                                task.callback(task_output)
+                        except Exception as e:
+                            logger.error(f"Error executing task callback for task {task_id}: {e}")
+                            logger.exception(e)
                            
                    self.save_output_to_file(task, task_output)
                    
@@ -1969,8 +1993,6 @@ class AgentTeam(SpawnAnnounceProtocol):
            # Define the endpoint handler
            @app.post(path)
            async def handle_query(request: Request, query_data: Optional[AgentQuery] = None):
-                if not _authorise_launch_request(request):
-                    raise HTTPException(status_code=401, detail="Unauthorized")
                # Handle both direct JSON with query field and form data
                if query_data is None:
                    try:
@@ -2049,9 +2071,7 @@ class AgentTeam(SpawnAnnounceProtocol):
            
            # Add GET endpoint to list available agents
            @app.get(f"{path}/list")
-            async def list_agents(request: Request):
-                if not _authorise_launch_request(request):
-                    raise HTTPException(status_code=401, detail="Unauthorized")
+            async def list_agents():
                return {
                    "agents": [
                        {"name": agent.display_name, "id": agent.display_name.lower().replace(' ', '_')}
@@ -2066,8 +2086,6 @@ class AgentTeam(SpawnAnnounceProtocol):
                # Create a closure to capture the agent instance
                def create_agent_handler(agent):
                    async def handle_single_agent(request: Request):
-                        if not _authorise_launch_request(request):
-                            raise HTTPException(status_code=401, detail="Unauthorized")
                        try:
                            request_data = await request.json()
                            query = request_data.get("query", "")
@@ -53,4 +53,3 @@ class AgentAppConfig:
    workers: int = 1
    timeout: int = 60
    metadata: Dict[str, Any] = field(default_factory=dict)
-    api_key: Optional[str] = None
@@ -42,7 +42,7 @@ from typing import Callable, Dict, List, Literal, Optional, Set
 # ── New protocol-driven exports ──────────────────────────────────────────────

 from .protocols import ApprovalProtocol, ApprovalRequest, ApprovalConfig  # noqa: F401
-from .protocols import ApprovalDecision, ApprovalStoreProtocol  # noqa: F401
+from .protocols import ApprovalDecision  # noqa: F401
 from .backends import AutoApproveBackend, ConsoleBackend, CallbackBackend, AgentApproval  # noqa: F401
 from .registry import ApprovalRegistry, DEFAULT_DANGEROUS_TOOLS  # noqa: F401
Author	SHA1	Message	Date
praisonai-triage-agent[bot]andMervin Praison	3e9826bda7	fix: resolve missing AgentTeam import and RouterAgent kwargs forwarding - Fix missing AgentTeam import in self-reflection-optimization-patterns.py - Forward kwargs properly in RouterAgent._build_system_prompt to super() - Addresses critical issues identified by Gemini code reviewer Co-authored-by: Mervin Praison <[email protected]>	2026-06-12 23:25:32 +00:00
praisonai-triage-agent[bot]andMervin Praison	fc298049e2	fix: resolve critical RouterAgent kwargs bug and add missing if __name__ guards - Fixed TypeError in RouterAgent._build_system_prompt() by removing kwargs from super() call - Added if __name__ == '__main__' guards to 6 example files to prevent auto-execution on import - Addresses Gemini Code Assist review feedback Co-authored-by: Mervin Praison <[email protected]>	2026-06-12 23:25:32 +00:00
Dhivya-Bharathy	0b7bf10820	fix(examples): phase-4 API alignment for 40 validation failures on v4.6.52	2026-06-12 23:25:32 +00:00
Dhivya-Bharathy	cbb7c726aa	fix(examples): phase-3 API alignment for consolidated params on v4.6.52	2026-06-12 23:25:12 +00:00
Dhivya-Bharathy	21d1986d0c	fix(examples): align AgentTeam memory and knowledge configs with v4.6.52 API	2026-06-12 23:25:12 +00:00
praisonai-triage-agent[bot]andMervin Praison	26b5792192	fix: correct TypeError in 3 example files by using proper parameter names - Fix MemoryConfig(memory=...) → user_id=... in agent_centric_api.py - Fix Session(memory=...) → user_id=... in session-example.py - Fix build_context_for_task(memory=...) → user_id=... in memory-quality-example.py Co-authored-by: Mervin Praison <[email protected]>	2026-06-12 23:25:12 +00:00
Dhivya-Bharathy	afcea5800e	fix(examples): resolve syntax errors and align with consolidated-params API	2026-06-12 23:25:12 +00:00