fix: surface competing block-anchor spans; make write-barrier test observe cross-probe order

- edit_file: blockAnchorReplacer now returns every span clearing the
  similarity threshold (best first) instead of silently picking the
  single most-similar block. Hiding competitors from the cascade's
  distinct-candidate check could edit the wrong block when two blocks
  share anchors and both interiors are plausible; one clear winner still
  resolves, two plausible blocks now report ambiguity (covered by a
  two-block test that asserts the file is untouched).
- agent tests: the parallel write-barrier test now records both probes
  into one mutex-guarded shared log and asserts the cross-probe ordering
  it always claimed to (write starts after the first read batch ends;
  second read batch starts after the write ends). The previous version
  only compared indices within the read probe's own log, which a
  regression letting the write overlap a batch could still pass; the
  stray no-op mutex assignment is gone.

.github/workflows/zero-action-smoke.yml

@@ -4,12 +4,14 @@ on:
   pull_request:
     paths:
       - action.yml
+      - docs/GITHUB_ACTION.md
       - .github/workflows/zero-action-smoke.yml
   push:
     branches:
       - main
     paths:
       - action.yml
+      - docs/GITHUB_ACTION.md
       - .github/workflows/zero-action-smoke.yml
 
 jobs:
@@ -29,7 +31,8 @@ jobs:
           import sys, yaml
 
           with open("action.yml", "r", encoding="utf-8") as handle:
-              doc = yaml.safe_load(handle)
+              action_raw = handle.read()
+          doc = yaml.safe_load(action_raw)
 
           assert isinstance(doc, dict), "action.yml must be a mapping"
           assert doc.get("name"), "action must declare a name"
@@ -65,6 +68,19 @@ jobs:
           assert "--skip-permissions-unsafe" not in serialized, \
               "action must never pass --skip-permissions-unsafe"
 
+          run_zero = next((step for step in steps if step.get("name") == "Run ZERO"), None)
+          assert run_zero is not None, "action.yml must contain a Run ZERO step"
+          run_zero_script = run_zero.get("run") or ""
+          assert 'export "${ZERO_INPUT_API_KEY_ENV}=${ZERO_INPUT_API_KEY}"' in run_zero_script, \
+              "Run ZERO must export the provider key for the current step"
+          assert "later steps inherit" not in action_raw.lower(), \
+              "provider keys must not be documented as inherited by later steps"
+          for line in run_zero_script.splitlines():
+              if "GITHUB_ENV" in line and (
+                  "ZERO_INPUT_API_KEY" in line or "ZERO_INPUT_API_KEY_ENV" in line
+              ):
+                  raise AssertionError("provider key must not be persisted to GITHUB_ENV")
+
           print(f"action.yml OK: {len(inputs)} inputs, {len(outputs)} outputs, {len(steps)} steps")
           PY
 

AGENTS.md

@@ -26,7 +26,7 @@ Accepted file names, in priority order at each level:
 | `./ZERO.md` | Brand-specific alias. Same format, lower priority. |
 | `./.zero/AGENTS.md` | Project-local, hidden, gitignored. Personal notes that stay out of git. |
 
-Matching is **case-insensitive** on the basename, so `AGENTS.md`, `Agents.md`, and `agents.md` resolve to the same file on Windows and macOS. The git-tracked filename in this repo is `AGENTS.MD` (uppercase `MD`) — keep that on case-sensitive filesystems (Linux, the WSL filesystem, or a CI runner) to match what the loader looks for.
+Matching is **case-insensitive** on the basename, so `AGENTS.md`, `Agents.md`, and `agents.md` resolve to the same file on Windows and macOS. The git-tracked filename in this repo is `AGENTS.md` — keep that on case-sensitive filesystems (Linux, the WSL filesystem, or a CI runner) to match what the loader looks for.
 
 Both files use the same format. YAML frontmatter is optional; the markdown body is loaded as instructions for the agent. Zero reads the file once at session start, so changes take effect on the next `zero` launch — not mid-session.
 

Makefile

@@ -1,4 +1,4 @@
-# Zero build/test/lint targets. AGENTS.MD says "Build with `make`" and "Run `make
+# Zero build/test/lint targets. AGENTS.md says "Build with `make`" and "Run `make
 # lint` before opening a PR" — these targets back those instructions.
 .DEFAULT_GOAL := build
 .PHONY: build build-all test test-race vet fmt fmt-check lint tidy clean help

README.md

@@ -59,20 +59,24 @@ Bun does not run dependency lifecycle scripts by default, so the `postinstall`
 that fetches the Zero binary is skipped and the first run fails with
 `No native binary found next to the npm wrapper`.
 
-Install with Bun, then fetch the binary in one of two ways:
+The simplest fix is to trust the package after installing, which runs the
+blocked postinstall. This works for project and global installs:
 
 ```bash
-# Option A: run the installer manually
+# project install
 bun add @gitlawb/zero
-node node_modules/@gitlawb/zero/scripts/postinstall.mjs
+bun pm trust @gitlawb/zero
 
-# Option B: allow the postinstall to run on install
-# add to your package.json:  "trustedDependencies": ["@gitlawb/zero"]
-bun add @gitlawb/zero
+# global install
+bun add -g @gitlawb/zero
+bun pm -g trust @gitlawb/zero
 ```
 
-For global installs (`bun add -g @gitlawb/zero`), use Option A against the
-global install path, or prefer the install scripts below.
+Alternatives: allow the postinstall up front by adding
+`"trustedDependencies": ["@gitlawb/zero"]` to your project's package.json
+before `bun add`, or run the installer manually
+(`node node_modules/@gitlawb/zero/scripts/postinstall.mjs`) on Bun versions
+that do not have `bun pm trust`.
 
 ### Install scripts
 
@@ -144,16 +148,9 @@ the key in the wizard:
 export OPENAI_API_KEY=sk-...
 export ANTHROPIC_API_KEY=...
 export GEMINI_API_KEY=...
-export AIMLAPI_API_KEY=...
 export LONGCAT_API_KEY=...
 ```
 
-To configure AI/ML API directly, run:
-
-```bash
-zero providers setup aimlapi --set-active
-```
-
 To configure Meituan LongCat (LongCat-2.0) directly, run:
 
 ```bash

README_ZH.md

@@ -44,19 +44,19 @@ npm 包安装一个小型包装器以及与你平台匹配的 Zero 二进制文
 
 Bun 默认不运行依赖的生命周期脚本，因此获取 Zero 二进制文件的 `postinstall` 会被跳过，首次运行会失败并显示 `No native binary found next to the npm wrapper`。
 
-使用 Bun 安装后，通过以下两种方式之一获取二进制文件：
+最简单的解决方法是安装后信任该包，这会运行被阻止的 postinstall。项目安装和全局安装均适用：
 
 ```bash
-# 选项 A：手动运行安装程序
+# 项目安装
 bun add @gitlawb/zero
-node node_modules/@gitlawb/zero/scripts/postinstall.mjs
+bun pm trust @gitlawb/zero
 
-# 选项 B：允许 postinstall 在安装时运行
-# 添加到你的 package.json:  "trustedDependencies": ["@gitlawb/zero"]
-bun add @gitlawb/zero
+# 全局安装
+bun add -g @gitlawb/zero
+bun pm -g trust @gitlawb/zero
 ```
 
-对于全局安装（`bun add -g @gitlawb/zero`），请对全局安装路径使用选项 A，或使用下面的安装脚本。
+其他方式：在 `bun add` 之前将 `"trustedDependencies": ["@gitlawb/zero"]` 添加到项目的 package.json 以提前允许 postinstall；或在不支持 `bun pm trust` 的旧版 Bun 上手动运行安装程序（`node node_modules/@gitlawb/zero/scripts/postinstall.mjs`）。
 
 ### 安装脚本
 

action.yml

@@ -27,14 +27,14 @@ inputs:
   api-key:
     description: >-
       The provider API key. Pass it from a repository or organization secret;
-      it is exported into the step environment and never written to the log.
+      it is exported only for the ZERO step and never written to the log.
     required: false
     default: ""
   api-key-env:
     description: >-
       Name of the environment variable the chosen provider reads its key from
       (for example OPENAI_API_KEY or ANTHROPIC_API_KEY). When set together with
-      api-key, that variable is exported for the run. Leave the action
+      api-key, that variable is exported only for the ZERO step. Leave the action
       provider-agnostic by passing the env name your provider expects rather
       than hardcoding one here.
     required: false
@@ -192,10 +192,9 @@ runs:
             echo "::error::api-key-env must be a valid environment variable name." >&2
             exit 2
           fi
-          # Export in-process so the zero invocation in this same run block sees
-          # it, and append to GITHUB_ENV so any later steps inherit it too.
+          # Export in-process for this Run ZERO step so zero can read it; do
+          # not persist provider credentials to later job steps.
           export "${ZERO_INPUT_API_KEY_ENV}=${ZERO_INPUT_API_KEY}"
-          printf '%s=%s\n' "${ZERO_INPUT_API_KEY_ENV}" "${ZERO_INPUT_API_KEY}" >> "${GITHUB_ENV}"
         fi
         if [ -n "${ZERO_INPUT_PROVIDER}" ]; then
           export ZERO_PROVIDER="${ZERO_INPUT_PROVIDER}"

bin/zero.js

@@ -66,10 +66,12 @@ if (!existsSync(nativePath)) {
       '\n' +
       (ranByBun
         ? 'You installed with Bun, which does not run dependency lifecycle scripts\n' +
-          'by default. To let it run on future installs, add zero to your\n' +
-          'project\'s trustedDependencies:\n' +
+          'by default. Trust the package to run the blocked postinstall:\n' +
+          '  bun pm trust @gitlawb/zero       (project install)\n' +
+          '  bun pm -g trust @gitlawb/zero    (global install)\n' +
+          'On Bun versions without `bun pm trust`, add\n' +
           '  "trustedDependencies": ["@gitlawb/zero"]\n' +
-          'then reinstall.\n' +
+          'to your project package.json and reinstall.\n' +
           '\n'
         : '') +
       'If that fails, build from source: https://github.com/Gitlawb/zero\n' +

docs/GITHUB_ACTION.md

@@ -41,8 +41,9 @@ jobs:
 > provider-neutral: `api-key-env` is the environment variable name the chosen
 > provider reads its key from (for example `ANTHROPIC_API_KEY`, `GEMINI_API_KEY`,
 > `OPENROUTER_API_KEY`), and `api-key` is the secret value. The action exports
-> that variable for the run and never prints it. If your repository already
-> commits a `.zero/config.json` with an active provider, you can omit `provider`.
+> that variable only for the ZERO step and never prints it. If your repository
+> already commits a `.zero/config.json` with an active provider, you can omit
+> `provider`.
 
 ## Inputs
 
@@ -51,8 +52,8 @@ jobs:
 | `prompt` | one of `prompt`/`prompt-file` | `""` | The instruction for ZERO to execute. |
 | `prompt-file` | one of `prompt`/`prompt-file` | `""` | Path (relative to `working-directory`) to a file whose contents are the prompt. |
 | `provider` | no | `""` | Provider id to activate (e.g. `openai`, `anthropic`, `gemini`, `ollama`, or any compatible endpoint). |
-| `api-key` | no | `""` | The provider API key. Pass from a secret; never logged. |
-| `api-key-env` | no | `""` | Env var name the provider reads its key from (e.g. `OPENAI_API_KEY`). Exported with `api-key`. |
+| `api-key` | no | `""` | The provider API key. Pass from a secret; exported only for the ZERO step and never logged. |
+| `api-key-env` | no | `""` | Env var name the provider reads its key from (e.g. `OPENAI_API_KEY`). Exported only for the ZERO step when used with `api-key`. |
 | `model` | no | `""` | Model id. Defaults to the resolved provider's default. |
 | `mode` | no | `""` | Run mode (`zero exec --mode`), e.g. `smart`, `deep`, `fast`. |
 | `auto` | no | `low` | Autonomy ceiling (`zero exec --auto`): `low`, `medium`, or `high`. Conservative by default. |
@@ -162,9 +163,10 @@ jobs:
 ## Security notes
 
 - **Secrets are passed as Action secrets and never logged.** The action exports
-  `api-key` (under `api-key-env`) and `slack-webhook-url` into the step
-  environment only. They are never echoed, and ZERO redacts secret-shaped
-  strings from its own output.
+  `api-key` (under `api-key-env`) only for the ZERO step. `slack-webhook-url`
+  is passed only to the Slack post step. They are never echoed or persisted to
+  later workflow steps, and ZERO redacts secret-shaped strings from its own
+  output.
 - **The sandbox is always active.** This action never passes
   `--skip-permissions-unsafe`, so writes stay inside the checked-out repository
   (plus any roots you grant with `add-dir`). Unsafe mode is never enabled

docs/INSTALL.md

@@ -31,15 +31,24 @@ dependencies (only the installing project's own scripts), so the `postinstall`
 that fetches the Zero binary is silently skipped. The first run then fails with
 `No native binary found next to the npm wrapper`.
 
-To install with Bun, either run the installer manually after installing:
+The simplest fix is to trust the package after installing, which runs the
+blocked postinstall. This works for project and global installs:
 
 ```bash
+# project install
 bun add @gitlawb/zero
-node node_modules/@gitlawb/zero/scripts/postinstall.mjs
+bun pm trust @gitlawb/zero
+
+# global install
+bun add -g @gitlawb/zero
+bun pm -g trust @gitlawb/zero
 ```
 
-Or allow the postinstall to run by adding the package to your project's
-`trustedDependencies` before installing:
+`bun pm untrusted` (or `bun pm -g untrusted`) lists the blocked postinstalls if
+you want to inspect before trusting.
+
+Alternatively, allow the postinstall to run at install time by adding the
+package to your project's `trustedDependencies` before installing:
 
 ```json
 {
@@ -51,8 +60,12 @@ Or allow the postinstall to run by adding the package to your project's
 bun add @gitlawb/zero
 ```
 
-For global installs (`bun add -g @gitlawb/zero`), run the installer manually
-against the global install path, or use the install scripts below.
+On Bun versions that do not have `bun pm trust`, run the installer manually
+after installing:
+
+```bash
+node node_modules/@gitlawb/zero/scripts/postinstall.mjs
+```
 
 Reference: <https://bun.sh/docs/pm/lifecycle>
 

internal/agent/additional_permissions_validation_test.go

@@ -0,0 +1,146 @@
+package agent
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/tools"
+)
+
+// A shell call that sets sandbox_permissions: with_additional_permissions but
+// omits (or malforms) additional_permissions can never succeed: the same
+// validation runs again at grant time regardless of the user's decision. It
+// must be rejected as a plain tool error BEFORE any permission prompt, not
+// surfaced as a confusing "approved but denied" prompt outcome.
+func TestExecuteToolCallRejectsMissingAdditionalPermissionsBeforePrompt(t *testing.T) {
+	root := t.TempDir()
+	registry := tools.NewRegistry()
+	registry.Register(tools.NewBashTool(root))
+
+	promptCalls := 0
+	result, err := executeToolCall(
+		context.Background(),
+		registry,
+		ToolCall{
+			ID:        "c1",
+			Name:      "bash",
+			Arguments: `{"command":"echo hi","sandbox_permissions":"with_additional_permissions"}`,
+		},
+		PermissionModeAuto,
+		Options{
+			OnPermissionRequest: func(ctx context.Context, request PermissionRequest) (PermissionDecision, error) {
+				promptCalls++
+				return PermissionDecision{Action: PermissionDecisionAllow}, nil
+			},
+		},
+	)
+	if err != nil {
+		t.Fatalf("executeToolCall: %v", err)
+	}
+	if result.Status != tools.StatusError {
+		t.Fatalf("status = %q, want error", result.Status)
+	}
+	if promptCalls != 0 {
+		t.Fatalf("OnPermissionRequest called %d times, want 0 (must reject before prompting)", promptCalls)
+	}
+}
+
+// The mismatched-flag case (additional_permissions present without the
+// sandbox_permissions flag) must be rejected the same way.
+func TestExecuteToolCallRejectsAdditionalPermissionsWithoutFlagBeforePrompt(t *testing.T) {
+	root := t.TempDir()
+	registry := tools.NewRegistry()
+	registry.Register(tools.NewBashTool(root))
+
+	promptCalls := 0
+	result, err := executeToolCall(
+		context.Background(),
+		registry,
+		ToolCall{
+			ID:        "c1",
+			Name:      "bash",
+			Arguments: `{"command":"echo hi","additional_permissions":{"network":{"enabled":true}}}`,
+		},
+		PermissionModeAuto,
+		Options{
+			OnPermissionRequest: func(ctx context.Context, request PermissionRequest) (PermissionDecision, error) {
+				promptCalls++
+				return PermissionDecision{Action: PermissionDecisionAllow}, nil
+			},
+		},
+	)
+	if err != nil {
+		t.Fatalf("executeToolCall: %v", err)
+	}
+	if result.Status != tools.StatusError {
+		t.Fatalf("status = %q, want error", result.Status)
+	}
+	if promptCalls != 0 {
+		t.Fatalf("OnPermissionRequest called %d times, want 0 (must reject before prompting)", promptCalls)
+	}
+}
+
+// The rejection message must show a concrete, correctly-shaped example: a
+// model that gets this wrong once (attaching sandbox_permissions out of habit
+// from an earlier call, without a valid payload) needs enough in the error to
+// self-correct on retry instead of repeating the identical mistake.
+func TestExecuteToolCallMissingAdditionalPermissionsErrorIsActionable(t *testing.T) {
+	root := t.TempDir()
+	registry := tools.NewRegistry()
+	registry.Register(tools.NewBashTool(root))
+
+	result, err := executeToolCall(
+		context.Background(),
+		registry,
+		ToolCall{
+			ID:        "c1",
+			Name:      "bash",
+			Arguments: `{"command":"ping example.com","sandbox_permissions":"with_additional_permissions"}`,
+		},
+		PermissionModeAuto,
+		Options{},
+	)
+	if err != nil {
+		t.Fatalf("executeToolCall: %v", err)
+	}
+	if !strings.Contains(result.Output, `{"network": {"enabled": true}}`) {
+		t.Fatalf("output = %q, want a concrete additional_permissions example", result.Output)
+	}
+	if !strings.Contains(result.Output, "omit sandbox_permissions") {
+		t.Fatalf("output = %q, want guidance that the flag can simply be omitted", result.Output)
+	}
+}
+
+// A well-formed additional_permissions payload must still reach the normal
+// permission prompt: this fix only rejects malformed payloads.
+func TestExecuteToolCallStillPromptsForValidAdditionalPermissions(t *testing.T) {
+	root := t.TempDir()
+	registry := tools.NewRegistry()
+	registry.Register(tools.NewBashTool(root))
+
+	promptCalls := 0
+	_, err := executeToolCall(
+		context.Background(),
+		registry,
+		ToolCall{
+			ID:   "c1",
+			Name: "bash",
+			Arguments: `{"command":"echo hi","sandbox_permissions":"with_additional_permissions",` +
+				`"additional_permissions":{"network":{"enabled":true}}}`,
+		},
+		PermissionModeAuto,
+		Options{
+			OnPermissionRequest: func(ctx context.Context, request PermissionRequest) (PermissionDecision, error) {
+				promptCalls++
+				return PermissionDecision{Action: PermissionDecisionDeny}, nil
+			},
+		},
+	)
+	if err != nil {
+		t.Fatalf("executeToolCall: %v", err)
+	}
+	if promptCalls != 1 {
+		t.Fatalf("OnPermissionRequest called %d times, want exactly 1 for a valid elevation request", promptCalls)
+	}
+}

internal/agent/compaction.go

@@ -21,12 +21,17 @@ import (
 // the agent loop owns the provider wiring.
 
 // defaultCompactionPreserveLast is how many trailing messages are kept verbatim
-// when the caller does not specify a preserve count.
-const defaultCompactionPreserveLast = 8
+// when the caller does not specify a preserve count. Kept in sync with the replay
+// reconstruction default in internal/sessions/replay.go.
+const defaultCompactionPreserveLast = 6
 
 // compactionTriggerRatio is the fraction of the context window at which
-// proactive compaction kicks in (top of each turn).
-const compactionTriggerRatio = 0.8
+// proactive compaction kicks in (top of each turn). NOTE: this is speculative
+// tuning — validate against the per-turn latency/compaction metrics (roadmap 5.2)
+// before trusting it. Measured fixed overhead is ~7.4k tokens (not the ~16k the
+// roadmap assumed), so context fills slower; if data shows 0.7 triggers redundant
+// summarization, raise it back toward 0.8.
+const compactionTriggerRatio = 0.7
 
 // summaryLabel prefixes the injected summary so it is unmistakable in the
 // transcript (and so tests can assert on it).
@@ -320,6 +325,43 @@ type compactionState struct {
 	// OnText is deliberately NOT forwarded (compaction stays invisible to the user),
 	// but its token COST must still be counted so usage reports and budgets include it.
 	onUsage func(Usage)
+
+	// calibrationRatio scales the raw byte/4 token estimate toward the provider's
+	// real prompt-token count. ApproxTextTokens over-counts code-heavy content by
+	// ~15-20%, which would trip compaction early (at ~60% of true capacity). It
+	// starts at 1.0 and converges via an EMA as each turn reports actual usage, so
+	// later turns compact nearer to real capacity. Zero is treated as 1.0.
+	calibrationRatio float64
+}
+
+// calibrate folds one turn's (rawEstimate, actualPromptTokens) sample into the
+// running calibration ratio. A single sample is clamped to a sane band so an
+// outlier (a huge cache-read turn, a provider-overhead spike) can't skew the
+// estimate enough to disable or thrash compaction.
+func (state *compactionState) calibrate(rawEstimate int, actualPromptTokens int) {
+	if !state.enabled || rawEstimate <= 0 || actualPromptTokens <= 0 {
+		return
+	}
+	sample := float64(actualPromptTokens) / float64(rawEstimate)
+	if sample < 0.5 {
+		sample = 0.5
+	} else if sample > 2.0 {
+		sample = 2.0
+	}
+	if state.calibrationRatio <= 0 {
+		state.calibrationRatio = 1.0
+	}
+	const alpha = 0.3 // weight on the newest sample; smooths jitter across turns
+	state.calibrationRatio = state.calibrationRatio*(1-alpha) + sample*alpha
+}
+
+// calibratedTokens applies the learned ratio to a raw estimate. Before any sample
+// arrives (ratio unset) it returns the raw estimate unchanged.
+func (state *compactionState) calibratedTokens(raw int) int {
+	if state.calibrationRatio <= 0 {
+		return raw
+	}
+	return int(float64(raw) * state.calibrationRatio)
 }
 
 func newCompactionState(options Options) *compactionState {
@@ -348,7 +390,7 @@ func (state *compactionState) maybeCompact(
 	// the messages; both the threshold check and the shrink check below use the
 	// same term so they stay consistent.
 	toolTokens := estimateToolDefTokens(tools)
-	size := estimateTokens(messages) + toolTokens
+	size := state.calibratedTokens(estimateTokens(messages) + toolTokens)
 	if size <= state.threshold {
 		return messages
 	}
@@ -365,7 +407,7 @@ func (state *compactionState) maybeCompact(
 	// entirely and preserve recent turns verbatim.
 	if pruned, reclaimed := pruneStaleToolOutput(messages, state.preserveLast); reclaimed > 0 {
 		messages = pruned
-		size = estimateTokens(messages) + toolTokens
+		size = state.calibratedTokens(estimateTokens(messages) + toolTokens)
 		if size <= state.threshold {
 			state.lowWaterMark = size
 			return messages
@@ -381,7 +423,7 @@ func (state *compactionState) maybeCompact(
 		// later turn) can try again; we never drop messages on failure here.
 		return messages
 	}
-	newSize := estimateTokens(compacted) + toolTokens
+	newSize := state.calibratedTokens(estimateTokens(compacted) + toolTokens)
 	if newSize >= size {
 		// Compaction did not actually shrink anything (e.g. nothing to
 		// summarize). Leave the history untouched and don't churn next turn.
@@ -439,7 +481,7 @@ func (state *compactionState) recover(
 	// the SAME combined (messages + tool-defs) domain maybeCompact uses, so the
 	// proactive shrink-guard compares like with like.
 	state.reactiveAttempted = true
-	state.lowWaterMark = estimateTokens(result) + estimateToolDefTokens(tools)
+	state.lowWaterMark = state.calibratedTokens(estimateTokens(result) + estimateToolDefTokens(tools))
 	return result, true, nil
 }
 

internal/agent/compaction_calibrate_test.go

@@ -0,0 +1,50 @@
+package agent
+
+import "testing"
+
+func TestCalibrationConvergesTowardActual(t *testing.T) {
+	state := &compactionState{enabled: true}
+
+	// Before any sample, the estimate is unchanged.
+	if got := state.calibratedTokens(1000); got != 1000 {
+		t.Fatalf("uncalibrated tokens = %d, want 1000", got)
+	}
+
+	// The estimator over-counts: our raw estimate is 1000 but the provider reports
+	// 850 (a 0.85 ratio). Feeding that sample repeatedly should pull the ratio
+	// toward 0.85, so a 1000-token raw estimate calibrates downward.
+	for range 20 {
+		state.calibrate(1000, 850)
+	}
+	got := state.calibratedTokens(1000)
+	if got >= 1000 || got < 820 || got > 880 {
+		t.Fatalf("calibrated tokens = %d, want ~850 after convergence", got)
+	}
+}
+
+func TestCalibrateIgnoresDegenerateSamples(t *testing.T) {
+	state := &compactionState{enabled: true}
+	state.calibrate(0, 500) // zero estimate
+	state.calibrate(500, 0) // zero actual
+	if state.calibrationRatio != 0 {
+		t.Fatalf("degenerate samples must not move the ratio, got %v", state.calibrationRatio)
+	}
+	// Disabled state never calibrates.
+	disabled := &compactionState{enabled: false}
+	disabled.calibrate(1000, 850)
+	if disabled.calibrationRatio != 0 {
+		t.Fatalf("disabled compaction must not calibrate, got %v", disabled.calibrationRatio)
+	}
+}
+
+func TestCalibrateClampsOutliers(t *testing.T) {
+	state := &compactionState{enabled: true}
+	// A wild outlier (10x) is clamped to 2.0, so one bad sample can't blow up the
+	// ratio and disable compaction.
+	for range 50 {
+		state.calibrate(1000, 10000)
+	}
+	if state.calibrationRatio > 2.0 {
+		t.Fatalf("ratio should be clamped at 2.0, got %v", state.calibrationRatio)
+	}
+}

internal/agent/compaction_preserve.go

@@ -20,6 +20,16 @@ const (
 	toolNameUpdatePlan = "update_plan"
 	toolNameToolSearch = "tool_search"
 	toolNameSkill      = "skill"
+	toolNameWriteFile  = "write_file"
+	toolNameEditFile   = "edit_file"
+)
+
+// maxRecentEdits caps how many edited files are carried across a compaction, and
+// maxEditNoteBytes caps each edit's one-line note, so a long editing run can't
+// bloat the preserved-state block.
+const (
+	maxRecentEdits   = 20
+	maxEditNoteBytes = 160
 )
 
 const (
@@ -100,12 +110,123 @@ func formatPlanArguments(arguments string) string {
 }
 
 // skillEntry is a named preserved body. It began as loaded-skill state and is
-// reused for loaded tools and project instruction blocks.
+// reused for loaded tools, project instruction blocks, and recent file edits.
 type skillEntry struct {
 	name string
 	body string
 }
 
+// recentEdits returns the files mutated by write_file/edit_file calls in messages
+// — latest note per path, in last-seen order — as skillEntry{name: path, body:
+// note}. After compaction elides the editing turns, this tells the model WHAT it
+// changed in each file (from the tool's result) so it need not re-read to
+// rediscover its own footprint. Capped at maxRecentEdits paths.
+//
+// Ordering is by LAST edit, not first: re-editing a file moves it to the newest
+// position so the tail cap below keeps the file the model most recently touched
+// rather than an earlier, now-stale entry.
+func recentEdits(messages []zeroruntime.Message) []skillEntry {
+	pathByID := map[string]string{}
+	sequence := make([]string, 0)
+	for _, message := range messages {
+		for _, call := range message.ToolCalls {
+			if call.Name != toolNameWriteFile && call.Name != toolNameEditFile {
+				continue
+			}
+			path := editPathFromArguments(call.Arguments)
+			if path == "" {
+				continue
+			}
+			if call.ID != "" {
+				pathByID[call.ID] = path
+			}
+			sequence = append(sequence, path)
+		}
+	}
+	order := lastSeenOrder(sequence)
+	if len(order) == 0 {
+		return nil
+	}
+
+	noteByPath := map[string]string{}
+	for _, message := range messages {
+		if message.Role != zeroruntime.MessageRoleTool || message.ToolCallID == "" {
+			continue
+		}
+		if path, ok := pathByID[message.ToolCallID]; ok {
+			noteByPath[path] = editNote(message.Content)
+		}
+	}
+
+	// Keep the most recent maxRecentEdits paths (the tail of last-seen order).
+	if len(order) > maxRecentEdits {
+		order = order[len(order)-maxRecentEdits:]
+	}
+	entries := make([]skillEntry, 0, len(order))
+	for _, path := range order {
+		entries = append(entries, skillEntry{name: path, body: noteByPath[path]})
+	}
+	return entries
+}
+
+// lastSeenOrder dedupes paths keeping each at its LAST occurrence, preserving
+// chronological order otherwise. A re-edited path therefore lands at the newest
+// position rather than staying pinned to where it first appeared.
+func lastSeenOrder(paths []string) []string {
+	lastIdx := make(map[string]int, len(paths))
+	for i, p := range paths {
+		lastIdx[p] = i
+	}
+	order := make([]string, 0, len(lastIdx))
+	for i, p := range paths {
+		if lastIdx[p] == i {
+			order = append(order, p)
+		}
+	}
+	return order
+}
+
+// editPathFromArguments pulls the target file path from a write_file/edit_file
+// call's JSON arguments (path/file/file_path/filename aliases). Returns "" on
+// malformed arguments or no path.
+func editPathFromArguments(arguments string) string {
+	var parsed struct {
+		Path     string `json:"path"`
+		File     string `json:"file"`
+		FilePath string `json:"file_path"`
+		Filename string `json:"filename"`
+	}
+	if err := json.Unmarshal([]byte(strings.TrimSpace(arguments)), &parsed); err != nil {
+		return ""
+	}
+	for _, candidate := range []string{parsed.Path, parsed.File, parsed.FilePath, parsed.Filename} {
+		if s := strings.TrimSpace(candidate); s != "" {
+			return s
+		}
+	}
+	return ""
+}
+
+// editNote reduces an edit tool's result to a single short line (its first
+// non-empty line, byte-capped on a rune boundary) for the preserved summary.
+func editNote(content string) string {
+	for _, line := range strings.Split(content, "\n") {
+		s := strings.TrimSpace(line)
+		if s == "" {
+			continue
+		}
+		if len(s) > maxEditNoteBytes {
+			limit := maxEditNoteBytes
+			for limit > 0 && !utf8.RuneStart(s[limit]) {
+				limit--
+			}
+			s = s[:limit] + "…"
+		}
+		return s
+	}
+	return ""
+}
+
 // loadedSkills returns the skills loaded via the skill tool in messages — the
 // latest body per name, in first-seen order — matching each skill tool call to
 // its tool result by id.
@@ -310,11 +431,17 @@ func capBody(body string) string {
 // preservedState is the JSON shape of the carried-across-compaction block.
 type preservedState struct {
 	Plan                string                 `json:"plan,omitempty"`
+	RecentEdits         []preservedEdit        `json:"recent_edits,omitempty"`
 	Tools               []preservedTool        `json:"tools,omitempty"`
 	Skills              []preservedSkill       `json:"skills,omitempty"`
 	ProjectInstructions []preservedInstruction `json:"project_instructions,omitempty"`
 }
 
+type preservedEdit struct {
+	Path string `json:"path"`
+	Note string `json:"note,omitempty"`
+}
+
 type preservedTool struct {
 	Name string `json:"name"`
 	Body string `json:"body"`
@@ -347,6 +474,12 @@ func appendPreservedState(summary string, middle []zeroruntime.Message) string {
 		plan = priorState.Plan
 	}
 
+	// Recent edits: merge edits preserved earlier with fresh write_file/edit_file
+	// results in middle (newer note per path wins AND moves to the newest
+	// position), so the tail cap keeps the files the model most recently touched
+	// across repeated compactions rather than dropping a just-re-edited file.
+	edits := capRecentEdits(mergeRecentEdits(preservedEditsToEntries(priorState.RecentEdits), recentEdits(middle)))
+
 	// Tools: preserve deferred tool_search schemas from the transcript. Fresh
 	// loads override older carried copies by name.
 	tools := mergeSkillEntries(preservedToolsToEntries(priorState.Tools), loadedToolSchemas(middle))
@@ -360,19 +493,76 @@ func appendPreservedState(summary string, middle []zeroruntime.Message) string {
 		projectInstructionEntries(middle),
 	)
 
-	if block := formatPreservedState(plan, tools, skills, instructions); block != "" {
+	if block := formatPreservedState(plan, edits, tools, skills, instructions); block != "" {
 		summary += "\n\n" + block
 	}
 	return summary
 }
 
+// mergeRecentEdits overlays fresh edits onto edits preserved by an earlier
+// compaction. Unlike mergeSkillEntries (which keeps refreshed entries in their
+// original slot), a path touched again by a fresh edit MOVES to the newest
+// position and takes the fresh note, so capRecentEdits — which keeps only the
+// most-recent tail — never drops a file the model just re-edited. Paths present
+// only in the older set keep their relative order, ahead of the fresh ones.
+func mergeRecentEdits(older, newer []skillEntry) []skillEntry {
+	if len(newer) == 0 {
+		return older
+	}
+	freshBody := make(map[string]string, len(newer))
+	freshOrder := make([]string, 0, len(newer))
+	for _, e := range newer {
+		if _, ok := freshBody[e.name]; !ok {
+			freshOrder = append(freshOrder, e.name)
+		}
+		freshBody[e.name] = e.body
+	}
+	merged := make([]skillEntry, 0, len(older)+len(freshOrder))
+	// Older entries that were NOT re-edited keep their position, oldest first.
+	for _, e := range older {
+		if _, refreshed := freshBody[e.name]; refreshed {
+			continue // re-appended at its newest position below
+		}
+		merged = append(merged, e)
+	}
+	// Fresh edits follow, in last-seen order, each carrying the fresh note — so
+	// the most recently touched files sit at the tail the cap preserves.
+	for _, name := range freshOrder {
+		merged = append(merged, skillEntry{name: name, body: freshBody[name]})
+	}
+	return merged
+}
+
+// capRecentEdits bounds the preserved edit list to the most recent maxRecentEdits
+// entries after a merge, so repeated compactions can't grow it without limit.
+func capRecentEdits(entries []skillEntry) []skillEntry {
+	if len(entries) > maxRecentEdits {
+		return entries[len(entries)-maxRecentEdits:]
+	}
+	return entries
+}
+
+func preservedEditsToEntries(edits []preservedEdit) []skillEntry {
+	entries := make([]skillEntry, 0, len(edits))
+	for _, e := range edits {
+		if e.Path == "" {
+			continue
+		}
+		entries = append(entries, skillEntry{name: e.Path, body: e.Note})
+	}
+	return entries
+}
+
 // formatPreservedState renders state as the labelled, single-line
 // JSON block. Returns "" when there is nothing to preserve.
-func formatPreservedState(plan string, tools, skills, instructions []skillEntry) string {
-	if plan == "" && len(tools) == 0 && len(skills) == 0 && len(instructions) == 0 {
+func formatPreservedState(plan string, edits, tools, skills, instructions []skillEntry) string {
+	if plan == "" && len(edits) == 0 && len(tools) == 0 && len(skills) == 0 && len(instructions) == 0 {
 		return ""
 	}
 	state := preservedState{Plan: plan}
+	for _, e := range edits {
+		state.RecentEdits = append(state.RecentEdits, preservedEdit{Path: e.name, Note: e.body})
+	}
 	for _, t := range tools {
 		state.Tools = append(state.Tools, preservedTool{Name: t.name, Body: t.body})
 	}

internal/agent/compaction_recent_edits_test.go

@@ -0,0 +1,121 @@
+package agent
+
+import (
+	"fmt"
+	"strings"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+// recentEdits extracts each mutated file's path and a one-line note from the
+// matching tool result, latest note per path in last-seen order.
+func TestRecentEditsExtractsPathsAndNotes(t *testing.T) {
+	messages := []zeroruntime.Message{
+		{Role: zeroruntime.MessageRoleAssistant, ToolCalls: []zeroruntime.ToolCall{
+			{ID: "e1", Name: "write_file", Arguments: `{"path":"internal/foo.go","content":"package foo"}`},
+		}},
+		{Role: zeroruntime.MessageRoleTool, ToolCallID: "e1", Content: "Wrote internal/foo.go (12 lines)"},
+		{Role: zeroruntime.MessageRoleAssistant, ToolCalls: []zeroruntime.ToolCall{
+			{ID: "e2", Name: "edit_file", Arguments: `{"path":"internal/bar.go","old_string":"a","new_string":"b"}`},
+		}},
+		{Role: zeroruntime.MessageRoleTool, ToolCallID: "e2", Content: "Applied edit to internal/bar.go"},
+	}
+
+	edits := recentEdits(messages)
+	if len(edits) != 2 {
+		t.Fatalf("expected 2 edited files, got %d: %#v", len(edits), edits)
+	}
+	if edits[0].name != "internal/foo.go" || !strings.Contains(edits[0].body, "12 lines") {
+		t.Fatalf("first edit = %#v, want foo.go with its note", edits[0])
+	}
+	if edits[1].name != "internal/bar.go" || !strings.Contains(edits[1].body, "Applied edit") {
+		t.Fatalf("second edit = %#v, want bar.go with its note", edits[1])
+	}
+}
+
+// After compaction elides the editing turns, the preserved-state block still
+// names the edited files and what changed, so the model needn't re-read them.
+func TestCompactionPreservesRecentEdits(t *testing.T) {
+	messages := []zeroruntime.Message{
+		{Role: zeroruntime.MessageRoleSystem, Content: "system"},
+		{Role: zeroruntime.MessageRoleUser, Content: "add a flag"},
+		{Role: zeroruntime.MessageRoleAssistant, Content: "editing", ToolCalls: []zeroruntime.ToolCall{
+			{ID: "e1", Name: "write_file", Arguments: `{"path":"cmd/main.go","content":"..."}`},
+		}},
+		{Role: zeroruntime.MessageRoleTool, ToolCallID: "e1", Content: "Wrote cmd/main.go (adds --version flag)"},
+		{Role: zeroruntime.MessageRoleAssistant, Content: "done"},
+		{Role: zeroruntime.MessageRoleUser, Content: "continue"},
+		{Role: zeroruntime.MessageRoleAssistant, Content: "continuing"},
+	}
+	summary := compactStateConversation(t, messages)
+
+	state := parsePreservedStateBlock(summary)
+	if len(state.RecentEdits) != 1 {
+		t.Fatalf("expected 1 preserved edit, got %#v", state.RecentEdits)
+	}
+	if state.RecentEdits[0].Path != "cmd/main.go" || !strings.Contains(state.RecentEdits[0].Note, "--version") {
+		t.Fatalf("preserved edit = %#v, want cmd/main.go + its note", state.RecentEdits[0])
+	}
+}
+
+// A fresh edit note for a path overrides the one carried from an earlier
+// compaction (newer wins) and moves the path to the newest position rather than
+// duplicating it or leaving it pinned to its old slot.
+func TestRecentEditsMergeNewerWins(t *testing.T) {
+	prior := preservedState{RecentEdits: []preservedEdit{{Path: "a.go", Note: "old note"}, {Path: "z.go", Note: "z"}}}
+	older := preservedEditsToEntries(prior.RecentEdits)
+	newer := []skillEntry{{name: "a.go", body: "new note"}, {name: "b.go", body: "added"}}
+
+	merged := mergeRecentEdits(older, newer)
+	// z.go (untouched) keeps its lead; a.go is re-edited so it takes the new note
+	// and moves behind z.go, ahead of the brand-new b.go: [z.go, a.go, b.go].
+	if len(merged) != 3 {
+		t.Fatalf("expected z.go + refreshed a.go + b.go, got %#v", merged)
+	}
+	if merged[0].name != "z.go" {
+		t.Fatalf("untouched z.go should stay first, got %#v", merged)
+	}
+	if merged[1].name != "a.go" || merged[1].body != "new note" {
+		t.Fatalf("re-edited a.go should move to the newest edits with the new note, got %#v", merged[1])
+	}
+	if merged[2].name != "b.go" {
+		t.Fatalf("brand-new b.go should be last, got %#v", merged[2])
+	}
+}
+
+// Regression: once more than maxRecentEdits distinct files are tracked, re-editing
+// an early file must move it into the capped tail rather than leaving it to be
+// dropped — otherwise the next compaction omits exactly the file the model most
+// recently touched.
+func TestRecentEditsCapKeepsReeditedFile(t *testing.T) {
+	// Earlier compaction preserved exactly maxRecentEdits files, f0 the oldest.
+	older := make([]skillEntry, 0, maxRecentEdits)
+	for i := 0; i < maxRecentEdits; i++ {
+		older = append(older, skillEntry{name: fmt.Sprintf("f%d.go", i), body: "old"})
+	}
+	// Fresh window re-edits the OLDEST file (f0) and adds a brand-new file, so the
+	// merged list exceeds the cap and something must be dropped.
+	newer := []skillEntry{{name: "f0.go", body: "re-edited"}, {name: "fresh.go", body: "new"}}
+
+	merged := capRecentEdits(mergeRecentEdits(older, newer))
+	if len(merged) != maxRecentEdits {
+		t.Fatalf("cap should keep %d entries, got %d", maxRecentEdits, len(merged))
+	}
+
+	var f0 *skillEntry
+	for i := range merged {
+		if merged[i].name == "f0.go" {
+			f0 = &merged[i]
+		}
+		if merged[i].name == "f1.go" {
+			t.Fatalf("f1.go (now the least-recently touched) should have been dropped, got %#v", merged)
+		}
+	}
+	if f0 == nil {
+		t.Fatalf("re-edited f0.go must survive the cap, got %#v", merged)
+	}
+	if f0.body != "re-edited" {
+		t.Fatalf("surviving f0.go should carry the fresh note, got %q", f0.body)
+	}
+}

internal/agent/file_diagnostics.go

@@ -0,0 +1,62 @@
+package agent
+
+import (
+	"context"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/Gitlawb/zero/internal/lsp"
+)
+
+// fileDiagnosticsTimeout bounds one inline post-edit diagnostics check so a
+// slow or wedged language server can never hang a tool call; on timeout the
+// edit simply reports without a diagnostics block.
+const fileDiagnosticsTimeout = 10 * time.Second
+
+// NewFileDiagnostics adapts an *lsp.Manager to the per-edit inline diagnostics
+// callback (tools.RunOptions.Diagnostics): it reads the just-written file,
+// checks it against the file's language server, and formats error-severity
+// diagnostics for the model. Warnings and hints are excluded — nagging about
+// style on every edit is noise, while a type error the edit just introduced is
+// exactly what the model should see before its next step. Diagnostics are
+// rendered with workspace-relative paths: the absolute path would put the
+// local username/home directory into the model prompt and session transcript
+// on every edit. Returns nil when manager is nil, disabling inline diagnostics
+// entirely.
+func NewFileDiagnostics(manager *lsp.Manager, workspaceRoot string) func(context.Context, string) string {
+	if manager == nil {
+		return nil
+	}
+	return func(ctx context.Context, absPath string) string {
+		text, err := os.ReadFile(absPath)
+		if err != nil {
+			return ""
+		}
+		checkCtx, cancel := context.WithTimeout(ctx, fileDiagnosticsTimeout)
+		defer cancel()
+		diagnostics, err := manager.Check(checkCtx, absPath, string(text))
+		if err != nil {
+			return ""
+		}
+		errors := lsp.FilterBySeverity(diagnostics, lsp.SeverityError)
+		if len(errors) == 0 {
+			return ""
+		}
+		return lsp.FormatDiagnostics(diagnosticsDisplayPath(workspaceRoot, absPath), errors)
+	}
+}
+
+// diagnosticsDisplayPath renders absPath relative to the workspace root for
+// model-facing output, falling back to the file's base name when the path is
+// outside the workspace (a bare name still identifies the file without
+// exposing the directory layout).
+func diagnosticsDisplayPath(workspaceRoot, absPath string) string {
+	if workspaceRoot != "" {
+		if rel, err := filepath.Rel(workspaceRoot, absPath); err == nil && !strings.HasPrefix(rel, "..") {
+			return rel
+		}
+	}
+	return filepath.Base(absPath)
+}

internal/agent/file_diagnostics_test.go

@@ -0,0 +1,24 @@
+package agent
+
+import (
+	"path/filepath"
+	"testing"
+)
+
+// Diagnostics are model-facing: absolute paths would leak the local username
+// and directory layout into the prompt and session transcript on every edit.
+func TestDiagnosticsDisplayPath(t *testing.T) {
+	root := filepath.Join("/Users", "someone", "project")
+	cases := []struct {
+		root, abs, want string
+	}{
+		{root, filepath.Join(root, "internal", "a.go"), filepath.Join("internal", "a.go")},
+		{root, filepath.Join("/etc", "other.go"), "other.go"}, // outside root -> base name only
+		{"", filepath.Join("/home", "user", "x.go"), "x.go"},  // no root -> base name only
+	}
+	for _, c := range cases {
+		if got := diagnosticsDisplayPath(c.root, c.abs); got != c.want {
+			t.Errorf("diagnosticsDisplayPath(%q, %q) = %q, want %q", c.root, c.abs, got, c.want)
+		}
+	}
+}

internal/agent/guardrails.go

@@ -22,6 +22,12 @@ const (
 	// calls have executed since the last update_plan call.
 	staleToolCallThreshold = 10
 
+	// stalePlanTurnThreshold injects the same one-shot reminder once this many
+	// turns have passed since the last update_plan while plan items are still
+	// pending — the turn-based complement to staleToolCallThreshold, catching a
+	// plan that drifts stale across many turns that each make few tool calls.
+	stalePlanTurnThreshold = 8
+
 	// toolOnlyProgressReminderAt injects a one-shot progress nudge after this
 	// many consecutive turns contain tool calls but no visible assistant text.
 	// It does not stop the run; it tells the model to synthesize what it already
@@ -43,8 +49,12 @@ const (
 	toolFailureHintAt = 2
 	// toolFailureStopAt halts the run after a tool fails this many times in a row
 	// with the same error, so NO model (weak or strong) burns turns looping on a
-	// bad call.
-	toolFailureStopAt = 4
+	// bad call. Set to 6 (not 4): a corrective hint fires at toolFailureHintAt (2),
+	// and a model iterating on a genuinely tricky edit can legitimately fail a few
+	// times after the hint while converging — stopping at 4 cut those runs short.
+	// The streak still resets the moment the tool succeeds or hits a different
+	// error, so this only affects true same-error loops.
+	toolFailureStopAt = 6
 
 	// maxContinueNudges bounds how many times the headless completion gate
 	// (Options.RequireCompletionSignal) re-prompts a model that stopped without a
@@ -354,8 +364,13 @@ type guardState struct {
 	emptyTurns               int
 	totalToolCalls           int
 	toolCallsSincePlanUpdate int
-	planEverCalled           bool
-	notCalledReminderSent    bool
+	// turnsSincePlanUpdate counts turns (not individual tool calls) since the last
+	// update_plan, so a plan that goes stale across many low-tool-call turns is
+	// still caught — the tool-call counter alone can take many turns to trip when
+	// the model makes only one call per turn.
+	turnsSincePlanUpdate  int
+	planEverCalled        bool
+	notCalledReminderSent bool
 	// staleReminderSent records whether the stale reminder has already fired for
 	// the current stale interval. It is cleared when a plan update opens a new
 	// interval, making the reminder one-shot per interval rather than per turn.
@@ -427,11 +442,16 @@ func (state *guardState) observeTurn(collected zeroruntime.CollectedStream) (sto
 		state.toolOnlyReminderSent = false
 	}
 
+	// One turn has passed; the plan-update below resets this to 0 when the model
+	// refreshes the plan this turn.
+	state.turnsSincePlanUpdate++
+
 	for _, call := range collected.ToolCalls {
 		state.totalToolCalls++
 		if call.Name == planToolName {
 			state.planEverCalled = true
 			state.toolCallsSincePlanUpdate = 0
+			state.turnsSincePlanUpdate = 0
 			// A fresh plan update opens a new stale interval.
 			state.staleReminderSent = false
 			// Record how many items remain so the completion gate knows whether
@@ -485,10 +505,12 @@ func (state *guardState) progressReminder() string {
 // number of turns completed so far).
 func (state *guardState) planReminder(turn int) string {
 	// STALE reminder takes priority: a long run without a plan update is the
-	// stronger signal. One-shot per stale interval.
-	if state.planEverCalled &&
-		!state.staleReminderSent &&
-		state.toolCallsSincePlanUpdate >= staleToolCallThreshold {
+	// stronger signal. Fires on either the tool-call streak OR a turn streak with
+	// pending items (so a plan drifting stale across many low-call turns is caught,
+	// while a fully-completed plan is left alone). One-shot per stale interval.
+	if state.planEverCalled && !state.staleReminderSent &&
+		(state.toolCallsSincePlanUpdate >= staleToolCallThreshold ||
+			(state.turnsSincePlanUpdate >= stalePlanTurnThreshold && state.planItemsPending > 0)) {
 		state.staleReminderSent = true
 		return planStaleReminder(state.toolCallsSincePlanUpdate)
 	}

internal/agent/loop.go

@@ -25,7 +25,12 @@ const maxTurnsFinalAnswerPrompt = "You have reached the tool-turn limit. Do not
 // WITH NO OUTPUT yet is re-issued on a fresh connection before giving up. Only
 // the no-output case is retried (a partial turn would duplicate), so this is a
 // safe recovery for a stalled/dead pooled connection.
-const maxStreamStallRetries = 2
+//
+// Set to 1 (not 2): each attempt can itself idle for the full stream timeout
+// (~5min) before the stall is even detected, so 2 retries left an interactive
+// session frozen for ~15min. One retry keeps the common single-hiccup recovery
+// while bounding the worst case to ~2× the idle timeout.
+const maxStreamStallRetries = 1
 
 const (
 	toolResultMetaControl       = "control"
@@ -145,6 +150,11 @@ func Run(ctx context.Context, prompt string, provider Provider, options Options)
 	// has already been demanded this run, so it fires at most once.
 	acceptanceRequested := false
 
+	// toolDefCache memoizes each tool's rendered JSON-schema definition across
+	// turns (a tool's advertised schema is stable for the run), so partitionTools
+	// doesn't re-run the recursive schema→map conversion for every tool every turn.
+	toolDefCache := map[string]zeroruntime.ToolDefinition{}
+
 	result := Result{Messages: copyMessages(messages)}
 	for turn := 0; turn < maxTurns; turn++ {
 		result.Turns = turn + 1
@@ -153,7 +163,7 @@ func Run(ctx context.Context, prompt string, provider Provider, options Options)
 		// the tool-definition tokens (they ride on every request) in its estimate.
 		// partitionTools depends only on registry/permissions/options/loaded, not on
 		// the messages, so computing it before compaction is safe.
-		exposed, _ := partitionTools(registry, permissionMode, options, loaded)
+		exposed, _ := partitionToolsCached(registry, permissionMode, options, loaded, toolDefCache)
 
 		// PROACTIVE compaction: if the history is approaching the model's
 		// context window, summarize the oldest middle before building the
@@ -340,6 +350,16 @@ func Run(ctx context.Context, prompt string, provider Provider, options Options)
 			}
 		}
 
+		// Calibrate the compaction token estimator against the provider's real
+		// prompt-token count for the request we just sent, so later turns trigger
+		// compaction near true capacity instead of ~15% early on code-heavy history.
+		// Recompute the estimate HERE (not at request-build time): a reactive
+		// compaction may have replaced `messages` with a smaller set and re-sent, so
+		// this reflects the request that actually produced collected.Usage. The
+		// assistant reply is appended below, after this, so `messages` is still the
+		// sent request.
+		compactor.calibrate(estimateTokens(messages)+estimateToolDefTokens(exposed), collected.Usage.InputTokens)
+
 		// Carry the turn's terminal stop reason so a final answer cut off at the
 		// output token cap (or by a content filter) is reported as truncated. A
 		// tool-call turn normalizes to "" and clears any prior reason.
@@ -475,11 +495,35 @@ func Run(ctx context.Context, prompt string, provider Provider, options Options)
 		// between tool_results breaks strict provider replay) — same after-batch
 		// rationale as turnRequestedModel above.
 		var changedFilesThisBatch []string
+		// Parallel read-ahead state: results for calls[precomputedStart:precomputedEnd]
+		// executed concurrently, consumed strictly in order below.
+		var precomputed []precomputedToolResult
+		precomputedStart, precomputedEnd := 0, 0
 		for index, call := range collected.ToolCalls {
+			// When this call starts a consecutive run of >= 2 auto-allowed read-only
+			// calls, execute the whole run concurrently now (see parallel_tools.go).
+			// The scan happens lazily at the run's first index — never ahead of a
+			// pending mutating call — so read-after-write ordering is preserved.
+			if index >= precomputedEnd {
+				runEnd := index
+				for runEnd < len(collected.ToolCalls) && parallelSafeToolCall(registry, collected.ToolCalls[runEnd], options) {
+					runEnd++
+				}
+				if runEnd-index >= 2 {
+					precomputed = executeParallelReadBatch(ctx, registry, collected.ToolCalls, index, runEnd, permissionMode, options)
+					precomputedStart, precomputedEnd = index, runEnd
+				}
+			}
 			if options.OnToolCall != nil {
 				options.OnToolCall(call)
 			}
-			toolResult, abortErr := executeToolCall(ctx, registry, call, permissionMode, options)
+			var toolResult ToolResult
+			var abortErr error
+			if index >= precomputedStart && index < precomputedEnd {
+				toolResult, abortErr = precomputed[index-precomputedStart].result, precomputed[index-precomputedStart].abortErr
+			} else {
+				toolResult, abortErr = executeToolCall(ctx, registry, call, permissionMode, options)
+			}
 			if options.OnToolResult != nil {
 				options.OnToolResult(toolResult)
 			}
@@ -816,6 +860,25 @@ func executeToolCall(ctx context.Context, registry *tools.Registry, call ToolCal
 		}
 	}
 
+	// A shell command that sets sandbox_permissions: with_additional_permissions
+	// must carry a valid additional_permissions payload; inlineAdditionalPermissionsProfile
+	// is the single source of truth for that shape, and buildPermissionEvent (below)
+	// calls it again to render the prompt's scope text. Check it here, before any
+	// permission prompt is shown: a malformed payload can never be satisfied no
+	// matter what the user decides, so surfacing it as a plain tool error (which the
+	// model can see and retry) is both clearer and avoids presenting a normal-looking
+	// prompt whose "allow" path is guaranteed to fail with a confusing denial.
+	if toolFound && isShellCommandTool(call.Name) {
+		if _, _, err := inlineAdditionalPermissionsProfile(args, options.Cwd); err != nil {
+			return ToolResult{
+				ToolCallID: call.ID,
+				Name:       call.Name,
+				Status:     tools.StatusError,
+				Output:     "Error: " + err.Error(),
+			}, nil
+		}
+	}
+
 	// ask_user is intercepted in the loop (like permissions) so the question can
 	// be routed to an interactive front-end instead of blocking inside the tool.
 	// When no front-end is wired up it degrades to the tool's own graceful Run().
@@ -1032,6 +1095,8 @@ func executeToolCall(ctx context.Context, registry *tools.Registry, call ToolCal
 		// Per-session file version tracker so write_file/edit_file refuse to clobber
 		// a file that changed on disk outside Zero since it was last read.
 		FileTracker: options.FileTracker,
+		// Inline post-edit diagnostics for mutating tools (nil = disabled).
+		Diagnostics: options.FileDiagnostics,
 		// Forward the run's operator tool filters so a filter-aware tool
 		// (tool_search) never discloses or loads an operator-hidden deferred tool.
 		EnabledTools:  options.EnabledTools,
@@ -2322,7 +2387,12 @@ func inlineAdditionalPermissionsProfile(args map[string]any, basePath string) (s
 	}
 	raw, ok := args["additional_permissions"]
 	if !ok || raw == nil {
-		return sandbox.RequestPermissionProfile{}, true, fmt.Errorf("missing additional_permissions; provide at least one of network or file_system")
+		return sandbox.RequestPermissionProfile{}, true, fmt.Errorf(
+			"sandbox_permissions was set to %q but no additional_permissions object was provided. "+
+				`Include one, for example additional_permissions: {"network": {"enabled": true}} or `+
+				`{"file_system": {"write": ["/path"]}}. If this command does not need elevated permissions, `+
+				"omit sandbox_permissions entirely and retry",
+			tools.SandboxPermissionsWithAdditionalPermissions)
 	}
 	data, err := json.Marshal(raw)
 	if err != nil {
@@ -2337,7 +2407,9 @@ func inlineAdditionalPermissionsProfile(args map[string]any, basePath string) (s
 		return sandbox.RequestPermissionProfile{}, true, err
 	}
 	if normalized.Empty() {
-		return sandbox.RequestPermissionProfile{}, true, fmt.Errorf("additional_permissions must include at least one requested permission in network or file_system")
+		return sandbox.RequestPermissionProfile{}, true, fmt.Errorf(
+			"additional_permissions must include at least one of network or file_system, for example " +
+				`{"network": {"enabled": true}} or {"file_system": {"write": ["/path"]}}`)
 	}
 	grantProfile, err := sandbox.RequestPermissionGrantProfile(normalized)
 	if err != nil {
@@ -2436,6 +2508,17 @@ func permissionActionFromSandbox(action sandbox.Action) PermissionAction {
 // exposed. The exposed slice is alpha-sorted by name, matching the legacy order
 // so the inactive path is stable.
 func partitionTools(registry *tools.Registry, permissionMode PermissionMode, options Options, loaded map[string]bool) ([]zeroruntime.ToolDefinition, string) {
+	return partitionToolsCached(registry, permissionMode, options, loaded, nil)
+}
+
+// partitionToolsCached is partitionTools with an optional per-tool definition
+// cache. The partitioning itself (visibility, deferral, ordering) is recomputed
+// every call — it must be, because a tool's deferred state can flip mid-run (e.g.
+// swarm tools un-defer once a swarm is active). Only the expensive part —
+// rendering each tool's JSON-schema parameters — is memoized by tool name, since a
+// tool's advertised name/description/schema is stable for the run. defCache nil
+// disables caching (used by tests and the plain partitionTools entrypoint).
+func partitionToolsCached(registry *tools.Registry, permissionMode PermissionMode, options Options, loaded map[string]bool, defCache map[string]zeroruntime.ToolDefinition) ([]zeroruntime.ToolDefinition, string) {
 	registeredTools := registry.All()
 
 	visible := make([]tools.Tool, 0, len(registeredTools))
@@ -2470,76 +2553,115 @@ func partitionTools(registry *tools.Registry, permissionMode PermissionMode, opt
 
 	active := options.DeferThreshold > 0 && eligible >= options.DeferThreshold && loaderUsable
 
-	definitions := make([]zeroruntime.ToolDefinition, 0, len(visible))
-	exposedNames := make(map[string]bool, len(visible))
+	// INACTIVE: every visible tool eager (except tool_search), alpha-sorted. With no
+	// deferral there is no mid-session loading, so this is byte-stable across turns
+	// and byte-identical to the pre-deferral output.
+	if !active {
+		definitions := make([]zeroruntime.ToolDefinition, 0, len(visible))
+		for _, tool := range visible {
+			if tool.Name() == tools.ToolSearchToolName {
+				continue
+			}
+			definitions = append(definitions, cachedRuntimeToolDefinition(defCache, tool))
+		}
+		sort.Slice(definitions, func(left int, right int) bool {
+			return definitions[left].Name < definitions[right].Name
+		})
+		return definitions, ""
+	}
+
+	// ACTIVE: lay the tools array out so its cacheable prefix does NOT shift when a
+	// deferred tool loads mid-session. The tools block is part of the provider's
+	// cached prefix; if a load reorders it, the cache invalidates from that point
+	// through the system prompt and messages — on EVERY load. Previously the whole
+	// array was alpha-sorted each turn, so a newly loaded tool was INSERTED into the
+	// middle and shifted every later definition. Instead we build three append-only
+	// regions:
+	//   1. non-deferred tools, alpha-sorted — always present, identical every turn;
+	//   2. tool_search — always present, right after the stable block;
+	//   3. loaded deferred tools, alpha-sorted — APPENDED, so an unloaded->loaded
+	//      transition grows the tail instead of inserting into the eager block.
+	// (tool_search's discovery text still shrinks as tools load, so keeping it and
+	// the loaded tail AFTER the eager block preserves the eager tools' cache across a
+	// load; fully stabilizing the loader's own description is a scoped follow-up.)
+	eager := make([]zeroruntime.ToolDefinition, 0, len(visible))
+	loadedTail := make([]zeroruntime.ToolDefinition, 0)
 	var hiddenTools []tools.Tool
 	for _, tool := range visible {
 		name := tool.Name()
-		deferred := tools.IsDeferred(tool)
-
-		if !active {
-			// Inactive: byte-identical to legacy, but tool_search is never advertised.
-			if name == tools.ToolSearchToolName {
-				continue
+		if name == tools.ToolSearchToolName {
+			continue // added explicitly between the eager block and the loaded tail
+		}
+		if tools.IsDeferred(tool) {
+			if loaded[name] {
+				loadedTail = append(loadedTail, cachedRuntimeToolDefinition(defCache, tool))
+			} else {
+				hiddenTools = append(hiddenTools, tool)
 			}
-			definitions = append(definitions, zeroruntime.ToolDefinition{
-				Name:        name,
-				Description: tool.Description(),
-				Parameters:  schemaToRuntimeMap(tool.Parameters()),
-			})
-			exposedNames[name] = true
 			continue
 		}
-
-		// Active path.
-		if deferred && !loaded[name] {
-			hiddenTools = append(hiddenTools, tool)
-			continue
-		}
-		definitions = append(definitions, zeroruntime.ToolDefinition{
-			Name:        name,
-			Description: tool.Description(),
-			Parameters:  schemaToRuntimeMap(tool.Parameters()),
-		})
-		exposedNames[name] = true
+		eager = append(eager, cachedRuntimeToolDefinition(defCache, tool))
 	}
-
-	// On the ACTIVE path tool_search is guaranteed runnable (active implies
-	// loaderUsable), so it must ALWAYS be reachable — even when a non-empty
-	// EnabledTools allowlist omits it (the operator allowlisted the deferred
-	// tools, not the loader). Expose its full definition whenever it is not
-	// already in the exposed set. This never runs on the inactive path, so the
-	// byte-identical below-threshold output is preserved.
-	discovery := ""
-	if active && len(hiddenTools) > 0 {
-		discovery = tools.BuildToolSearchDescription(hiddenTools)
-	}
-	if active && !exposedNames[tools.ToolSearchToolName] {
-		description := loader.Description()
-		if discovery != "" {
-			description = discovery
-		}
-		definitions = append(definitions, zeroruntime.ToolDefinition{
-			Name:        loader.Name(),
-			Description: description,
-			Parameters:  schemaToRuntimeMap(loader.Parameters()),
-		})
-	} else if active && discovery != "" {
-		for i := range definitions {
-			if definitions[i].Name == tools.ToolSearchToolName {
-				definitions[i].Description = discovery
-				break
-			}
-		}
-	}
-
-	sort.Slice(definitions, func(left int, right int) bool {
-		return definitions[left].Name < definitions[right].Name
+	sort.Slice(eager, func(left int, right int) bool {
+		return eager[left].Name < eager[right].Name
+	})
+	sort.Slice(loadedTail, func(left int, right int) bool {
+		return loadedTail[left].Name < loadedTail[right].Name
 	})
 
+	discovery := ""
+	if len(hiddenTools) > 0 {
+		discovery = tools.BuildToolSearchDescription(hiddenTools)
+	}
+
+	// tool_search is guaranteed runnable on the active path, so it is ALWAYS exposed
+	// — even when a non-empty EnabledTools allowlist omits it (the operator
+	// allowlisted the deferred tools, not the loader). It sits right after the stable
+	// eager block and carries the discovery list for still-hidden tools.
+	description := loader.Description()
+	if discovery != "" {
+		description = discovery
+	}
+	definitions := make([]zeroruntime.ToolDefinition, 0, len(eager)+1+len(loadedTail))
+	definitions = append(definitions, eager...)
+	definitions = append(definitions, zeroruntime.ToolDefinition{
+		Name:        loader.Name(),
+		Description: description,
+		Parameters:  schemaToRuntimeMap(loader.Parameters()),
+	})
+	definitions = append(definitions, loadedTail...)
+
 	return definitions, discovery
 }
 
+// cachedRuntimeToolDefinition returns the tool's rendered definition, reusing a
+// cached render when defCache holds one for this tool name. A tool's advertised
+// definition is stable across a run, so caching skips the recursive schema→map
+// conversion (schemaToRuntimeMap) that would otherwise run for every tool on every
+// turn. tool_search is excluded by its callers (its description is dynamic), so it
+// never poisons the cache. A nil cache computes fresh.
+func cachedRuntimeToolDefinition(defCache map[string]zeroruntime.ToolDefinition, tool tools.Tool) zeroruntime.ToolDefinition {
+	if defCache == nil {
+		return runtimeToolDefinition(tool)
+	}
+	if def, ok := defCache[tool.Name()]; ok {
+		return def
+	}
+	def := runtimeToolDefinition(tool)
+	defCache[tool.Name()] = def
+	return def
+}
+
+// runtimeToolDefinition renders a tool's advertised definition (name, description,
+// JSON-schema parameters) as sent to the provider.
+func runtimeToolDefinition(tool tools.Tool) zeroruntime.ToolDefinition {
+	return zeroruntime.ToolDefinition{
+		Name:        tool.Name(),
+		Description: tool.Description(),
+		Parameters:  schemaToRuntimeMap(tool.Parameters()),
+	}
+}
+
 func ToolVisible(tool tools.Tool, permissionMode PermissionMode, enabledTools []string, disabledTools []string) bool {
 	return ToolAllowedByFilters(tool.Name(), enabledTools, disabledTools) && ToolAdvertised(tool, permissionMode)
 }

internal/agent/parallel_tools.go

@@ -0,0 +1,97 @@
+package agent
+
+import (
+	"context"
+	"sync"
+
+	"github.com/Gitlawb/zero/internal/tools"
+)
+
+// Parallel read-ahead for tool batches. When a turn requests several
+// independent lookups (read_file + grep + glob is the common shape), executing
+// them one after another serializes pure I/O waits. A consecutive run of
+// auto-allowed read-only calls is executed concurrently instead; results are
+// then consumed in the original call order, so guard counters, message
+// ordering, abort semantics, and the surface's call/result event pairing are
+// byte-identical to sequential execution. Runs never span a mutating call: a
+// read that follows a write must observe the write, so eligibility is decided
+// per consecutive run, not per batch.
+
+// maxParallelReadTools bounds concurrent read-only tool executions in a turn.
+const maxParallelReadTools = 8
+
+// precomputedToolResult is one parallel read-ahead execution, keyed back to
+// its batch index by the caller.
+type precomputedToolResult struct {
+	result   ToolResult
+	abortErr error
+}
+
+// parallelSafeToolCall reports whether call may run concurrently with its
+// neighbors: the tool must exist, be side-effect-free (SideEffectRead), and be
+// auto-allowed for these args, so no interactive prompt or workspace mutation
+// is on the hot path. Loop-intercepted tools (ask_user, request_permissions)
+// and tool_search (mutates the deferred-tool set) stay sequential.
+func parallelSafeToolCall(registry *tools.Registry, call ToolCall, options Options) bool {
+	switch call.Name {
+	case "ask_user", tools.RequestPermissionsToolName, tools.ToolSearchToolName:
+		return false
+	}
+	tool, found := registry.Get(call.Name)
+	if !found || tool.Safety().SideEffect != tools.SideEffectRead {
+		return false
+	}
+	args := map[string]any{}
+	if call.Arguments != "" {
+		if err := decodeToolArguments(call.Arguments, &args); err != nil {
+			return false
+		}
+	}
+	return effectivePermission(tool, args) == tools.PermissionAllow
+}
+
+// executeParallelReadBatch runs calls[start:end] concurrently (bounded by
+// maxParallelReadTools) and returns results indexed relative to start. All
+// execution-side callbacks that can fire inside executeToolCall are serialized
+// behind one mutex: a permission prompt (a sandbox preflight can demand one
+// even for an auto-allowed read) must never appear twice at once on an
+// interactive front-end, and OnPermission event handlers append to shared
+// session-recording state without their own locking — two batched reads under
+// a granted extra root would otherwise race (the pre-batch serial loop never
+// had two callbacks in flight at once).
+func executeParallelReadBatch(ctx context.Context, registry *tools.Registry, calls []ToolCall, start, end int, permissionMode PermissionMode, options Options) []precomputedToolResult {
+	batchOptions := options
+	var callbackMutex sync.Mutex
+	if options.OnPermissionRequest != nil {
+		inner := options.OnPermissionRequest
+		batchOptions.OnPermissionRequest = func(ctx context.Context, request PermissionRequest) (PermissionDecision, error) {
+			callbackMutex.Lock()
+			defer callbackMutex.Unlock()
+			return inner(ctx, request)
+		}
+	}
+	if options.OnPermission != nil {
+		inner := options.OnPermission
+		batchOptions.OnPermission = func(event PermissionEvent) {
+			callbackMutex.Lock()
+			defer callbackMutex.Unlock()
+			inner(event)
+		}
+	}
+
+	results := make([]precomputedToolResult, end-start)
+	semaphore := make(chan struct{}, maxParallelReadTools)
+	var waitGroup sync.WaitGroup
+	for index := start; index < end; index++ {
+		waitGroup.Add(1)
+		go func(index int) {
+			defer waitGroup.Done()
+			semaphore <- struct{}{}
+			defer func() { <-semaphore }()
+			result, abortErr := executeToolCall(ctx, registry, calls[index], permissionMode, batchOptions)
+			results[index-start] = precomputedToolResult{result: result, abortErr: abortErr}
+		}(index)
+	}
+	waitGroup.Wait()
+	return results
+}

internal/agent/parallel_tools_test.go

@@ -0,0 +1,204 @@
+package agent
+
+import (
+	"context"
+	"fmt"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/Gitlawb/zero/internal/tools"
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+// probeTool records execution overlap and ordering so tests can assert what
+// actually ran concurrently.
+type probeTool struct {
+	name       string
+	sideEffect tools.SideEffect
+	delay      time.Duration
+	// shared, when set, receives this probe's start/end entries too, so a test
+	// can observe ordering ACROSS probes (e.g. reads vs a write barrier).
+	shared *probeLog
+
+	mu        sync.Mutex
+	active    int
+	maxActive int
+	log       []string
+}
+
+// probeLog is a mutex-guarded event log shared between probes.
+type probeLog struct {
+	mu      sync.Mutex
+	entries []string
+}
+
+func (log *probeLog) append(entry string) {
+	log.mu.Lock()
+	defer log.mu.Unlock()
+	log.entries = append(log.entries, entry)
+}
+
+func (log *probeLog) snapshot() []string {
+	log.mu.Lock()
+	defer log.mu.Unlock()
+	return append([]string(nil), log.entries...)
+}
+
+func (tool *probeTool) Name() string        { return tool.name }
+func (tool *probeTool) Description() string { return "test probe tool" }
+func (tool *probeTool) Parameters() tools.Schema {
+	return tools.Schema{
+		Type:                 "object",
+		Properties:           map[string]tools.PropertySchema{"id": {Type: "string"}},
+		AdditionalProperties: false,
+	}
+}
+func (tool *probeTool) Safety() tools.Safety {
+	return tools.Safety{SideEffect: tool.sideEffect, Permission: tools.PermissionAllow, Reason: "test"}
+}
+func (tool *probeTool) Run(_ context.Context, args map[string]any) tools.Result {
+	id, _ := args["id"].(string)
+	tool.mu.Lock()
+	tool.active++
+	if tool.active > tool.maxActive {
+		tool.maxActive = tool.active
+	}
+	tool.log = append(tool.log, "start:"+id)
+	tool.mu.Unlock()
+	if tool.shared != nil {
+		tool.shared.append("start:" + id)
+	}
+	time.Sleep(tool.delay)
+	if tool.shared != nil {
+		tool.shared.append("end:" + id)
+	}
+	tool.mu.Lock()
+	tool.active--
+	tool.log = append(tool.log, "end:"+id)
+	tool.mu.Unlock()
+	return tools.Result{Status: tools.StatusOK, Output: "probe " + id}
+}
+
+func probeCallEvents(callID, toolName, id string) []zeroruntime.StreamEvent {
+	return []zeroruntime.StreamEvent{
+		{Type: zeroruntime.StreamEventToolCallStart, ToolCallID: callID, ToolName: toolName},
+		{Type: zeroruntime.StreamEventToolCallDelta, ToolCallID: callID, ArgumentsFragment: fmt.Sprintf(`{"id":%q}`, id)},
+		{Type: zeroruntime.StreamEventToolCallEnd, ToolCallID: callID},
+	}
+}
+
+func TestParallelSafeToolCall(t *testing.T) {
+	registry := tools.NewRegistry()
+	registry.Register(&probeTool{name: "probe_read", sideEffect: tools.SideEffectRead})
+	registry.Register(&probeTool{name: "probe_write", sideEffect: tools.SideEffectWrite})
+
+	call := func(name, args string) ToolCall { return ToolCall{ID: "c", Name: name, Arguments: args} }
+	if !parallelSafeToolCall(registry, call("probe_read", `{"id":"a"}`), Options{}) {
+		t.Fatal("auto-allowed read tool must be parallel-safe")
+	}
+	if parallelSafeToolCall(registry, call("probe_write", `{"id":"a"}`), Options{}) {
+		t.Fatal("mutating tool must not be parallel-safe")
+	}
+	if parallelSafeToolCall(registry, call("unknown_tool", `{}`), Options{}) {
+		t.Fatal("unknown tool must not be parallel-safe")
+	}
+	if parallelSafeToolCall(registry, call("probe_read", `{"id":`), Options{}) {
+		t.Fatal("undecodable arguments must not be parallel-safe")
+	}
+	if parallelSafeToolCall(registry, call("ask_user", `{}`), Options{}) {
+		t.Fatal("loop-intercepted tools must stay sequential")
+	}
+}
+
+func TestRunExecutesConsecutiveReadsConcurrently(t *testing.T) {
+	probe := &probeTool{name: "probe_read", sideEffect: tools.SideEffectRead, delay: 60 * time.Millisecond}
+	registry := tools.NewRegistry()
+	registry.Register(probe)
+
+	turnOne := append(probeCallEvents("call-1", "probe_read", "a"), probeCallEvents("call-2", "probe_read", "b")...)
+	turnOne = append(turnOne, probeCallEvents("call-3", "probe_read", "c")...)
+	turnOne = append(turnOne, zeroruntime.StreamEvent{Type: zeroruntime.StreamEventDone})
+	provider := &mockProvider{
+		turns: [][]zeroruntime.StreamEvent{
+			turnOne,
+			{{Type: zeroruntime.StreamEventText, Content: "done"}, {Type: zeroruntime.StreamEventDone}},
+		},
+	}
+
+	var results []ToolResult
+	_, err := Run(context.Background(), "probe", provider, Options{
+		Registry:     registry,
+		OnToolResult: func(result ToolResult) { results = append(results, result) },
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if probe.maxActive < 2 {
+		t.Fatalf("consecutive read-only calls must overlap, max concurrency was %d", probe.maxActive)
+	}
+	// Results must still be recorded in original call order.
+	if len(results) != 3 || results[0].ToolCallID != "call-1" || results[1].ToolCallID != "call-2" || results[2].ToolCallID != "call-3" {
+		t.Fatalf("tool results out of order: %#v", results)
+	}
+	messages := provider.requests[1].Messages
+	var toolOrder []string
+	for _, message := range messages {
+		if message.Role == zeroruntime.MessageRoleTool {
+			toolOrder = append(toolOrder, message.ToolCallID)
+		}
+	}
+	if len(toolOrder) != 3 || toolOrder[0] != "call-1" || toolOrder[1] != "call-2" || toolOrder[2] != "call-3" {
+		t.Fatalf("recorded tool messages out of order: %v", toolOrder)
+	}
+}
+
+func TestRunParallelReadsNeverSpanMutatingCall(t *testing.T) {
+	shared := &probeLog{}
+	read := &probeTool{name: "probe_read", sideEffect: tools.SideEffectRead, delay: 30 * time.Millisecond, shared: shared}
+	write := &probeTool{name: "probe_write", sideEffect: tools.SideEffectWrite, shared: shared}
+	registry := tools.NewRegistry()
+	registry.Register(read)
+	registry.Register(write)
+
+	turnOne := append(probeCallEvents("call-1", "probe_read", "r1"), probeCallEvents("call-2", "probe_read", "r2")...)
+	turnOne = append(turnOne, probeCallEvents("call-3", "probe_write", "w")...)
+	turnOne = append(turnOne, probeCallEvents("call-4", "probe_read", "r3")...)
+	turnOne = append(turnOne, probeCallEvents("call-5", "probe_read", "r4")...)
+	turnOne = append(turnOne, zeroruntime.StreamEvent{Type: zeroruntime.StreamEventDone})
+	provider := &mockProvider{
+		turns: [][]zeroruntime.StreamEvent{
+			turnOne,
+			{{Type: zeroruntime.StreamEventText, Content: "done"}, {Type: zeroruntime.StreamEventDone}},
+		},
+	}
+
+	_, err := Run(context.Background(), "probe", provider, Options{Registry: registry})
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Cross-probe ordering on the SHARED log: the write must start only after
+	// both first-batch reads finished, and both second-batch reads must start
+	// only after the write finished — batches never cross a mutating call.
+	log := shared.snapshot()
+	index := func(entry string) int {
+		for i, e := range log {
+			if e == entry {
+				return i
+			}
+		}
+		t.Fatalf("entry %q missing from shared log: %v", entry, log)
+		return -1
+	}
+	writeStart, writeEnd := index("start:w"), index("end:w")
+	if firstBatchMaxEnd := max(index("end:r1"), index("end:r2")); writeStart < firstBatchMaxEnd {
+		t.Fatalf("write started before the first read batch finished: %v", log)
+	}
+	if secondBatchMinStart := min(index("start:r3"), index("start:r4")); secondBatchMinStart < writeEnd {
+		t.Fatalf("second read batch started before the write finished: %v", log)
+	}
+	if read.maxActive < 2 {
+		t.Fatalf("reads within a batch must overlap, max concurrency was %d", read.maxActive)
+	}
+}

internal/agent/partition_cache_stable_test.go

@@ -0,0 +1,44 @@
+package agent
+
+import (
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/tools"
+)
+
+func TestPartitionToolsActiveAppendsLoadedToolAfterEagerBlock(t *testing.T) {
+	// Cache-stability property: loading a deferred tool must APPEND it after the
+	// always-eager tools, not sort it into the middle of them — otherwise the tool
+	// definitions (the provider's cacheable prefix) shift on every load. The old
+	// behaviour alpha-sorted the whole array, so "mcp__srv__alpha" jumped ahead of
+	// "read_file".
+	root := t.TempDir()
+	registry := tools.NewRegistry()
+	registry.Register(tools.NewReadFileTool(root)) // non-deferred, "read_file"
+	registry.Register(fakeToolSearchTool{})
+	registry.Register(fakeDeferredTool{name: "mcp__srv__alpha", desc: "alpha"})
+	registry.Register(fakeDeferredTool{name: "mcp__srv__beta", desc: "beta"})
+
+	exposed, _ := partitionTools(registry, PermissionModeAuto, Options{DeferThreshold: 2}, map[string]bool{"mcp__srv__alpha": true})
+
+	pos := func(name string) int {
+		for i := range exposed {
+			if exposed[i].Name == name {
+				return i
+			}
+		}
+		return -1
+	}
+	readPos, searchPos, alphaPos := pos("read_file"), pos("tool_search"), pos("mcp__srv__alpha")
+	if readPos < 0 || searchPos < 0 || alphaPos < 0 {
+		t.Fatalf("expected read_file, tool_search and loaded alpha all exposed, got %#v", exposed)
+	}
+	// The eager tools must precede the loaded deferred tool (appended, not inserted).
+	if !(readPos < alphaPos && searchPos < alphaPos) {
+		t.Fatalf("loaded deferred tool must be appended after the eager block; got read=%d search=%d alpha=%d", readPos, searchPos, alphaPos)
+	}
+	// beta was never loaded — it stays hidden.
+	if pos("mcp__srv__beta") != -1 {
+		t.Fatalf("unloaded deferred tool must stay hidden, got %#v", exposed)
+	}
+}

internal/agent/partition_cache_test.go

@@ -0,0 +1,63 @@
+package agent
+
+import (
+	"context"
+	"reflect"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/tools"
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+// countingSchemaTool records how many times its Parameters() schema is read, so a
+// test can prove the definition cache avoids re-rendering across turns.
+type countingSchemaTool struct {
+	name  string
+	calls *int
+}
+
+func (t countingSchemaTool) Name() string        { return t.name }
+func (t countingSchemaTool) Description() string { return "counts schema reads" }
+func (t countingSchemaTool) Parameters() tools.Schema {
+	*t.calls++
+	return tools.Schema{Type: "object", AdditionalProperties: false, Properties: map[string]tools.PropertySchema{
+		"x": {Type: "string"},
+	}}
+}
+func (t countingSchemaTool) Safety() tools.Safety {
+	return tools.Safety{SideEffect: tools.SideEffectRead, Permission: tools.PermissionAllow}
+}
+func (t countingSchemaTool) Run(_ context.Context, _ map[string]any) tools.Result {
+	return tools.Result{Status: tools.StatusOK}
+}
+
+// The cache renders a tool's schema once and reuses it across calls, and its
+// output is identical to the uncached path.
+func TestPartitionToolsCacheRendersOnceAndMatchesUncached(t *testing.T) {
+	calls := 0
+	registry := tools.NewRegistry()
+	registry.Register(countingSchemaTool{name: "alpha", calls: &calls})
+	registry.Register(countingSchemaTool{name: "beta", calls: &calls})
+
+	uncached, _ := partitionTools(registry, PermissionModeAuto, Options{}, map[string]bool{})
+
+	base := calls
+	cache := map[string]zeroruntime.ToolDefinition{}
+	first, _ := partitionToolsCached(registry, PermissionModeAuto, Options{}, map[string]bool{}, cache)
+	rendersAfterFirst := calls - base
+	second, _ := partitionToolsCached(registry, PermissionModeAuto, Options{}, map[string]bool{}, cache)
+	rendersAfterSecond := calls - base
+
+	if rendersAfterFirst != 2 {
+		t.Fatalf("first cached call should render 2 tools once each, rendered %d times", rendersAfterFirst)
+	}
+	if rendersAfterSecond != 2 {
+		t.Fatalf("second cached call must reuse the cache (no new schema reads), total renders = %d", rendersAfterSecond)
+	}
+	if !reflect.DeepEqual(first, second) {
+		t.Fatal("cached calls must return identical definitions")
+	}
+	if !reflect.DeepEqual(first, uncached) {
+		t.Fatal("cached output must match the uncached partitionTools output")
+	}
+}

internal/agent/plan_staleness_test.go

@@ -0,0 +1,64 @@
+package agent
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+func planUpdateTurn(status string) zeroruntime.CollectedStream {
+	return zeroruntime.CollectedStream{ToolCalls: []zeroruntime.ToolCall{
+		{Name: "update_plan", Arguments: `{"plan":[{"content":"step one","status":"` + status + `"}]}`},
+	}}
+}
+
+// Turn-based staleness: a plan that goes many turns without an update — while
+// items are still pending — draws the stale reminder even though few tool calls
+// have run (the tool-call streak alone would not have tripped).
+func TestPlanReminderFiresOnTurnStaleness(t *testing.T) {
+	state := newGuardState()
+	state.observeTurn(planUpdateTurn("in_progress"))
+
+	// Text-only turns advance the turn counter without adding tool calls.
+	for range stalePlanTurnThreshold {
+		state.observeTurn(zeroruntime.CollectedStream{Text: "still working…"})
+	}
+
+	if state.toolCallsSincePlanUpdate >= staleToolCallThreshold {
+		t.Fatalf("precondition: the tool-call trigger must not fire (%d calls)", state.toolCallsSincePlanUpdate)
+	}
+	if got := state.planReminder(stalePlanTurnThreshold + 2); !strings.Contains(got, planStaleReminderMarker) {
+		t.Fatalf("expected a turn-based stale reminder, got %q", got)
+	}
+}
+
+// A fully-completed plan is NOT stale: no pending items means the turn-based
+// trigger stays quiet even after many turns, so a finished plan isn't nagged.
+func TestPlanReminderSkipsTurnStalenessWhenComplete(t *testing.T) {
+	state := newGuardState()
+	state.observeTurn(planUpdateTurn("completed"))
+
+	for range stalePlanTurnThreshold + 3 {
+		state.observeTurn(zeroruntime.CollectedStream{Text: "wrapping up…"})
+	}
+
+	if got := state.planReminder(stalePlanTurnThreshold + 5); got != "" {
+		t.Fatalf("a completed plan must not draw a stale reminder, got %q", got)
+	}
+}
+
+// A plan update resets the turn counter, so the stale reminder does not fire
+// right after a fresh update.
+func TestPlanUpdateResetsTurnStaleness(t *testing.T) {
+	state := newGuardState()
+	state.observeTurn(planUpdateTurn("in_progress"))
+	for range stalePlanTurnThreshold - 1 {
+		state.observeTurn(zeroruntime.CollectedStream{Text: "working…"})
+	}
+	// Refresh the plan — resets the turn streak.
+	state.observeTurn(planUpdateTurn("in_progress"))
+	if got := state.planReminder(stalePlanTurnThreshold + 1); got != "" {
+		t.Fatalf("a just-refreshed plan must not be stale, got %q", got)
+	}
+}

internal/agent/prompt_budget_test.go

@@ -0,0 +1,51 @@
+package agent
+
+import (
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/tools"
+)
+
+// These ceilings are a deliberate ratchet on Zero's fixed per-turn overhead — the
+// system prompt and the eager tool schemas that ride on EVERY request. They are
+// set ~10% above the current measured cost, using ApproxTextTokens (the same
+// estimate the compaction loop and /context use, ~non-whitespace-bytes/4). A change
+// that pushes either past its ceiling must be justified (and the ceiling raised
+// deliberately) or trimmed — the per-turn floor should not creep up silently.
+//
+// Measured baselines (2026-07): base system prompt ~3160 tokens; the tools a normal
+// (auto permission-mode) interactive turn sends ~2430 tokens. The tool figure is
+// the auto-mode advertised set, not the full core registry — higher-risk tools that
+// only advertise in other modes are excluded, matching what a real turn actually
+// pays.
+const (
+	maxBaseSystemPromptTokens = 3500
+	maxEagerToolSchemaTokens  = 2700
+)
+
+func TestSystemPromptTokenBudget(t *testing.T) {
+	// Minimal render: a model is set (so the session block renders) but no Cwd, so
+	// the workspace map, project guidelines, and repo map are excluded — this is the
+	// fixed base every session pays regardless of the workspace it runs in.
+	prompt := buildSystemPrompt(Options{Model: "claude-opus-4-8"})
+	got := ApproxTextTokens(prompt)
+	t.Logf("base system prompt: %d tokens (%d bytes)", got, len(prompt))
+	if got > maxBaseSystemPromptTokens {
+		t.Fatalf("base system prompt is %d tokens, over the %d ceiling — trim it or raise the ceiling deliberately", got, maxBaseSystemPromptTokens)
+	}
+}
+
+func TestEagerToolSchemaTokenBudget(t *testing.T) {
+	registry := tools.NewRegistry()
+	for _, tool := range tools.CoreToolsScoped(t.TempDir(), nil) {
+		registry.Register(tool)
+	}
+	// Options{} keeps DeferThreshold at 0, so deferral is inactive and every core
+	// tool is exposed eagerly — exactly what a plugin-free session sends each turn.
+	exposed, _ := partitionTools(registry, PermissionModeAuto, Options{}, map[string]bool{})
+	got := estimateToolDefTokens(exposed)
+	t.Logf("eager core tool schemas: %d tokens across %d tools", got, len(exposed))
+	if got > maxEagerToolSchemaTokens {
+		t.Fatalf("eager tool schemas are %d tokens, over the %d ceiling — defer a tool or raise the ceiling deliberately", got, maxEagerToolSchemaTokens)
+	}
+}

internal/agent/reconnect.go

@@ -3,9 +3,11 @@ package agent
 import (
 	"context"
 	"fmt"
+	"math/rand"
 	"strings"
 	"time"
 
+	"github.com/Gitlawb/zero/internal/errhint"
 	"github.com/Gitlawb/zero/internal/zeroruntime"
 )
 
@@ -17,10 +19,20 @@ import (
 // times. We retry ONLY the connect (not a partially-consumed stream), so no
 // already-forwarded OnText is ever duplicated.
 const (
-	maxStreamReconnects = 2
-	streamReconnectBase = 500 * time.Millisecond
+	// maxStreamReconnects is how many times the connect is re-issued after a
+	// transient disconnect. 4 (not 2): with jittered exponential backoff this rides
+	// out a multi-second network blip (~0.5s + 1s + 2s + 4s worst case) instead of
+	// dying on a 2s hiccup and re-burning every token on a restart.
+	maxStreamReconnects = 4
+	// streamReconnectMax caps a single backoff so the tail attempts don't wait
+	// minutes on a long outage.
+	streamReconnectMax = 8 * time.Second
 )
 
+// streamReconnectBase is the first-retry delay (doubled each subsequent attempt).
+// A var, not a const, so tests can shrink it to keep the exhaustion path fast.
+var streamReconnectBase = 500 * time.Millisecond
+
 // reconnectNotifier is called before each retry with the 1-based attempt number
 // and the max, so the caller can surface a "Reconnecting N/max" notice. Nil is
 // fine.
@@ -73,7 +85,7 @@ func streamWithReconnect(ctx context.Context, provider Provider, request zerorun
 		if notify != nil {
 			notify(attempt, maxStreamReconnects)
 		}
-		if waitErr := sleepWithContext(ctx, backoffFor(attempt)); waitErr != nil {
+		if waitErr := sleepWithContext(ctx, jitteredBackoff(attempt)); waitErr != nil {
 			return nil, err // ctx cancelled while waiting; surface the original error
 		}
 		stream, err = provider.StreamCompletion(ctx, request)
@@ -96,6 +108,19 @@ func shouldReconnect(ctx context.Context, err error) bool {
 	if isContextLimitError(msg) || isImageRejectionError(err) {
 		return false
 	}
+	// HTTP 5xx statuses are non-reconnectable and must be excluded BEFORE the
+	// transport substring match below — otherwise "504 Gateway Timeout" would slip
+	// through on the generic "timeout" needle. 503 already exhausted
+	// providerio.SendWithRetry (retrying is a redundant double-retry); 500/502/504
+	// are non-idempotent by providerio's rule — the completion POST may already
+	// have reached the model before the upstream/gateway gave up, so replaying the
+	// connect risks duplicate billable work. Digit-boundary matched so an
+	// incidental "504" inside a latency/id number is not mistaken for a status.
+	if errhint.HasStatusCode(msg, "500", "502", "503", "504") {
+		return false
+	}
+	// Transport-level disconnects only. A genuine transport failure (EOF, reset,
+	// refused, timeout) means no response was received, which is safe to reconnect.
 	for _, needle := range []string{
 		"eof",
 		"connection reset",
@@ -106,8 +131,6 @@ func shouldReconnect(ctx context.Context, err error) bool {
 		"timed out",
 		"temporarily unavailable",
 		"i/o timeout",
-		"503",
-		"502",
 		"server closed",
 		"unexpected end",
 	} {
@@ -118,14 +141,31 @@ func shouldReconnect(ctx context.Context, err error) bool {
 	return false
 }
 
+// backoffFor is the deterministic exponential base delay for a 1-based attempt,
+// capped at streamReconnectMax. Jitter is layered on separately (jitteredBackoff).
 func backoffFor(attempt int) time.Duration {
 	d := streamReconnectBase
 	for i := 1; i < attempt; i++ {
+		if d >= streamReconnectMax {
+			return streamReconnectMax
+		}
 		d *= 2
 	}
+	if d > streamReconnectMax {
+		d = streamReconnectMax
+	}
 	return d
 }
 
+// jitteredBackoff adds up to 50% random jitter on top of backoffFor so concurrent
+// runs (swarm members, a cron fleet) that all trip on the same outage don't
+// reconnect in lockstep and hammer a recovering endpoint. Never shorter than the
+// deterministic base, so backoff still grows attempt over attempt.
+func jitteredBackoff(attempt int) time.Duration {
+	base := backoffFor(attempt)
+	return base + time.Duration(rand.Int63n(int64(base/2)+1))
+}
+
 func sleepWithContext(ctx context.Context, d time.Duration) error {
 	timer := time.NewTimer(d)
 	defer timer.Stop()

internal/agent/reconnect_test.go

@@ -44,6 +44,10 @@ func TestStreamWithReconnectRecoversFromTransientDisconnect(t *testing.T) {
 }
 
 func TestStreamWithReconnectGivesUpAfterMax(t *testing.T) {
+	// Shrink the backoff so exhausting all retries stays fast (real base would
+	// sleep ~7.5s across 4 attempts).
+	defer func(orig time.Duration) { streamReconnectBase = orig }(streamReconnectBase)
+	streamReconnectBase = time.Millisecond
 	// Always fails with a disconnect error → exhausts retries and returns it.
 	p := &flakyProvider{failBefore: 99, failErr: errors.New("connection reset by peer")}
 	_, err := streamWithReconnect(context.Background(), p, zeroruntime.CompletionRequest{}, nil)
@@ -87,7 +91,7 @@ func TestShouldReconnectClassification(t *testing.T) {
 	ctx := context.Background()
 	disconnects := []string{
 		"unexpected EOF", "connection reset by peer", "broken pipe",
-		"i/o timeout", "503 Service Unavailable", "server closed the connection",
+		"i/o timeout", "server closed the connection", "connection refused",
 	}
 	for _, m := range disconnects {
 		if !shouldReconnect(ctx, errors.New(m)) {
@@ -97,6 +101,13 @@ func TestShouldReconnectClassification(t *testing.T) {
 	notDisconnects := []string{
 		"context length exceeded", "invalid api key", "model not found",
 		"400 bad request: unsupported parameter",
+		// HTTP 5xx statuses are handled by providerio.SendWithRetry (503) or are
+		// non-idempotent (500/502/504); the reconnect path must not double-retry
+		// them. "504 Gateway Timeout" must NOT slip through on the generic "timeout"
+		// transport signal — a gateway timeout is non-idempotent, since the upstream
+		// may have already processed the completion POST before giving up.
+		"503 Service Unavailable", "502 Bad Gateway",
+		"504 Gateway Timeout", "500 Internal Server Error",
 	}
 	for _, m := range notDisconnects {
 		if shouldReconnect(ctx, errors.New(m)) {
@@ -112,6 +123,24 @@ func TestBackoffGrows(t *testing.T) {
 	if backoffFor(2) != 2*streamReconnectBase {
 		t.Fatalf("attempt 2 backoff = %v, want %v", backoffFor(2), 2*streamReconnectBase)
 	}
+	// The exponential base is capped so late attempts don't wait minutes.
+	if got := backoffFor(20); got != streamReconnectMax {
+		t.Fatalf("attempt 20 backoff = %v, want cap %v", got, streamReconnectMax)
+	}
+}
+
+func TestJitteredBackoffStaysInBounds(t *testing.T) {
+	// Jitter never drops below the deterministic base and never exceeds base*1.5,
+	// so backoff still grows attempt over attempt while decorrelating retries.
+	for attempt := 1; attempt <= 5; attempt++ {
+		base := backoffFor(attempt)
+		for i := 0; i < 200; i++ {
+			got := jitteredBackoff(attempt)
+			if got < base || got > base+base/2 {
+				t.Fatalf("attempt %d jittered backoff %v out of [%v, %v]", attempt, got, base, base+base/2)
+			}
+		}
+	}
 }
 
 func TestReconnectNoticeRoutesThroughReasoning(t *testing.T) {

internal/agent/skills_context_test.go

@@ -0,0 +1,63 @@
+package agent
+
+import (
+	"strconv"
+	"strings"
+	"testing"
+)
+
+func TestSkillsContextCapsLongList(t *testing.T) {
+	skills := make([]SkillInfo, 0, 40)
+	for i := 0; i < 40; i++ {
+		n := strconv.Itoa(i)
+		skills = append(skills, SkillInfo{Name: "skill-" + n, Description: "does something useful, number " + n})
+	}
+	got := skillsContext(Options{Skills: skills})
+	if len(got) > 1200 {
+		t.Fatalf("skills block should stay bounded, got %d bytes:\n%s", len(got), got)
+	}
+	if !strings.Contains(got, "more (call skill") {
+		t.Fatalf("expected an overflow summary line, got:\n%s", got)
+	}
+	// The first skill is always listed regardless of budget.
+	if !strings.Contains(got, "- skill-0:") {
+		t.Fatalf("expected the first skill to always be listed, got:\n%s", got)
+	}
+}
+
+func TestSkillsContext(t *testing.T) {
+	if got := skillsContext(Options{}); got != "" {
+		t.Fatalf("no skills should yield an empty section, got %q", got)
+	}
+	got := skillsContext(Options{Skills: []SkillInfo{
+		{Name: "commit-writer", Description: "Write a conventional-commit message."},
+		{Name: "  ", Description: "nameless, should be skipped"},
+		{Name: "reviewer"},
+	}})
+	if !strings.Contains(got, "<available_skills>") || !strings.Contains(got, "</available_skills>") {
+		t.Fatalf("missing available_skills block: %q", got)
+	}
+	if !strings.Contains(got, "- commit-writer: Write a conventional-commit message.") {
+		t.Fatalf("missing commit-writer line: %q", got)
+	}
+	if !strings.Contains(got, "- reviewer\n") {
+		t.Fatalf("reviewer (no description) line missing: %q", got)
+	}
+	if strings.Contains(got, "nameless") {
+		t.Fatalf("nameless entry should be skipped: %q", got)
+	}
+}
+
+func TestSystemPromptIncludesSkillsOnlyWhenInstalled(t *testing.T) {
+	with := buildSystemPrompt(Options{Skills: []SkillInfo{
+		{Name: "commit-writer", Description: "Write a commit message."},
+	}})
+	if !strings.Contains(with, "<available_skills>") || !strings.Contains(with, "skill tool") {
+		t.Fatalf("expected available_skills guidance in system prompt: %q", with)
+	}
+	// Default (no skills) must reproduce the prior prompt: no skills block.
+	without := buildSystemPrompt(Options{})
+	if strings.Contains(without, "<available_skills>") {
+		t.Fatalf("available_skills block must not appear without skills")
+	}
+}

internal/agent/system_prompt.go

@@ -5,6 +5,7 @@ import (
 	"os"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"strings"
 	"unicode/utf8"
 
@@ -86,6 +87,9 @@ func buildSystemPrompt(options Options) string {
 	if delegation := specialistDelegationContext(options); delegation != "" {
 		sections = append(sections, delegation)
 	}
+	if skillsBlock := skillsContext(options); skillsBlock != "" {
+		sections = append(sections, skillsBlock)
+	}
 	if style := responseStyleContext(options); style != "" {
 		sections = append(sections, style)
 	}
@@ -164,6 +168,64 @@ func specialistDelegationContext(options Options) string {
 	return b.String()
 }
 
+// skillsContext lists the reusable skills the model can pull in on demand via the
+// skill tool, so it invokes the right one on the first try instead of guessing a
+// name, failing, and reading the error. It mirrors specialistDelegationContext:
+// names + one-line descriptions only (never skill bodies), and it renders nothing
+// when no skills are installed, so a skill-less run reproduces the previous prompt
+// byte-for-byte.
+// skillsContextListBudget bounds the bytes spent listing individual skills so a
+// workspace with dozens of them can't bloat every turn's prompt. Skills past the
+// budget are summarized as a count rather than dropped — the model can still load
+// any of them by name (an unknown name returns the full list from the skill tool).
+const skillsContextListBudget = 640
+
+func skillsContext(options Options) string {
+	if len(options.Skills) == 0 {
+		return ""
+	}
+	var b strings.Builder
+	b.WriteString("<available_skills>\n")
+	b.WriteString("Reusable, on-demand instruction sets you can load with the skill tool. When a request matches one, call skill with its exact name to pull in its guidance before acting — do not guess names or skip a relevant skill.\n")
+	listed, spent, omitted := 0, 0, 0
+	for _, info := range options.Skills {
+		name := strings.TrimSpace(info.Name)
+		if name == "" {
+			continue
+		}
+		line := "- " + name
+		if desc := strings.TrimSpace(info.Description); desc != "" {
+			line += ": " + truncateForSkillLine(desc)
+		}
+		line += "\n"
+		// Always list at least one skill; past the budget, summarize the remainder
+		// as a count instead of silently dropping it.
+		if listed > 0 && spent+len(line) > skillsContextListBudget {
+			omitted++
+			continue
+		}
+		b.WriteString(line)
+		spent += len(line)
+		listed++
+	}
+	if omitted > 0 {
+		b.WriteString("- …and " + strconv.Itoa(omitted) + " more (call skill with a name; an unknown name lists them all)\n")
+	}
+	b.WriteString("</available_skills>")
+	return b.String()
+}
+
+// truncateForSkillLine keeps a skill's one-line description short so a single
+// verbose description can't dominate the skills-list budget.
+func truncateForSkillLine(desc string) string {
+	const maxDescRunes = 100
+	runes := []rune(desc)
+	if len(runes) <= maxDescRunes {
+		return desc
+	}
+	return strings.TrimSpace(string(runes[:maxDescRunes])) + "…"
+}
+
 func sessionRuntimeContext(options Options) string {
 	provider := strings.TrimSpace(options.ProviderName)
 	model := strings.TrimSpace(options.Model)
@@ -310,9 +372,9 @@ func projectGuidelineLabel(match, gitRoot string) string {
 
 // findProjectContextFile returns the first matching project guideline file
 // in dir. Lookup is case-insensitive on the basename and on the actual
-// filename returned, so a git-tracked file like AGENTS.MD (uppercase MD)
-// still resolves to its true filename on case-sensitive filesystems — the
-// returned path is what should appear in the project guidelines label.
+// filename returned, so a git-tracked file like AGENTS.md still resolves to
+// its true filename on case-sensitive filesystems — the returned path is what
+// should appear in the project guidelines label.
 // Returns "" when nothing matches.
 func findProjectContextFile(dir string) string {
 	for _, name := range projectContextFiles {

internal/agent/system_prompt.md

@@ -46,14 +46,10 @@ work.
    formatting-only edits unless the user asked for them.
 4. **Verify.** Verify after edits; see the testing gate below. This is
    mandatory.
-5. **Summarize.** For a substantial task, close with an ELABORATE, well-structured
-   summary — never a single throwaway line. Use short headings and bullets: a
-   one-line "what I built/changed"; **Where it lives** — the key files and how they
-   fit together; **How each requirement is met** — walk through each requirement
-   from the request and name the specifics that satisfy it; what validation ran and
-   what was NOT verified; and a brief "Next steps" or "Want me to…?" offer. Scale
-   the depth to the work — a big build earns several sections, a trivial fix earns
-   a line or two.
+5. **Summarize.** Close with a summary scaled to the work: a trivial fix earns one
+   line; a substantial change earns a few short bullets covering what changed, where
+   it lives, and what you did and did not verify. Lead with the outcome, and never
+   pad — do not restate the task or narrate steps the user already watched you take.
 
 ## Editing discipline
 
@@ -64,7 +60,9 @@ work.
 - Make one tool call per file. Do not batch multi-file writes into a single
   shell or script invocation.
 - For edits to existing files, prefer edit_file or apply_patch with minimal,
-  targeted diffs. Match the existing indentation, imports, and idioms.
+  targeted diffs. Match the existing indentation, imports, and idioms. Match the
+  file's comment density: do not add explanatory comments unless the user asks or
+  the code is already comment-dense.
 - Preserve behavior you were not asked to change. Do not delete or rewrite code
   you did not author unless the task requires it; if you must, say so.
 

internal/agent/system_prompt_models.go

@@ -6,11 +6,13 @@ import "strings"
 //
 // Different model families respond to different framing: the GPT family benefits
 // from an explicit markdown/output spec and a strong "prefer native tools" nudge;
-// Gemini benefits from explicit tool-preference and conciseness guidance; the
-// Claude family is already well-aligned with the core prompt and only needs a
-// light comment-discipline nudge. modelPromptAddendum returns the family-specific
-// block appended after the core prompt. Classification is by model id (the agent
-// only has the model string — no registry dependency).
+// Gemini benefits from explicit tool-preference and conciseness guidance. The
+// Claude family is already aligned with the core prompt and needs no addendum —
+// comment discipline is universal in the core prompt now, so every family gets it.
+// modelPromptAddendum returns the family-specific block appended after the core
+// prompt, or "" when the family is unknown or needs no specialization.
+// Classification is by model id (the agent only has the model string — no registry
+// dependency).
 
 const (
 	familyOpenAI    = "openai"
@@ -45,9 +47,10 @@ func modelPromptAddendum(model string) string {
 		return openAIPromptAddendum
 	case familyGemini:
 		return geminiPromptAddendum
-	case familyAnthropic:
-		return anthropicPromptAddendum
 	default:
+		// Anthropic (aligned with the core prompt) and unknown families get no
+		// family-specific block; comment discipline now lives in the core prompt
+		// for every model.
 		return ""
 	}
 }
@@ -70,8 +73,3 @@ const geminiPromptAddendum = `<model_guidance>
   in one short clause why it is needed.
 - Use update_plan for any multi-step task and keep it current.
 </model_guidance>`
-
-const anthropicPromptAddendum = `<model_guidance>
-- Do not add code comments unless the user asks or the surrounding code is
-  comment-dense; match the file's existing comment density.
-</model_guidance>`

internal/agent/system_prompt_models_test.go

@@ -29,9 +29,11 @@ func TestBuildSystemPromptAppendsModelAddendum(t *testing.T) {
 	if got := buildSystemPrompt(Options{Model: "gpt-5"}); !strings.Contains(got, openAIPromptAddendum) {
 		t.Fatalf("expected the OpenAI addendum in the gpt-5 prompt")
 	}
+	// Claude is aligned with the core prompt and gets no family addendum now that
+	// comment discipline is universal; it must not pick up another family's block.
 	claude := buildSystemPrompt(Options{Model: "claude-opus-4-6"})
-	if !strings.Contains(claude, anthropicPromptAddendum) {
-		t.Fatalf("expected the Anthropic addendum in the claude prompt")
+	if strings.Contains(claude, "<model_guidance>") {
+		t.Fatalf("expected no model_guidance block for Claude (aligned with core prompt)")
 	}
 	if strings.Contains(claude, openAIPromptAddendum) {
 		t.Fatalf("the claude prompt must not contain the OpenAI addendum")

internal/agent/system_prompt_test.go

@@ -23,7 +23,8 @@ func TestCoreSystemPromptIncludesCodingQualityRules(t *testing.T) {
 		"avoid broad refactors",
 		"search the web before answering",
 		"do not recognize",
-		"how each requirement is met",
+		"scaled to the work",
+		"comment density",
 	} {
 		if !strings.Contains(prompt, want) {
 			t.Fatalf("expected core system prompt to include %q, got:\n%s", want, buildSystemPrompt(Options{}))
@@ -128,7 +129,7 @@ func systemPromptTestBlock(t *testing.T, prompt, start, end string) string {
 }
 
 func TestBuildSystemPromptInjectsProjectGuidelinesCaseInsensitive(t *testing.T) {
-	// Git tracks AGENTS.MD (uppercase MD) on a case-sensitive filesystem; the
+	// Git tracks AGENTS.md on a case-sensitive filesystem; the
 	// loader must still resolve it when the cwd lookup uses lowercase.
 	cwd := t.TempDir()
 	if err := os.WriteFile(filepath.Join(cwd, "AGENTS.MD"), []byte("Always run `make lint`."), 0o644); err != nil {

internal/agent/types.go

@@ -174,6 +174,16 @@ type SpecialistInfo struct {
 	WhenToUse string
 }
 
+// SkillInfo is a one-line summary of a reusable, on-demand skill (its name and
+// frontmatter description) surfaced to the system prompt so the model can invoke
+// the right skill with the skill tool on the first try instead of guessing a name
+// and reading the failure. Like SpecialistInfo it is plain data, so the agent
+// package needs no dependency on internal/skills.
+type SkillInfo struct {
+	Name        string
+	Description string
+}
+
 type Options struct {
 	MaxTurns int
 	// DeferThreshold activates deferred MCP-tool loading: when the number of
@@ -189,6 +199,12 @@ type Options struct {
 	// populated only where the Task tool is actually registered, so an empty slice
 	// (the default) reproduces the previous prompt byte-for-byte.
 	Specialists []SpecialistInfo
+	// Skills lists the reusable skills installed for this run (the default skills
+	// dir merged with any plugin skill roots). When non-empty the system prompt
+	// gains an <available_skills> block naming them so the model loads the right one
+	// via the skill tool on the first try. Empty (the default) reproduces the
+	// previous prompt byte-for-byte.
+	Skills []SkillInfo
 	// Specialist/sub-agent metadata is carried through exec now and consumed by
 	// the specialist runtime in later slices.
 	SessionID        string
@@ -269,6 +285,12 @@ type Options struct {
 	// ceiling and the autonomy gate. nil disables it entirely (the loop is
 	// byte-identical to before). One instance per run — it holds attempt state.
 	SelfCorrect *SelfCorrector
+	// FileDiagnostics, when set, is passed to mutating tools so edit_file /
+	// write_file append error-severity language diagnostics for the file they
+	// just wrote directly to their tool output — the model sees an error it
+	// introduced in the same turn instead of a later verification pass. Build
+	// one with NewFileDiagnostics. nil disables inline diagnostics.
+	FileDiagnostics func(ctx context.Context, absPath string) string
 
 	// RequireCompletionSignal gates run completion for HEADLESS exec. Without it,
 	// any assistant turn that produces text but no tool call is accepted as the

internal/cli/app.go

@@ -19,6 +19,7 @@ import (
 	"github.com/Gitlawb/zero/internal/hooks"
 	"github.com/Gitlawb/zero/internal/localcontrol"
 	"github.com/Gitlawb/zero/internal/mcp"
+	"github.com/Gitlawb/zero/internal/modelregistry"
 	"github.com/Gitlawb/zero/internal/observability"
 	"github.com/Gitlawb/zero/internal/plugins"
 	"github.com/Gitlawb/zero/internal/providerhealth"
@@ -80,6 +81,7 @@ type appDeps struct {
 	runTUI                 func(context.Context, tui.Options) int
 	runEditor              func(string) error
 	checkUpdate            func(context.Context, update.Options) (update.Result, error)
+	applyUpdate            func(context.Context, update.Options) (update.ApplyResult, error)
 	now                    func() time.Time
 }
 
@@ -180,6 +182,7 @@ func defaultAppDeps() appDeps {
 		runTUI:           tui.Run,
 		runEditor:        openEditor,
 		checkUpdate:      update.Check,
+		applyUpdate:      update.Apply,
 		now:              time.Now,
 	}
 }
@@ -227,6 +230,12 @@ func runWithDeps(args []string, stdout io.Writer, stderr io.Writer, deps appDeps
 	defer observability.Recover(observability.DefaultCrashDir(), "cli", stderr, &exitCode)
 	deps = fillAppDeps(deps)
 
+	// CLI runs opt into the models.dev overlay (cached live context limits and
+	// pricing on top of the curated catalog). Explicitly enabled here — and only
+	// here — so library consumers and hermetic tests are never perturbed by a
+	// cache file on the machine. The refresh itself is fired in exec/TUI startup.
+	modelregistry.EnableModelsDevOverlay()
+
 	addDirs, args, err := splitLeadingAddDirFlags(args)
 	if err != nil {
 		return writeAppError(stderr, err.Error(), 1)
@@ -383,6 +392,8 @@ func runWithDeps(args []string, stdout io.Writer, stderr io.Writer, deps appDeps
 		return runSandbox(args[1:], stdout, stderr, deps)
 	case "update":
 		return runUpdate(args[1:], stdout, stderr, deps)
+	case "upgrade":
+		return runUpgrade(args[1:], stdout, stderr, deps)
 	case "worktrees", "worktree":
 		return runWorktrees(args[1:], stdout, stderr, deps)
 	case "verify":
@@ -515,6 +526,9 @@ func fillAppDeps(deps appDeps) appDeps {
 	if deps.checkUpdate == nil {
 		deps.checkUpdate = defaults.checkUpdate
 	}
+	if deps.applyUpdate == nil {
+		deps.applyUpdate = defaults.applyUpdate
+	}
 	if deps.now == nil {
 		deps.now = defaults.now
 	}
@@ -537,6 +551,11 @@ func runInteractiveTUI(stderr io.Writer, deps appDeps, permissionMode agent.Perm
 }
 
 func runInteractiveTUIWithSetup(stderr io.Writer, deps appDeps, permissionMode agent.PermissionMode, addDirs []string, theme string, forceSetup bool) int {
+	// Refresh the models.dev pricing/limits cache in the background when stale;
+	// the overlay is read at registry construction from the cache file, so this
+	// benefits the next run and never blocks or fails this one.
+	go func() { _ = modelregistry.RefreshModelsDevCache(context.Background()) }()
+
 	workspaceRoot, err := deps.getwd()
 	if err != nil {
 		return writeAppError(stderr, "failed to resolve workspace: "+err.Error(), 1)
@@ -557,8 +576,19 @@ func runInteractiveTUIWithSetup(stderr io.Writer, deps appDeps, permissionMode a
 		if !errors.Is(err, config.ErrNoActiveProvider) && !errors.Is(err, config.ErrProviderRequiresModel) {
 			return writeAppError(stderr, err.Error(), 1)
 		}
-		resolved = config.ResolvedConfig{}
-		forceSetup = true
+		// ErrNoActiveProvider can mean "nothing configured yet" (needs onboarding)
+		// or "providers ARE configured, just none marked active" (e.g. config.json's
+		// activeProvider is blank/stale). In the second case resolved.Providers still
+		// carries the already-normalized list — prefer falling back to one of those
+		// over wiping everything and forcing the user to re-enter credentials they
+		// already saved.
+		if usable, ok := firstUsableProvider(resolved.Providers); errors.Is(err, config.ErrNoActiveProvider) && ok {
+			resolved.Provider = usable
+			resolved.ActiveProvider = usable.Name
+		} else {
+			resolved = config.ResolvedConfig{}
+			forceSetup = true
+		}
 	}
 	userConfigPath, err := deps.userConfigPath()
 	if err != nil {
@@ -740,9 +770,11 @@ func runInteractiveTUIWithSetup(stderr io.Writer, deps appDeps, permissionMode a
 			Hooks:          newHookDispatcherWithExtra(workspaceRoot, pluginActivation.hooks),
 			DeferThreshold: resolved.Tools.DeferThreshold,
 			Specialists:    specialistRuntime.specialists,
+			Skills:         pluginActivation.skillInfos(deps.skillsDir()),
 		},
 		PermissionMode: permissionMode,
 		Notify:         resolved.Notify,
+		KeyBindings:    resolved.KeyBindings,
 		Setup: tui.SetupOptions{
 			Visible:    setupVisible,
 			Required:   needsSetup,
@@ -1158,7 +1190,8 @@ func profileHasCredential(profile config.ProviderProfile) bool {
 }
 
 // baseURLIsLoopback reports whether a provider base_url points at a loopback host
-// (a keyless local provider like Ollama/LM Studio), which needs no API key.
+// or a private-network host (192.168.x.y / 10.x.y.z / 172.16.x.y) — a local
+// provider like Ollama/LM Studio that needs no API key.
 func baseURLIsLoopback(baseURL string) bool {
 	u := strings.TrimSpace(baseURL)
 	if u == "" {
@@ -1173,7 +1206,7 @@ func baseURLIsLoopback(baseURL string) bool {
 		return true
 	}
 	if addr, err := netip.ParseAddr(host); err == nil {
-		return addr.IsLoopback()
+		return addr.IsLoopback() || addr.IsPrivate()
 	}
 	return false
 }

internal/cli/app_test.go

@@ -205,6 +205,67 @@ func TestRunNoArgsEntersSetupWhenResolveReportsNoActiveProvider(t *testing.T) {
 	}
 }
 
+func TestRunNoArgsFallsBackToUsableProviderWhenNoneMarkedActive(t *testing.T) {
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+	cwd := t.TempDir()
+	setCLIUserConfigRoot(t)
+	userConfigPath := filepath.Join(t.TempDir(), "zero", "config.json")
+	var launchedOptions tui.Options
+	launched := false
+	var providerProfile config.ProviderProfile
+	fake := &cliFakeProvider{}
+
+	usable := config.ProviderProfile{
+		Name:         "work",
+		ProviderKind: config.ProviderKindOpenAI,
+		BaseURL:      config.OpenAIBaseURL,
+		APIKey:       "sk-test",
+		Model:        "gpt-test",
+	}
+
+	exitCode := runWithDeps([]string{}, &stdout, &stderr, appDeps{
+		getwd: func() (string, error) {
+			return cwd, nil
+		},
+		resolveConfig: func(workspaceRoot string, overrides config.Overrides) (config.ResolvedConfig, error) {
+			// config.json has providers configured but none marked active (e.g. a
+			// blank/stale activeProvider field) — Resolve returns the successfully
+			// normalized providers list alongside ErrNoActiveProvider.
+			return config.ResolvedConfig{Providers: []config.ProviderProfile{usable}},
+				fmt.Errorf("%w: active provider %q not found", config.ErrNoActiveProvider, "")
+		},
+		newProvider: func(profile config.ProviderProfile) (zeroruntime.Provider, error) {
+			providerProfile = profile
+			return fake, nil
+		},
+		userConfigPath: func() (string, error) {
+			return userConfigPath, nil
+		},
+		registerMCPTools: func(context.Context, *tools.Registry, config.MCPConfig, mcp.RegisterOptions) (mcpToolRuntime, error) {
+			return noopMCPRuntime{}, nil
+		},
+		runTUI: func(ctx context.Context, options tui.Options) int {
+			launched = true
+			launchedOptions = options
+			return 0
+		},
+	})
+
+	if exitCode != 0 {
+		t.Fatalf("exit code = %d, want 0, stderr=%q", exitCode, stderr.String())
+	}
+	if !launched {
+		t.Fatal("TUI was not launched")
+	}
+	if launchedOptions.Setup.Visible || launchedOptions.Setup.Required {
+		t.Fatalf("Setup = %#v, want no setup wizard — a usable saved provider exists", launchedOptions.Setup)
+	}
+	if providerProfile.Name != "work" {
+		t.Fatalf("provider used = %q, want fallback to the usable saved provider %q", providerProfile.Name, "work")
+	}
+}
+
 func TestRunNoArgsFailsWhenResolveErrorIsNotProviderRelated(t *testing.T) {
 	var stdout bytes.Buffer
 	var stderr bytes.Buffer
@@ -1267,6 +1328,153 @@ func TestRunUpdateReportsUpToDate(t *testing.T) {
 	}
 }
 
+func TestRunUpdateApplyTextAndJSON(t *testing.T) {
+	applyResult := update.ApplyResult{
+		Result: update.Result{
+			CurrentVersion:  "dev",
+			LatestVersion:   "0.2.0",
+			UpdateAvailable: true,
+		},
+		Applied:       true,
+		InstallMethod: update.InstallMethodStandalone,
+		BinaryPath:    "/usr/local/bin/zero",
+		Message:       "updated to 0.2.0",
+	}
+	var gotOptions update.Options
+	deps := appDeps{
+		applyUpdate: func(_ context.Context, options update.Options) (update.ApplyResult, error) {
+			gotOptions = options
+			return applyResult, nil
+		},
+		checkUpdate: func(context.Context, update.Options) (update.Result, error) {
+			t.Fatal("checkUpdate should not run for --apply")
+			return update.Result{}, nil
+		},
+	}
+
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+	exitCode := runWithDeps([]string{"update", "--apply"}, &stdout, &stderr, deps)
+	if exitCode != exitSuccess {
+		t.Fatalf("expected exit code %d, got %d: %s", exitSuccess, exitCode, stderr.String())
+	}
+	if !strings.Contains(stdout.String(), "updated to 0.2.0") {
+		t.Fatalf("unexpected apply text: %q", stdout.String())
+	}
+	if gotOptions.CurrentVersion != "dev" {
+		t.Fatalf("unexpected options passed to applyUpdate: %#v", gotOptions)
+	}
+
+	stdout.Reset()
+	stderr.Reset()
+	exitCode = runWithDeps([]string{"update", "--apply", "--json"}, &stdout, &stderr, deps)
+	if exitCode != exitSuccess {
+		t.Fatalf("expected exit code %d, got %d: %s", exitSuccess, exitCode, stderr.String())
+	}
+	var payload struct {
+		Applied       bool   `json:"applied"`
+		InstallMethod string `json:"installMethod"`
+		BinaryPath    string `json:"binaryPath"`
+	}
+	if err := json.Unmarshal(stdout.Bytes(), &payload); err != nil {
+		t.Fatalf("apply JSON did not decode: %v\n%s", err, stdout.String())
+	}
+	if !payload.Applied || payload.InstallMethod != string(update.InstallMethodStandalone) || payload.BinaryPath != applyResult.BinaryPath {
+		t.Fatalf("unexpected apply JSON: %#v", payload)
+	}
+}
+
+func TestRunUpgradeDefaultsToApply(t *testing.T) {
+	var applyCalled bool
+	deps := appDeps{
+		applyUpdate: func(context.Context, update.Options) (update.ApplyResult, error) {
+			applyCalled = true
+			return update.ApplyResult{Message: "already up to date"}, nil
+		},
+		checkUpdate: func(context.Context, update.Options) (update.Result, error) {
+			t.Fatal("checkUpdate should not run for `zero upgrade`")
+			return update.Result{}, nil
+		},
+	}
+
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+	exitCode := runWithDeps([]string{"upgrade"}, &stdout, &stderr, deps)
+	if exitCode != exitSuccess {
+		t.Fatalf("expected exit code %d, got %d: %s", exitSuccess, exitCode, stderr.String())
+	}
+	if !applyCalled {
+		t.Fatal("expected `zero upgrade` to call applyUpdate")
+	}
+}
+
+func TestRunUpdateRejectsCheckAndApplyTogether(t *testing.T) {
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	exitCode := runWithDeps([]string{"update", "--check", "--apply"}, &stdout, &stderr, appDeps{
+		checkUpdate: func(context.Context, update.Options) (update.Result, error) {
+			t.Fatal("checkUpdate should not run")
+			return update.Result{}, nil
+		},
+		applyUpdate: func(context.Context, update.Options) (update.ApplyResult, error) {
+			t.Fatal("applyUpdate should not run")
+			return update.ApplyResult{}, nil
+		},
+	})
+
+	if exitCode != exitUsage {
+		t.Fatalf("expected usage exit code, got %d", exitCode)
+	}
+	if got := stderr.String(); !strings.Contains(got, "only one of --check or --apply") {
+		t.Fatalf("expected --check/--apply usage error, got %q", got)
+	}
+}
+
+func TestRunUpdateRejectsTargetWithApply(t *testing.T) {
+	for _, args := range [][]string{
+		{"update", "--apply", "--target", "linux-arm64"},
+		{"upgrade", "--target", "linux-arm64"},
+	} {
+		t.Run(strings.Join(args, " "), func(t *testing.T) {
+			var stdout bytes.Buffer
+			var stderr bytes.Buffer
+
+			exitCode := runWithDeps(args, &stdout, &stderr, appDeps{
+				applyUpdate: func(context.Context, update.Options) (update.ApplyResult, error) {
+					t.Fatal("applyUpdate should not run when --target is combined with --apply")
+					return update.ApplyResult{}, nil
+				},
+			})
+
+			if exitCode != exitUsage {
+				t.Fatalf("expected usage exit code, got %d", exitCode)
+			}
+			if got := stderr.String(); !strings.Contains(got, "--target cannot be combined with --apply") {
+				t.Fatalf("expected --target/--apply usage error, got %q", got)
+			}
+		})
+	}
+}
+
+func TestRunUpdateReportsApplyError(t *testing.T) {
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	exitCode := runWithDeps([]string{"update", "--apply"}, &stdout, &stderr, appDeps{
+		applyUpdate: func(context.Context, update.Options) (update.ApplyResult, error) {
+			return update.ApplyResult{}, errors.New("download failed")
+		},
+	})
+
+	if exitCode == exitSuccess {
+		t.Fatalf("expected non-success exit code, got %d", exitCode)
+	}
+	if got := stderr.String(); !strings.Contains(got, "download failed") {
+		t.Fatalf("expected apply error, got %q", got)
+	}
+}
+
 func assertHelpOutput(t *testing.T, args []string) {
 	t.Helper()
 

internal/cli/auth.go

@@ -301,12 +301,27 @@ func runAuthLogin(args []string, stdout io.Writer, stderr io.Writer, deps appDep
 	if len(parsed.positional) != 1 {
 		return writeExecUsageError(stderr, "usage: zero auth login <provider> [--device] [--scope <scope>]")
 	}
+	provider := parsed.positional[0]
+	// ChatGPT (Codex) requires a fixed redirect_uri (http://localhost:1455/
+	// auth/callback) and mandatory authorize params (id_token_add_organizations,
+	// codex_cli_simplified_flow, originator) that the generic loopback flow
+	// cannot supply. Route it to the dedicated ChatGPT login so
+	// `zero auth login chatgpt` behaves identically to `zero auth chatgpt`.
+	if strings.EqualFold(provider, "chatgpt") {
+		if parsed.device {
+			return writeExecUsageError(stderr, "ChatGPT login does not support --device (it is loopback-only)")
+		}
+		if len(parsed.scopes) > 0 {
+			return writeExecUsageError(stderr, "ChatGPT login does not support --scope (the required scopes are fixed by the Codex client registration)")
+		}
+		return runAuthChatGPT(nil, stdout, stderr, deps)
+	}
 	manager, err := newAuthManager(deps, stdout)
 	if err != nil {
 		return writeAppError(stderr, redaction.ErrorMessage(err, redaction.Options{}), exitCrash)
 	}
 	status, err := manager.Login(context.Background(), oauth.LoginOptions{
-		Provider:    parsed.positional[0],
+		Provider:    provider,
 		Device:      parsed.device,
 		ExtraScopes: parsed.scopes,
 	})
@@ -489,8 +504,9 @@ Commands:
 A provider is any OAuth 2.0 / OIDC server. "openrouter" ('zero auth openrouter')
 works out of the box. "xai" ('zero auth login xai') uses a built-in preset that is
 off by default — enable it with ZERO_OAUTH_ALLOW_PRESETS=1, or set the
-ZERO_OAUTH_XAI_* vars yourself. Any preset field is overridable via the env vars
-below. For a custom provider named <name>, set:
+ZERO_OAUTH_XAI_* vars yourself. "chatgpt" ('zero auth login chatgpt' or
+'zero auth chatgpt') uses a fixed-port loopback flow against the Codex backend.
+Any preset field is overridable via the env vars below. For a custom provider named <name>, set:
   ZERO_OAUTH_<NAME>_CLIENT_ID       (required)
   ZERO_OAUTH_<NAME>_CLIENT_SECRET   (optional)
   ZERO_OAUTH_<NAME>_AUTHORIZE_URL   ZERO_OAUTH_<NAME>_TOKEN_URL

internal/cli/auth_test.go

@@ -156,3 +156,44 @@ func TestRunAuthHelp(t *testing.T) {
 		}
 	}
 }
+
+// TestRunAuthLoginChatGPTRoutesToDedicatedFlow verifies `zero auth login
+// chatgpt` reaches the dedicated ChatGPT login (fixed-port loopback + mandatory
+// authorize params), not the generic manager path. The generic login accepts
+// --device, so a ChatGPT-specific rejection proves the routing took effect.
+// See issue #430: the generic path built a random-port 127.0.0.1 redirect_uri
+// without the required extra params, so OpenAI's authorize endpoint rejected it.
+func TestRunAuthLoginChatGPTRoutesToDedicatedFlow(t *testing.T) {
+	withAuthStore(t)
+	var stdout, stderr bytes.Buffer
+	if code := runWithDeps([]string{"auth", "login", "chatgpt", "--device"}, &stdout, &stderr, appDeps{}); code == exitSuccess {
+		t.Fatal("auth login chatgpt --device should be rejected (ChatGPT is loopback-only)")
+	}
+	if !strings.Contains(stderr.String(), "ChatGPT login does not support --device") {
+		t.Fatalf("stderr = %q, want the ChatGPT-specific --device rejection", stderr.String())
+	}
+	// Case-insensitive provider name should still route.
+	stdout.Reset()
+	stderr.Reset()
+	if code := runWithDeps([]string{"auth", "login", "ChatGPT", "--device"}, &stdout, &stderr, appDeps{}); code == exitSuccess {
+		t.Fatal("auth login ChatGPT --device should be rejected")
+	}
+	if !strings.Contains(stderr.String(), "ChatGPT login does not support --device") {
+		t.Fatalf("stderr = %q, want the ChatGPT-specific rejection (case-insensitive)", stderr.String())
+	}
+}
+
+// TestRunAuthLoginChatGPTRejectsScope mirrors the --device rejection: --scope
+// must not be silently dropped on the ChatGPT path. The Codex client
+// registration pins a fixed scope set (incl. api.connectors.*), so custom
+// scopes are rejected up front rather than plumbed through.
+func TestRunAuthLoginChatGPTRejectsScope(t *testing.T) {
+	withAuthStore(t)
+	var stdout, stderr bytes.Buffer
+	if code := runWithDeps([]string{"auth", "login", "chatgpt", "--scope", "custom-scope"}, &stdout, &stderr, appDeps{}); code == exitSuccess {
+		t.Fatal("auth login chatgpt --scope should be rejected")
+	}
+	if !strings.Contains(stderr.String(), "ChatGPT login does not support --scope") {
+		t.Fatalf("stderr = %q, want the ChatGPT-specific --scope rejection", stderr.String())
+	}
+}

internal/cli/exec.go

@@ -12,6 +12,7 @@ import (
 
 	"github.com/Gitlawb/zero/internal/agent"
 	"github.com/Gitlawb/zero/internal/config"
+	"github.com/Gitlawb/zero/internal/errhint"
 	"github.com/Gitlawb/zero/internal/imageinput"
 	"github.com/Gitlawb/zero/internal/lsp"
 	"github.com/Gitlawb/zero/internal/modelregistry"
@@ -135,6 +136,11 @@ func runExec(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) in
 		return exitSuccess
 	}
 
+	// Refresh the models.dev pricing/limits cache in the background when stale;
+	// the overlay is read at registry construction from the cache file, so this
+	// benefits the next run and never blocks or fails this one.
+	go func() { _ = modelregistry.RefreshModelsDevCache(context.Background()) }()
+
 	// A mode seeds model/effort/max-turns/tool filters as a preset. Expand it up
 	// front — before tool-filter validation and the --list-tools branch — so a
 	// mode-injected tool filter is validated and reflected in --list-tools, and a
@@ -499,15 +505,17 @@ func runExec(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) in
 	// by default the corrector is nil, leaving the agent loop byte-identical. When
 	// on we verify with both the workspace test plan and LSP diagnostics over the
 	// changed files; the autonomy gate inside the corrector still decides whether
-	// failures auto-fix or just report. lspShutdown tears down any language-server
-	// sessions the LSP half spawned (no-op when self-correct is off).
-	selfCorrector, lspShutdown := newExecSelfCorrector(options.selfCorrect, workspaceRoot, options.autonomy)
+	// failures auto-fix or just report. fileDiagnostics (always on, lazy) gives
+	// edit_file/write_file inline error diagnostics for the file they just wrote.
+	// lspShutdown tears down any language-server sessions either half spawned.
+	selfCorrector, fileDiagnostics, lspShutdown := newExecSelfCorrector(options.selfCorrect, workspaceRoot, options.autonomy)
 	defer lspShutdown()
 	result, err := agent.Run(runCtx, agentPrompt, provider, agent.Options{
 		MaxTurns:         resolved.MaxTurns,
 		ContextWindow:    resolveAgentContextWindow(runCtx, modelRegistry, resolved.Provider),
 		DeferThreshold:   effectiveDeferThreshold,
 		Specialists:      specialistRuntime.specialistInfos(),
+		Skills:           pluginActivation.skillInfos(deps.skillsDir()),
 		SessionID:        preparedSession.Session.SessionID,
 		CallingSessionID: options.callingSessionID,
 		CallingToolUseID: options.callingToolUseID,
@@ -524,6 +532,7 @@ func runExec(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) in
 		PermissionMode:   permissionMode,
 		Autonomy:         options.autonomy,
 		SelfCorrect:      selfCorrector,
+		FileDiagnostics:  fileDiagnostics,
 		// Headless exec: don't accept a no-tool-call turn as "done" while work
 		// clearly remains (pending plan items / a mid-step continuation cue) —
 		// nudge to continue, and finalize as INCOMPLETE rather than false success
@@ -699,23 +708,28 @@ func runExec(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) in
 // unless a changed file's language actually has one installed on PATH. The
 // returned cleanup shuts that manager down (terminating any spawned server
 // sessions); it is a no-op when self-correct is off.
-func newExecSelfCorrector(enabled bool, workspaceRoot string, autonomy string) (*agent.SelfCorrector, func()) {
-	if !enabled {
-		return nil, func() {}
-	}
+func newExecSelfCorrector(enabled bool, workspaceRoot string, autonomy string) (*agent.SelfCorrector, func(context.Context, string) string, func()) {
+	// The manager is created regardless of --self-correct: it also backs the
+	// always-on inline post-edit diagnostics (agent.NewFileDiagnostics), and it
+	// stays lazy — no language server is spawned unless an edited file's
+	// language actually has one installed on PATH.
 	manager := lsp.NewManager(workspaceRoot)
+	cleanup := func() {
+		ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+		defer cancel()
+		_ = manager.Shutdown(ctx)
+	}
+	fileDiagnostics := agent.NewFileDiagnostics(manager, workspaceRoot)
+	if !enabled {
+		return nil, fileDiagnostics, cleanup
+	}
 	corrector := agent.NewSelfCorrector(workspaceRoot, agent.NewLSPDiagnosticsChecker(manager), agent.NewProjectVerifier(workspaceRoot), agent.SelfCorrectConfig{
 		Enabled:      true,
 		IncludeTests: true,
 		IncludeLSP:   true,
 		Autonomy:     autonomy,
 	})
-	cleanup := func() {
-		ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
-		defer cancel()
-		_ = manager.Shutdown(ctx)
-	}
-	return corrector, cleanup
+	return corrector, fileDiagnostics, cleanup
 }
 
 func deferredEligibleCount(registry *tools.Registry, permissionMode agent.PermissionMode, enabledTools []string, disabledTools []string) int {
@@ -948,6 +962,14 @@ func writeExecProviderError(stdout io.Writer, stderr io.Writer, format execOutpu
 	if _, err := fmt.Fprintf(stderr, "[zero] %s\n", message); err != nil {
 		return exitCrash
 	}
+	// Append a one-line next step for recognized provider failures (auth /
+	// rate-limit / connectivity / …). The classifier gates on a provider-origin
+	// marker, so non-provider codes (sandbox_error, mcp_error) never draw a hint.
+	if hint := errhint.CLIHint(errors.New(message)); hint != "" {
+		if _, err := fmt.Fprintf(stderr, "[zero] %s\n", hint); err != nil {
+			return exitCrash
+		}
+	}
 	return exitProvider
 }
 

internal/cli/exec_provider_hint_test.go

@@ -0,0 +1,50 @@
+package cli
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+)
+
+// A recognized provider error in text mode prints the raw message plus a one-line
+// actionable hint; a non-provider error (or JSON mode) prints no hint.
+func TestWriteExecProviderErrorAppendsHint(t *testing.T) {
+	t.Run("provider auth error gets a hint", func(t *testing.T) {
+		var stdout, stderr bytes.Buffer
+		code := writeExecProviderError(&stdout, &stderr, execOutputText, "provider_error",
+			"auth error: your API key is missing or invalid")
+		if code != exitProvider {
+			t.Fatalf("exit code = %d, want exitProvider", code)
+		}
+		out := stderr.String()
+		if !strings.Contains(out, "auth error:") {
+			t.Fatalf("expected raw message, got %q", out)
+		}
+		if !strings.Contains(out, "zero auth") {
+			t.Fatalf("expected an actionable hint referencing `zero auth`, got %q", out)
+		}
+	})
+
+	t.Run("non-provider error gets no hint", func(t *testing.T) {
+		var stdout, stderr bytes.Buffer
+		writeExecProviderError(&stdout, &stderr, execOutputText, "sandbox_error",
+			"sandbox setup failed: permission denied")
+		out := stderr.String()
+		// Exactly one "[zero]" line — no spurious hint attached to a local error.
+		if n := strings.Count(out, "[zero]"); n != 1 {
+			t.Fatalf("expected exactly one [zero] line for a non-provider error, got %d:\n%s", n, out)
+		}
+	})
+
+	t.Run("json mode never appends a hint line", func(t *testing.T) {
+		var stdout, stderr bytes.Buffer
+		writeExecProviderError(&stdout, &stderr, execOutputJSON, "provider_error",
+			"auth error: bad key")
+		if stderr.Len() != 0 {
+			t.Fatalf("json mode must not write to stderr, got %q", stderr.String())
+		}
+		if strings.Contains(stdout.String(), "zero auth") {
+			t.Fatalf("json mode must not inject a hint into the structured payload, got %q", stdout.String())
+		}
+	})
+}

internal/cli/plugin_activate.go

@@ -3,7 +3,9 @@ package cli
 import (
 	"fmt"
 	"io"
+	"strings"
 
+	"github.com/Gitlawb/zero/internal/agent"
 	"github.com/Gitlawb/zero/internal/hooks"
 	"github.com/Gitlawb/zero/internal/plugins"
 	"github.com/Gitlawb/zero/internal/tools"
@@ -57,6 +59,27 @@ func activatePlugins(workspaceRoot string, registry *tools.Registry, deps appDep
 	return pluginActivation{hooks: result.Hooks, skillRoots: result.SkillRoots}
 }
 
+// skillInfos resolves the reusable skills the model can load via the skill tool —
+// the default skills dir merged with any plugin skill roots, the same set the
+// plugin-aware skill tool resolves against — as plain data for the agent's system
+// prompt. It returns nil when no skills are installed, so a skill-less run leaves
+// the prompt byte-identical.
+func (a pluginActivation) skillInfos(defaultDir string) []agent.SkillInfo {
+	merged, _ := plugins.MergedSkills(defaultDir, a.skillRoots)
+	if len(merged) == 0 {
+		return nil
+	}
+	infos := make([]agent.SkillInfo, 0, len(merged))
+	for _, skill := range merged {
+		name := strings.TrimSpace(skill.Name)
+		if name == "" {
+			continue
+		}
+		infos = append(infos, agent.SkillInfo{Name: name, Description: strings.TrimSpace(skill.Description)})
+	}
+	return infos
+}
+
 // formatLoadDiagnostic renders a plugin load diagnostic into a single warning
 // line, prefixing the diagnostic kind and appending the most specific locator
 // available (manifest path, then plugin path, then plugin ID) so an operator can

internal/cli/provider_catalog_onboarding_test.go

@@ -17,7 +17,6 @@ func TestProviderCatalogRuntimeProvidersShowOnboardingSetupHints(t *testing.T) {
 		defaultModel string
 	}{
 		{id: "openai", name: "OpenAI", transport: "openai", defaultModel: "gpt-4.1"},
-		{id: "aimlapi", name: "AI/ML API", transport: "openai-compatible", defaultModel: "openai/gpt-5-chat"},
 		{id: "groq", name: "Groq", transport: "openai-compatible", defaultModel: "llama-3.3-70b-versatile"},
 		{id: "longcat", name: "LongCat", transport: "openai-compatible", defaultModel: "LongCat-2.0"},
 		{id: "ollama-cloud", name: "Ollama Cloud", transport: "openai-compatible", defaultModel: "qwen3-coder:480b"},
@@ -48,21 +47,6 @@ func TestProviderCatalogRuntimeProvidersShowOnboardingSetupHints(t *testing.T) {
 	}
 }
 
-func TestProviderCatalogAIMLAPIRequiresAPIKey(t *testing.T) {
-	output := runProviderCatalogOnboarding(t)
-	block := providerCatalogOnboardingBlock(t, output, "aimlapi")
-
-	for _, want := range []string{
-		"requiresAuth=true",
-		"authEnvVars=AIMLAPI_API_KEY",
-		"setup: zero providers setup aimlapi --set-active",
-	} {
-		if !strings.Contains(block, want) {
-			t.Fatalf("expected AIMLAPI catalog block to contain %q, got:\n%s", want, block)
-		}
-	}
-}
-
 func TestProviderCatalogLongCatRequiresAPIKey(t *testing.T) {
 	output := runProviderCatalogOnboarding(t)
 	block := providerCatalogOnboardingBlock(t, output, "longcat")

internal/cli/setup.go

@@ -3,6 +3,7 @@ package cli
 import (
 	"fmt"
 	"io"
+	"net"
 	"net/url"
 	"strconv"
 	"strings"
@@ -453,6 +454,11 @@ func providerProfileIsLocal(profile config.ProviderProfile) bool {
 	case "localhost", "127.0.0.1", "::1":
 		return true
 	}
+	// Also treat explicit private-network IPs (192.168.x.y, 10.x.y.z,
+	// 172.16-31.x.y) as local — the user is pointing at their own LAN box.
+	if ip := net.ParseIP(parsed.Hostname()); ip != nil && ip.IsPrivate() {
+		return true
+	}
 	return false
 }
 

internal/cli/update.go

@@ -12,6 +12,7 @@ import (
 
 type updateOptions struct {
 	check      bool
+	apply      bool
 	json       bool
 	repository string
 	endpoint   string
@@ -20,6 +21,18 @@ type updateOptions struct {
 }
 
 func runUpdate(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) int {
+	return runUpdateCommand(args, stdout, stderr, deps, false)
+}
+
+func runUpgrade(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) int {
+	return runUpdateCommand(args, stdout, stderr, deps, true)
+}
+
+// runUpdateCommand backs both `zero update` and `zero upgrade`. defaultApply
+// makes `--apply` the implicit behavior for `zero upgrade` when neither
+// `--check` nor `--apply` is passed explicitly; `zero update` keeps requiring
+// one of the two so existing scripts around `zero update --check` don't change.
+func runUpdateCommand(args []string, stdout io.Writer, stderr io.Writer, deps appDeps, defaultApply bool) int {
 	options, help, err := parseUpdateArgs(args)
 	if err != nil {
 		return writeUsageError(stderr, err.Error())
@@ -30,10 +43,19 @@ func runUpdate(args []string, stdout io.Writer, stderr io.Writer, deps appDeps)
 		}
 		return exitSuccess
 	}
-	if !options.check {
-		return writeUsageError(stderr, "Only `zero update --check` is available right now.")
+	if options.check && options.apply {
+		return writeUsageError(stderr, "Pass only one of --check or --apply.")
 	}
-	checkOptions := update.Options{
+	if !options.check && !options.apply {
+		if !defaultApply {
+			return writeUsageError(stderr, "Pass --check to check for updates or --apply to install one (or use `zero upgrade`).")
+		}
+		options.apply = true
+	}
+	if options.apply && options.target != "" {
+		return writeUsageError(stderr, "--target cannot be combined with --apply; it only verifies release assets for --check.")
+	}
+	updateOptions := update.Options{
 		CurrentVersion: version,
 		Repository:     options.repository,
 		Endpoint:       options.endpoint,
@@ -44,10 +66,26 @@ func runUpdate(args []string, stdout io.Writer, stderr io.Writer, deps appDeps)
 		if err != nil {
 			return writeUsageError(stderr, err.Error())
 		}
-		checkOptions.GOOS = target.GOOS
-		checkOptions.GOARCH = target.GOARCH
+		updateOptions.GOOS = target.GOOS
+		updateOptions.GOARCH = target.GOARCH
 	}
-	result, err := deps.checkUpdate(context.Background(), checkOptions)
+	if options.apply {
+		result, err := deps.applyUpdate(context.Background(), updateOptions)
+		if err != nil {
+			return writeAppError(stderr, "Could not install update: "+err.Error(), exitCrash)
+		}
+		if options.json {
+			if err := writePrettyJSON(stdout, result); err != nil {
+				return exitCrash
+			}
+			return exitSuccess
+		}
+		if _, err := fmt.Fprintln(stdout, update.FormatApply(result)); err != nil {
+			return exitCrash
+		}
+		return exitSuccess
+	}
+	result, err := deps.checkUpdate(context.Background(), updateOptions)
 	if err != nil {
 		return writeAppError(stderr, "Could not check for updates: "+err.Error(), exitCrash)
 	}
@@ -72,6 +110,8 @@ func parseUpdateArgs(args []string) (updateOptions, bool, error) {
 			return options, true, nil
 		case arg == "--check":
 			options.check = true
+		case arg == "--apply":
+			options.apply = true
 		case arg == "--json":
 			options.json = true
 		case arg == "--repo":
@@ -155,14 +195,17 @@ func parseUpdateTarget(value string) (string, error) {
 func writeUpdateHelp(w io.Writer) error {
 	_, err := fmt.Fprint(w, `Usage:
   zero update --check [flags]
+  zero update --apply [flags]
+  zero upgrade [flags]
 
 Flags:
       --check                 Check the latest GitHub release without installing
-      --json                  Print the update check result as JSON
+      --apply                 Download, verify, and install the latest release
+      --json                  Print the update result as JSON
       --repo <owner/repo>     Repository to check when no endpoint is provided
       --endpoint <url|repo>   Release API URL or owner/repo slug to check
       --timeout <duration>    Release check timeout (default 5s)
-      --target <platform>     Release target to verify (for example windows-x64)
+      --target <platform>     Release target to verify with --check (for example windows-x64); not valid with --apply
   -h, --help                  Show this help
 `)
 	return err

internal/cli/workflow_test.go

@@ -662,3 +662,144 @@ func TestRunExecRejectsWorktreeDirWithoutWorktree(t *testing.T) {
 		t.Fatalf("expected empty stdout, got %q", stdout.String())
 	}
 }
+
+func TestParseChangesArgsAuto(t *testing.T) {
+	// 1. Correct parsing of auto
+	for _, args := range [][]string{
+		{"--auto"},
+		{"-a"},
+	} {
+		options, help, err := parseChangesArgs(args, "commit")
+		if err != nil {
+			t.Fatalf("parseChangesArgs(%v) error: %v", args, err)
+		}
+		if help {
+			t.Fatalf("parseChangesArgs(%v) returned help", args)
+		}
+		if !options.auto {
+			t.Fatalf("auto = false, want true (args %v)", args)
+		}
+	}
+
+	// 2. Reject --auto on inspect
+	_, _, err := parseChangesArgs([]string{"--auto"}, "inspect")
+	if err == nil || !strings.Contains(err.Error(), "--auto") {
+		t.Fatalf("expected --auto rejection on inspect, got %v", err)
+	}
+
+	// 3. Reject both --message and --auto on commit
+	_, _, err = parseChangesArgs([]string{"--message", "foo", "--auto"}, "commit")
+	if err == nil || !strings.Contains(err.Error(), "cannot specify both --message and --auto") {
+		t.Fatalf("expected --message and --auto conflict error, got %v", err)
+	}
+
+	// 4. Reject both empty message and --auto on commit
+	_, _, err = parseChangesArgs([]string{"--message=", "--auto"}, "commit")
+	if err == nil || !strings.Contains(err.Error(), "cannot specify both --message and --auto") {
+		t.Fatalf("expected empty --message and --auto conflict error, got %v", err)
+	}
+}
+
+type mockCommitMsgProvider struct {
+	response string
+	req      zeroruntime.CompletionRequest
+}
+
+func (p *mockCommitMsgProvider) StreamCompletion(ctx context.Context, request zeroruntime.CompletionRequest) (<-chan zeroruntime.StreamEvent, error) {
+	p.req = request
+	events := make(chan zeroruntime.StreamEvent, 2)
+	events <- zeroruntime.StreamEvent{Type: zeroruntime.StreamEventText, Content: p.response}
+	events <- zeroruntime.StreamEvent{Type: zeroruntime.StreamEventDone}
+	close(events)
+	return events, nil
+}
+
+func TestRunChangesCommitAuto(t *testing.T) {
+	cwd := t.TempDir()
+	summary := zerogit.ChangeSummary{
+		Root:   cwd,
+		Branch: "main",
+		Files:  []zerogit.FileChange{{Path: "README.md", Status: "modified"}},
+		Diff:   "some diff content with ghp_SECRETKEYHERE",
+	}
+
+	mockProv := &mockCommitMsgProvider{
+		response: "```\nfeat: auto commit message\n```",
+	}
+
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+	commitCalled := false
+
+	exitCode := runWithDeps([]string{"changes", "commit", "--auto"}, &stdout, &stderr, appDeps{
+		getwd: func() (string, error) { return cwd, nil },
+		inspectChanges: func(ctx context.Context, options zerogit.InspectOptions) (zerogit.ChangeSummary, error) {
+			return summary, nil
+		},
+		resolveConfig: func(workspaceRoot string, overrides config.Overrides) (config.ResolvedConfig, error) {
+			cfg := execResolvedConfig()
+			cfg.Provider.Name = "openai"
+			cfg.Provider.Model = "gpt-4o"
+			return cfg, nil
+		},
+		newProvider: func(profile config.ProviderProfile) (zeroruntime.Provider, error) {
+			return mockProv, nil
+		},
+		commitChanges: func(ctx context.Context, options zerogit.CommitOptions) (zerogit.CommitResult, error) {
+			commitCalled = true
+			if options.Message != "feat: auto commit message" {
+				t.Fatalf("expected commit message 'feat: auto commit message', got %q", options.Message)
+			}
+			return zerogit.CommitResult{
+				Root:       cwd,
+				Message:    options.Message,
+				Committed:  true,
+				CommitHash: "abc1234",
+				Before:     summary,
+			}, nil
+		},
+	})
+
+	if exitCode != exitSuccess {
+		t.Fatalf("expected exit code %d, got %d: %s", exitSuccess, exitCode, stderr.String())
+	}
+	if !commitCalled {
+		t.Fatal("expected commitChanges to be called")
+	}
+	// Verify that secret in the diff was redacted
+	promptContent := mockProv.req.Messages[0].Content
+	if strings.Contains(promptContent, "ghp_SECRETKEYHERE") {
+		t.Fatal("expected secret in diff to be redacted, but it was found in the prompt")
+	}
+	if !strings.Contains(promptContent, "[REDACTED]") && !strings.Contains(promptContent, "REDACTED") {
+		t.Fatalf("expected redacted diff content in prompt, got: %q", promptContent)
+	}
+	if !strings.Contains(stdout.String(), "Generating commit message using LLM...") {
+		t.Fatalf("expected status message in stdout, got %q", stdout.String())
+	}
+
+	t.Run("EmptyLLMResponse", func(t *testing.T) {
+		mockProvEmpty := &mockCommitMsgProvider{
+			response: "   \n\n   ",
+		}
+		var stdout, stderr bytes.Buffer
+		code := runWithDeps([]string{"changes", "commit", "--auto"}, &stdout, &stderr, appDeps{
+			getwd: func() (string, error) { return cwd, nil },
+			inspectChanges: func(ctx context.Context, options zerogit.InspectOptions) (zerogit.ChangeSummary, error) {
+				return summary, nil
+			},
+			resolveConfig: func(workspaceRoot string, overrides config.Overrides) (config.ResolvedConfig, error) {
+				return execResolvedConfig(), nil
+			},
+			newProvider: func(profile config.ProviderProfile) (zeroruntime.Provider, error) {
+				return mockProvEmpty, nil
+			},
+		})
+		if code == exitSuccess {
+			t.Fatal("expected command to fail when LLM returns empty response, got success")
+		}
+		if !strings.Contains(stderr.String(), "empty commit message") {
+			t.Fatalf("expected empty commit message error in stderr, got %q", stderr.String())
+		}
+	})
+}

internal/cli/workflows.go

@@ -6,13 +6,16 @@ import (
 	"io"
 	"strconv"
 	"strings"
+	"time"
 
+	"github.com/Gitlawb/zero/internal/config"
 	"github.com/Gitlawb/zero/internal/redaction"
 	"github.com/Gitlawb/zero/internal/selfverify"
 	"github.com/Gitlawb/zero/internal/testrunner"
 	"github.com/Gitlawb/zero/internal/verify"
 	"github.com/Gitlawb/zero/internal/worktrees"
 	"github.com/Gitlawb/zero/internal/zerogit"
+	"github.com/Gitlawb/zero/internal/zeroruntime"
 )
 
 type worktreeCommandOptions struct {
@@ -35,8 +38,10 @@ type changesCommandOptions struct {
 	cwd          string
 	baseRef      string
 	message      string
+	hasMessage   bool
 	dryRun       bool
 	maxDiffBytes int
+	auto         bool
 }
 
 func runWorktrees(args []string, stdout io.Writer, stderr io.Writer, deps appDeps) int {
@@ -198,9 +203,52 @@ func runChanges(args []string, stdout io.Writer, stderr io.Writer, deps appDeps)
 		}
 		return exitSuccess
 	case "commit":
+		var message string
+		if options.auto {
+			summary, err := deps.inspectChanges(context.Background(), zerogit.InspectOptions{
+				Cwd:          workspaceRoot,
+				MaxDiffBytes: options.maxDiffBytes,
+			})
+			if err != nil {
+				return writeExecUsageError(stderr, fmt.Sprintf("failed to inspect changes: %v", err))
+			}
+			if summary.Clean {
+				return writeExecUsageError(stderr, "no changes to commit")
+			}
+
+			resolved, err := deps.resolveConfig(workspaceRoot, config.Overrides{})
+			if err != nil {
+				return writeExecUsageError(stderr, fmt.Sprintf("failed to resolve config: %v", err))
+			}
+			if !config.HasProviderProfile(resolved.Provider) {
+				return writeExecUsageError(stderr, "no provider configured for auto-commit message")
+			}
+			provider, err := deps.newProvider(resolved.Provider)
+			if err != nil {
+				return writeExecUsageError(stderr, fmt.Sprintf("failed to create provider: %v", err))
+			}
+
+			if !options.json {
+				if _, err := fmt.Fprintln(stdout, "Generating commit message using LLM..."); err != nil {
+					return exitCrash
+				}
+			}
+
+			safeSummary := redactChangeSummary(summary)
+			genCtx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+			defer cancel()
+			msg, err := generateAutoCommitMessage(genCtx, provider, resolved.Provider.Model, safeSummary)
+			if err != nil {
+				return writeExecUsageError(stderr, fmt.Sprintf("failed to generate commit message: %v", err))
+			}
+			message = msg
+		} else {
+			message = options.message
+		}
+
 		result, err := deps.commitChanges(context.Background(), zerogit.CommitOptions{
 			Cwd:          workspaceRoot,
-			Message:      options.message,
+			Message:      message,
 			DryRun:       options.dryRun,
 			MaxDiffBytes: options.maxDiffBytes,
 		})
@@ -395,9 +443,11 @@ func parseChangesArgs(args []string, command string) (changesCommandOptions, boo
 				return options, false, err
 			}
 			options.message = value
+			options.hasMessage = true
 			index = next
 		case strings.HasPrefix(arg, "--message="):
 			options.message = strings.TrimSpace(strings.TrimPrefix(arg, "--message="))
+			options.hasMessage = true
 		case arg == "--diff-bytes":
 			value, next, err := nextFlagValue(args, index, arg)
 			if err != nil {
@@ -415,14 +465,19 @@ func parseChangesArgs(args []string, command string) (changesCommandOptions, boo
 				return options, false, err
 			}
 			options.maxDiffBytes = maxDiffBytes
+		case arg == "-a" || arg == "--auto":
+			options.auto = true
 		case strings.HasPrefix(arg, "-"):
 			return options, false, execUsageError{fmt.Sprintf("unknown changes flag %q", arg)}
 		default:
 			return options, false, execUsageError{fmt.Sprintf("unexpected changes argument %q", arg)}
 		}
 	}
-	if command != "commit" && (options.message != "" || options.dryRun) {
-		return options, false, execUsageError{"--message and --dry-run are only valid with `zero changes commit`"}
+	if command != "commit" && (options.hasMessage || options.dryRun || options.auto) {
+		return options, false, execUsageError{"--message, --dry-run, and --auto are only valid with `zero changes commit`"}
+	}
+	if command == "commit" && options.hasMessage && options.auto {
+		return options, false, execUsageError{"cannot specify both --message and --auto"}
 	}
 	if command == "commit" && options.baseRef != "" {
 		return options, false, execUsageError{"--base is only valid with `zero changes inspect`"}
@@ -712,9 +767,47 @@ Flags:
       --base <ref>        Diff against <ref>...HEAD instead of the working tree
       --diff-bytes <n>    Maximum diff bytes to include
   -m, --message <text>    Commit message for `+"`zero changes commit`"+`
+  -a, --auto              Auto-generate commit message using LLM (use --dry-run to preview)
       --dry-run           Preview commit metadata without mutating git state
       --json              Print JSON output
   -h, --help              Show this help
 `)
 	return err
 }
+
+func generateAutoCommitMessage(ctx context.Context, provider zeroruntime.Provider, model string, summary zerogit.ChangeSummary) (string, error) {
+	var promptBuilder strings.Builder
+	promptBuilder.WriteString("Analyze the following git diff and generate a concise, conventional commit message.\n")
+	promptBuilder.WriteString("The commit message subject line must be 72 characters or fewer, starting with a conventional commit type (e.g., feat, fix, docs, style, refactor, test, chore) followed by a colon and space, and a lowercase description.\n")
+	promptBuilder.WriteString("You may optionally include a blank line and a bulleted list of changes for the body if there are multiple files or complex changes.\n")
+	promptBuilder.WriteString("Output ONLY the raw commit message text. Do not wrap the message in markdown code block fence, do not include quotes or any introduction/explanation.\n\n")
+	promptBuilder.WriteString("Git Diff:\n")
+	promptBuilder.WriteString(summary.Diff)
+
+	request := zeroruntime.CompletionRequest{
+		Messages: []zeroruntime.Message{
+			{Role: zeroruntime.MessageRoleUser, Content: promptBuilder.String()},
+		},
+	}
+	stream, err := provider.StreamCompletion(ctx, request)
+	if err != nil {
+		return "", err
+	}
+	collected := zeroruntime.CollectStream(ctx, stream)
+	if collected.Error != "" {
+		return "", fmt.Errorf("%s", collected.Error)
+	}
+
+	msg := strings.TrimSpace(collected.Text)
+	if strings.HasPrefix(msg, "```") {
+		if idx := strings.Index(msg, "\n"); idx != -1 {
+			msg = msg[idx+1:]
+		}
+	}
+	msg = strings.TrimSuffix(msg, "```")
+	msg = strings.TrimSpace(msg)
+	if msg == "" {
+		return "", fmt.Errorf("provider returned empty commit message")
+	}
+	return msg, nil
+}

internal/config/resolver.go

@@ -130,7 +130,11 @@ func Resolve(options ResolveOptions) (ResolvedConfig, error) {
 
 	providers, active, err := normalizeProviders(cfg.Providers, cfg.ActiveProvider, options.Env)
 	if err != nil {
-		return ResolvedConfig{}, err
+		// On ErrNoActiveProvider, providers may still hold the successfully
+		// normalized (but active-less) profile list — keep it so a caller can fall
+		// back to an already-configured usable provider instead of treating this
+		// like a config with nothing set up at all.
+		return ResolvedConfig{Providers: providers}, err
 	}
 
 	return ResolvedConfig{
@@ -144,6 +148,7 @@ func Resolve(options ResolveOptions) (ResolvedConfig, error) {
 		Tools:          cfg.Tools,
 		Swarm:          cfg.Swarm,
 		Preferences:    cfg.Preferences,
+		KeyBindings:    cfg.KeyBindings,
 		LocalControl:   cfg.LocalControl,
 	}, nil
 }
@@ -225,6 +230,7 @@ func mergeConfig(dst *FileConfig, src FileConfig) {
 		dst.Preferences.Theme = strings.TrimSpace(src.Preferences.Theme)
 	}
 	mergeLocalControlConfig(&dst.LocalControl, src.LocalControl)
+	mergeKeyBindings(&dst.KeyBindings, src.KeyBindings)
 }
 
 func mergeProjectConfig(dst *FileConfig, src FileConfig) error {
@@ -277,6 +283,7 @@ func mergeProjectConfig(dst *FileConfig, src FileConfig) error {
 	if src.Swarm.MaxTeamSize != 0 {
 		dst.Swarm.MaxTeamSize = src.Swarm.MaxTeamSize
 	}
+	mergeKeyBindings(&dst.KeyBindings, src.KeyBindings)
 	// Local control is intentionally user-config/override only. A cloned project
 	// must not be able to make browser, desktop, or terminal automation tools
 	// appear in the model's tool surface.
@@ -583,9 +590,17 @@ func applyProviderEnv(cfg *FileConfig, providerKind ProviderKind, env envProfile
 	if profile.Name == "" {
 		profile.Name = string(providerKind)
 	}
+	// A non-official baseURL from the environment signals a proxy/gateway, so promote
+	// the first-party kind to its -compatible transport (which accepts a custom URL).
+	// Env-only by design: a custom baseURL in a project config.json is left as-is and
+	// rejected by Resolve, so an untrusted project can't redirect a first-party key to
+	// another host.
 	if providerKind == ProviderKindOpenAI && baseURL != "" && !isOfficialOpenAIBaseURL(baseURL) {
 		profile.ProviderKind = ProviderKindOpenAICompatible
 	}
+	if providerKind == ProviderKindAnthropic && baseURL != "" && !isOfficialAnthropicBaseURL(baseURL) {
+		profile.ProviderKind = ProviderKindAnthropicCompat
+	}
 	// When the env supplies only credentials (no baseURL) for a provider name that
 	// already exists, don't force the standard transport kind — a same-named
 	// openai-compatible/anthropic-compatible proxy or gateway must keep its kind.
@@ -665,6 +680,7 @@ func applyOverrides(cfg *FileConfig, overrides Overrides) {
 		cfg.Tools.deferThresholdSet = true
 	}
 	mergeLocalControlConfig(&cfg.LocalControl, overrides.LocalControl)
+	mergeKeyBindings(&cfg.KeyBindings, overrides.KeyBindings)
 	for _, provider := range overrides.Providers {
 		mergeProvider(cfg, provider)
 	}
@@ -700,6 +716,24 @@ func mergeLocalControlDriverConfig(dst *LocalControlDriverConfig, src LocalContr
 	}
 }
 
+func mergeKeyBindings(dst *KeyBindingsConfig, src KeyBindingsConfig) {
+	if src.ToggleDetailed != "" {
+		dst.ToggleDetailed = src.ToggleDetailed
+	}
+	if src.ToggleMouse != "" {
+		dst.ToggleMouse = src.ToggleMouse
+	}
+	if src.CycleReasoning != "" {
+		dst.CycleReasoning = src.CycleReasoning
+	}
+	if src.TogglePlan != "" {
+		dst.TogglePlan = src.TogglePlan
+	}
+	if src.ToggleSidebar != "" {
+		dst.ToggleSidebar = src.ToggleSidebar
+	}
+}
+
 func mergeMCPConfig(dst *MCPConfig, src MCPConfig) {
 	if len(src.Servers) == 0 {
 		return
@@ -834,7 +868,11 @@ func normalizeProvidersWithOptions(providers []ProviderProfile, activeName strin
 	}
 
 	if !activeFound {
-		return nil, ProviderProfile{}, fmt.Errorf("%w: active provider %q not found", ErrNoActiveProvider, activeName)
+		// Return the successfully normalized list alongside the error (rather than
+		// nil) so a caller like the interactive TUI can still fall back to an
+		// already-configured, usable provider instead of forcing a full
+		// re-onboarding wizard just because none was marked active.
+		return normalized, ProviderProfile{}, fmt.Errorf("%w: active provider %q not found", ErrNoActiveProvider, activeName)
 	}
 	if active.Model == "" {
 		return nil, ProviderProfile{}, &setupFixableError{

internal/config/resolver_test.go

@@ -617,6 +617,35 @@ func TestResolveUsesOpenAIEnvFallback(t *testing.T) {
 	}
 }
 
+func TestResolveUsesAnthropicEnvBaseURLAsCompatible(t *testing.T) {
+	// ANTHROPIC_BASE_URL pointing at a proxy/gateway must resolve to an
+	// anthropic-compatible provider (mirroring OPENAI_BASE_URL), not the "requires
+	// official baseURL" rejection — issue #479. The env is user-controlled, so a
+	// custom URL there is a deliberate gateway choice, unlike a project config.json.
+	resolved, err := Resolve(ResolveOptions{
+		Env: map[string]string{
+			"ANTHROPIC_API_KEY":  "sk-ant-env",
+			"ANTHROPIC_BASE_URL": "https://gateway.example/anthropic",
+			"ANTHROPIC_MODEL":    "claude-custom",
+		},
+	})
+	if err != nil {
+		t.Fatalf("Resolve() error = %v", err)
+	}
+	if resolved.ActiveProvider != "anthropic" {
+		t.Fatalf("ActiveProvider = %q, want anthropic", resolved.ActiveProvider)
+	}
+	if resolved.Provider.ProviderKind != ProviderKindAnthropicCompat {
+		t.Fatalf("ProviderKind = %q, want anthropic-compatible", resolved.Provider.ProviderKind)
+	}
+	if resolved.Provider.BaseURL != "https://gateway.example/anthropic" {
+		t.Fatalf("BaseURL = %q, want gateway URL", resolved.Provider.BaseURL)
+	}
+	if resolved.Provider.APIKey != "sk-ant-env" || resolved.Provider.Model != "claude-custom" {
+		t.Fatalf("Provider = %#v, want env credentials/model", resolved.Provider)
+	}
+}
+
 func TestResolveUsesOpenAIAPIKeyOnlyWithDefaultModel(t *testing.T) {
 	resolved, err := Resolve(ResolveOptions{
 		Env: map[string]string{
@@ -916,6 +945,25 @@ func TestResolveRejectsActiveProviderWithoutConfiguredProfiles(t *testing.T) {
 	}
 }
 
+func TestResolveKeepsNormalizedProvidersWhenNoneMarkedActive(t *testing.T) {
+	// Multiple providers configured (e.g. via `zero provider add`) but
+	// activeProvider is blank/stale — a caller like the interactive TUI still
+	// needs the normalized list to fall back to an already-usable provider
+	// instead of forcing a full re-onboarding wizard.
+	path := writeConfig(t, `{"providers":[
+		{"name":"work","provider_kind":"openai","apiKey":"sk-test","model":"gpt-test"},
+		{"name":"other","provider_kind":"openai","apiKey":"sk-other","model":"gpt-test"}
+	]}`)
+
+	resolved, err := Resolve(ResolveOptions{ProjectConfigPath: path, Env: map[string]string{}})
+	if !errors.Is(err, ErrNoActiveProvider) {
+		t.Fatalf("error = %v, want errors.Is(err, ErrNoActiveProvider)", err)
+	}
+	if len(resolved.Providers) != 2 {
+		t.Fatalf("Providers = %#v, want the 2 normalized profiles preserved despite the error", resolved.Providers)
+	}
+}
+
 func TestResolveTrimsProviderProfileAliasesBeforeFallback(t *testing.T) {
 	path := writeConfig(t, `{
 		"activeProvider": "custom",

internal/config/types.go

@@ -112,6 +112,25 @@ func (p PreferencesConfig) RecapsEnabled() bool {
 	return p.Recaps == nil || *p.Recaps
 }
 
+// KeyBindingDef defines one key binding string (e.g. "ctrl+o") that the TUI
+// can remap. An empty string means "use the built-in default" for that action.
+type KeyBindingDef string
+
+// KeyBindingsConfig holds the subset of TUI keybindings that users may remap
+// via config.json. Each field defaults to a sensible built-in when empty.
+type KeyBindingsConfig struct {
+	// ToggleDetailed toggles the detailed transcript view (default: ctrl+o).
+	ToggleDetailed KeyBindingDef `json:"toggleDetailed,omitempty"`
+	// ToggleMouse toggles mouse capture release (default: ctrl+e).
+	ToggleMouse KeyBindingDef `json:"toggleMouse,omitempty"`
+	// CycleReasoning cycles through reasoning effort levels (default: ctrl+t).
+	CycleReasoning KeyBindingDef `json:"cycleReasoning,omitempty"`
+	// TogglePlan toggles the plan panel expansion (default: ctrl+p).
+	TogglePlan KeyBindingDef `json:"togglePlan,omitempty"`
+	// ToggleSidebar toggles the right context sidebar (default: ctrl+b).
+	ToggleSidebar KeyBindingDef `json:"toggleSidebar,omitempty"`
+}
+
 // LocalControlConfig controls local browser/desktop/terminal automation helpers.
 // Helpers are discovered lazily by the tool that needs them; no setup command or
 // background probe is required during startup.
@@ -206,6 +225,7 @@ type FileConfig struct {
 	Tools          ToolsConfig        `json:"tools,omitempty"`
 	Swarm          SwarmConfig        `json:"swarm,omitempty"`
 	Preferences    PreferencesConfig  `json:"preferences,omitempty"`
+	KeyBindings    KeyBindingsConfig  `json:"keybindings,omitempty"`
 	LocalControl   LocalControlConfig `json:"localControl,omitempty"`
 }
 
@@ -220,6 +240,7 @@ func (cfg FileConfig) MarshalJSON() ([]byte, error) {
 		Tools          ToolsConfig         `json:"tools,omitempty"`
 		Swarm          SwarmConfig         `json:"swarm,omitempty"`
 		Preferences    PreferencesConfig   `json:"preferences,omitempty"`
+		KeyBindings    KeyBindingsConfig   `json:"keybindings,omitempty"`
 		LocalControl   *LocalControlConfig `json:"localControl,omitempty"`
 	}
 	raw := rawConfig{
@@ -232,6 +253,7 @@ func (cfg FileConfig) MarshalJSON() ([]byte, error) {
 		Tools:          cfg.Tools,
 		Swarm:          cfg.Swarm,
 		Preferences:    cfg.Preferences,
+		KeyBindings:    cfg.KeyBindings,
 	}
 	if !cfg.LocalControl.Empty() {
 		raw.LocalControl = &cfg.LocalControl
@@ -256,6 +278,7 @@ type Overrides struct {
 	Sandbox        SandboxConfig
 	Notify         NotifyConfig
 	Tools          ToolsConfig
+	KeyBindings    KeyBindingsConfig
 	LocalControl   LocalControlConfig
 }
 
@@ -270,6 +293,7 @@ type ResolvedConfig struct {
 	Tools          ToolsConfig
 	Swarm          SwarmConfig
 	Preferences    PreferencesConfig
+	KeyBindings    KeyBindingsConfig
 	LocalControl   LocalControlConfig
 }
 
@@ -316,6 +340,7 @@ func (cfg *FileConfig) UnmarshalJSON(data []byte) error {
 		Tools           ToolsConfig                `json:"tools"`
 		Swarm           SwarmConfig                `json:"swarm"`
 		Preferences     PreferencesConfig          `json:"preferences"`
+		KeyBindings     KeyBindingsConfig          `json:"keybindings"`
 		LocalControl    LocalControlConfig         `json:"localControl"`
 		MCPServers      map[string]MCPServerConfig `json:"mcpServers"`
 		MCPServersSnake map[string]MCPServerConfig `json:"mcp_servers"`
@@ -342,6 +367,7 @@ func (cfg *FileConfig) UnmarshalJSON(data []byte) error {
 	cfg.Tools = raw.Tools
 	cfg.Swarm = raw.Swarm
 	cfg.Preferences = raw.Preferences
+	cfg.KeyBindings = raw.KeyBindings
 	cfg.LocalControl = raw.LocalControl
 	if cfg.MCP.Servers == nil && (len(raw.MCPServers) > 0 || len(raw.MCPServersSnake) > 0) {
 		cfg.MCP.Servers = map[string]MCPServerConfig{}

internal/doctor/hardening.go

@@ -131,29 +131,45 @@ func doctorSandboxPolicy(cfg config.SandboxConfig) sandbox.Policy {
 // lspServersCheck reports which language servers ZERO would use are present on
 // PATH. Missing servers are not a failure — ZERO degrades to text-only edits for
 // those languages — so the worst status is WARN, and each missing server gets an
-// actionable install command keyed by its binary name.
+// actionable install command keyed by its binary name. Only the tier-1 servers
+// (lsp.CoreServerBinaries) drive the status: the long-tail servers configured
+// for breadth are listed informationally under missingOptional, because warning
+// about a missing zls on a machine with no Zig code would be permanent noise.
 func lspServersCheck(lookup func(string) (string, error)) Check {
 	if lookup == nil {
 		lookup = exec.LookPath
 	}
+	core := map[string]bool{}
+	for _, binary := range lsp.CoreServerBinaries() {
+		core[binary] = true
+	}
 	present := map[string]any{}
 	missing := map[string]any{}
+	missingOptional := map[string]any{}
 	for _, binary := range lsp.ServerBinaries() {
 		if _, err := lookup(binary); err == nil {
 			present[binary] = "on PATH"
 			continue
 		}
-		missing[binary] = lspRemedy(binary)
+		if core[binary] {
+			missing[binary] = lspRemedy(binary)
+		} else {
+			missingOptional[binary] = lspRemedy(binary)
+		}
+	}
+	details := map[string]any{"present": present}
+	if len(missingOptional) > 0 {
+		details["missingOptional"] = missingOptional
 	}
 	if len(missing) == 0 {
-		return check("lsp.servers", "LSP servers", StatusPass, "All known language servers are available on PATH.", map[string]any{
-			"present": present,
-		})
+		message := "All core language servers are available on PATH."
+		if len(missingOptional) > 0 {
+			message = fmt.Sprintf("All core language servers are available on PATH (%d optional server(s) not installed).", len(missingOptional))
+		}
+		return check("lsp.servers", "LSP servers", StatusPass, message, details)
 	}
-	return check("lsp.servers", "LSP servers", StatusWarn, fmt.Sprintf("%d language server(s) missing from PATH; affected files degrade to text-only edits.", len(missing)), map[string]any{
-		"present": present,
-		"missing": missing,
-	})
+	details["missing"] = missing
+	return check("lsp.servers", "LSP servers", StatusWarn, fmt.Sprintf("%d language server(s) missing from PATH; affected files degrade to text-only edits.", len(missing)), details)
 }
 
 // lspRemedy returns an actionable install command for a missing language-server

internal/errhint/errhint.go

@@ -0,0 +1,167 @@
+// Package errhint classifies provider/model failures into a few user-actionable
+// categories and turns them into a one-line "next step" hint.
+//
+// Provider errors already arrive with a classified string prefix from
+// providerio.ClassifiedError ("auth error:", "rate limit error:", …); lower-level
+// failures (DNS, TLS, timeouts, context-length) arrive as raw driver or library
+// messages. Classify matches both so the interactive error row (TUI) and the
+// `zero exec` provider-error path (CLI) can append one concrete next step instead
+// of dumping an identical red blob for every failure mode.
+package errhint
+
+import "strings"
+
+// Category buckets a provider/model failure into a small set of classes that each
+// map to a distinct recovery action.
+type Category int
+
+const (
+	// Unknown means the error didn't match any known signature; callers should
+	// emit no hint rather than guess.
+	Unknown Category = iota
+	Auth
+	RateLimit
+	Connectivity
+	ModelNotFound
+	ContextOverflow
+)
+
+// providerMarkers are the prefixes the provider layer attaches to every
+// provider-originated failure: providerio.ClassifiedError emits "auth error:",
+// "rate limit error:", "provider error:", and "provider request error:", and the
+// streaming paths wrap transport/decoding failures as "provider stream error:".
+// A UI surface's error can also be a *local* failure (a tool's "permission
+// denied", a "file does not exist", a config error), so Classify only proceeds
+// past this gate for messages that are recognizably from the provider — otherwise
+// a broad substring like "does not exist" would attach a bogus /model hint to an
+// unrelated local error.
+var providerMarkers = []string{
+	"auth error:",
+	"rate limit error:",
+	"provider error:",
+	"provider request error:",
+	"provider stream error:",
+}
+
+// Classify buckets err by scanning its message for known signatures. It is a
+// deliberately conservative string heuristic — the provider layer has already
+// discarded the numeric HTTP status by the time the error reaches a UI surface,
+// so the message is all we have. It first gates on a provider-origin marker (see
+// providerMarkers) so local failures never draw a provider hint, then sub-classifies.
+// Order matters: more specific signatures are tested before broader ones (e.g.
+// "context length" as overflow before the generic "timeout" as connectivity).
+func Classify(err error) Category {
+	if err == nil {
+		return Unknown
+	}
+	m := strings.ToLower(err.Error())
+	if !containsAny(m, providerMarkers...) {
+		return Unknown
+	}
+	switch {
+	case containsAny(m, "auth error:", "unauthorized", "api key", "api_key", "invalid_api_key",
+		"authentication", "permission denied", "forbidden") || containsStatusCode(m, "401", "403"):
+		return Auth
+	case containsAny(m, "rate limit", "rate_limit", "too many requests", "quota",
+		"resource_exhausted", "overloaded") || containsStatusCode(m, "429", "529"):
+		return RateLimit
+	case containsAny(m, "context length", "context window", "maximum context", "context_length_exceeded",
+		"too many tokens", "prompt is too long", "reduce the length", "maximum context length"):
+		return ContextOverflow
+	case containsAny(m, "model not found", "model_not_found", "does not exist", "unknown model",
+		"no such model", "unsupported model", "invalid model", "model is not"):
+		return ModelNotFound
+	case containsAny(m, "dial tcp", "no such host", "connection refused", "network is unreachable",
+		"i/o timeout", "context deadline exceeded", "tls handshake", "connection reset",
+		"unexpected eof", "lookup ", "timeout"):
+		return Connectivity
+	default:
+		return Unknown
+	}
+}
+
+// TUIHint returns a one-line hint referencing interactive slash commands, or ""
+// when the category is Unknown. Meant to sit under the raw error in the live
+// error row.
+func TUIHint(err error) string {
+	switch Classify(err) {
+	case Auth:
+		return "API key rejected — run /provider to re-check your credentials"
+	case RateLimit:
+		return "Rate limited — wait a moment, or switch model with /model"
+	case Connectivity:
+		return "Can't reach the provider — run /doctor --connectivity"
+	case ModelNotFound:
+		return "Model unavailable — pick another with /model"
+	case ContextOverflow:
+		return "Context window full — run /compact to free space"
+	default:
+		return ""
+	}
+}
+
+// CLIHint returns a one-line hint referencing `zero …` subcommands, or "" when the
+// category is Unknown. Meant for the non-interactive `zero exec` error path, where
+// slash commands don't apply.
+func CLIHint(err error) string {
+	switch Classify(err) {
+	case Auth:
+		return "API key rejected — run `zero auth` or set the provider's API key, then retry"
+	case RateLimit:
+		return "Rate limited — wait a moment, or switch model with --model"
+	case Connectivity:
+		return "Can't reach the provider — run `zero doctor`"
+	case ModelNotFound:
+		return "Model unavailable — run `zero doctor` or pick another with --model"
+	case ContextOverflow:
+		return "Context window full — shorten the prompt or start a fresh session"
+	default:
+		return ""
+	}
+}
+
+func containsAny(haystack string, needles ...string) bool {
+	for _, n := range needles {
+		if strings.Contains(haystack, n) {
+			return true
+		}
+	}
+	return false
+}
+
+// containsStatusCode reports whether haystack contains any of the given HTTP
+// status codes as a standalone number — not embedded in a longer digit run like
+// "completed in 4290ms" or "request id 14015" — so an incidental number can't be
+// mis-bucketed as an auth/rate-limit failure.
+func containsStatusCode(haystack string, codes ...string) bool {
+	return HasStatusCode(haystack, codes...)
+}
+
+// HasStatusCode reports whether haystack contains any of the given HTTP status
+// codes as a standalone number — not embedded in a longer digit run like
+// "completed in 4290ms" or "request id 14015". Exported so other packages (e.g.
+// the reconnect classifier) can gate on a status code without re-implementing
+// the digit-boundary check.
+func HasStatusCode(haystack string, codes ...string) bool {
+	for _, code := range codes {
+		for from := 0; ; {
+			rel := strings.Index(haystack[from:], code)
+			if rel < 0 {
+				break
+			}
+			pos := from + rel
+			beforeOK := pos == 0 || !isASCIIDigit(haystack[pos-1])
+			end := pos + len(code)
+			afterOK := end >= len(haystack) || !isASCIIDigit(haystack[end])
+			if beforeOK && afterOK {
+				return true
+			}
+			from = pos + 1
+		}
+	}
+	return false
+}
+
+func isASCIIDigit(b byte) bool {
+	return b >= '0' && b <= '9'
+}

internal/errhint/errhint_test.go

@@ -0,0 +1,94 @@
+package errhint
+
+import (
+	"errors"
+	"strings"
+	"testing"
+)
+
+func TestClassify(t *testing.T) {
+	cases := []struct {
+		name string
+		msg  string
+		want Category
+	}{
+		{"nil-ish empty", "", Unknown},
+		{"providerio auth prefix", "auth error: your API key is missing or invalid — run `zero auth`", Auth},
+		{"raw 401", "provider request error: 401 Unauthorized", Auth},
+		{"invalid api key", "provider request error: invalid_api_key: incorrect key provided", Auth},
+		{"rate limit prefix", "rate limit error: 429 too many requests", RateLimit},
+		{"overloaded", "provider error: model is overloaded, please retry", RateLimit},
+		{"resource exhausted gemini", "provider stream error: rpc error: code = ResourceExhausted desc = quota exceeded", RateLimit},
+		{"context length openai", "provider request error: this model's maximum context length is 128000 tokens", ContextOverflow},
+		{"context_length_exceeded", "provider request error: context_length_exceeded", ContextOverflow},
+		{"model not found", "provider request error: 404 the model `gpt-9` does not exist", ModelNotFound},
+		{"unknown model", "provider request error: unknown model: sonnet-99", ModelNotFound},
+		{"dns failure", "provider stream error: Post \"https://api...\": dial tcp: lookup api.foo.com: no such host", Connectivity},
+		{"connection refused", "provider stream error: dial tcp 127.0.0.1:443: connection refused", Connectivity},
+		{"deadline", "provider stream error: context deadline exceeded (Client.Timeout exceeded)", Connectivity},
+		{"provider-marked but no sub-signature", "provider error: something totally unexpected happened", Unknown},
+
+		// An incidental number that merely contains a status-code digit run must NOT
+		// be mis-bucketed (digit-boundary check), even behind a provider marker.
+		{"incidental latency number", "provider error: request completed in 4290ms then failed", Unknown},
+		{"incidental request id", "provider error: request id 14015 failed unexpectedly", Unknown},
+		{"standalone 401 still auth", "provider request error: got 401 from upstream", Auth},
+		{"standalone 429 still rate limit", "provider error: 429 returned", RateLimit},
+
+		// Local (non-provider) failures must NOT be classified — no provider marker,
+		// so no bogus recovery hint attaches to them.
+		{"local fs permission denied", "open /etc/shadow: permission denied", Unknown},
+		{"local file missing", "stat foo.go: no such file or directory", Unknown},
+		{"local model config typo", "unknown model: sonnet-99", Unknown},
+		{"tool timeout local", "tool bash timed out after 600s", Unknown},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := Classify(errors.New(tc.msg)); got != tc.want {
+				t.Fatalf("Classify(%q) = %v, want %v", tc.msg, got, tc.want)
+			}
+		})
+	}
+}
+
+func TestClassifyNil(t *testing.T) {
+	if got := Classify(nil); got != Unknown {
+		t.Fatalf("Classify(nil) = %v, want Unknown", got)
+	}
+}
+
+// Context-overflow must win over the connectivity "timeout" catch-all: a message
+// mentioning both "context length" and a timeout is an overflow, not a network
+// problem.
+func TestContextOverflowBeatsConnectivity(t *testing.T) {
+	err := errors.New("provider request error: maximum context length exceeded; request timeout")
+	if got := Classify(err); got != ContextOverflow {
+		t.Fatalf("Classify = %v, want ContextOverflow", got)
+	}
+}
+
+func TestHintsPresentForKnownCategories(t *testing.T) {
+	known := []error{
+		errors.New("auth error: bad key"),
+		errors.New("rate limit error: 429"),
+		errors.New("provider stream error: dial tcp: no such host"),
+		errors.New("provider request error: model does not exist"),
+		errors.New("provider request error: maximum context length is 128000 tokens"),
+	}
+	for _, err := range known {
+		if h := TUIHint(err); strings.TrimSpace(h) == "" {
+			t.Fatalf("TUIHint(%q) empty, want a hint", err)
+		}
+		if h := CLIHint(err); strings.TrimSpace(h) == "" {
+			t.Fatalf("CLIHint(%q) empty, want a hint", err)
+		}
+	}
+	// Unknown yields no hint on either surface.
+	unknown := errors.New("provider error: mystery")
+	if h := TUIHint(unknown); h != "" {
+		t.Fatalf("TUIHint(unknown) = %q, want empty", h)
+	}
+	if h := CLIHint(unknown); h != "" {
+		t.Fatalf("CLIHint(unknown) = %q, want empty", h)
+	}
+}

internal/lsp/registry.go

@@ -9,26 +9,134 @@ import (
 
 // serverCommands maps a file extension to the language-server command (argv) ZERO
 // will spawn. The first element is the binary looked up on PATH; missing binaries
-// are not an error, the agent just degrades to text-only for that file.
+// are not an error, the agent just degrades to text-only for that file. Every
+// command is the language's community-standard server invoked in stdio mode, so
+// a wrong guess can never spawn something surprising — only fail the PATH check.
 var serverCommands = map[string][]string{
-	".go":  {"gopls", "serve"},
-	".ts":  {"typescript-language-server", "--stdio"},
-	".tsx": {"typescript-language-server", "--stdio"},
-	".js":  {"typescript-language-server", "--stdio"},
-	".jsx": {"typescript-language-server", "--stdio"},
-	".py":  {"pyright-langserver", "--stdio"},
-	".rs":  {"rust-analyzer"},
+	".go":     {"gopls", "serve"},
+	".ts":     {"typescript-language-server", "--stdio"},
+	".tsx":    {"typescript-language-server", "--stdio"},
+	".mts":    {"typescript-language-server", "--stdio"},
+	".cts":    {"typescript-language-server", "--stdio"},
+	".js":     {"typescript-language-server", "--stdio"},
+	".jsx":    {"typescript-language-server", "--stdio"},
+	".mjs":    {"typescript-language-server", "--stdio"},
+	".cjs":    {"typescript-language-server", "--stdio"},
+	".py":     {"pyright-langserver", "--stdio"},
+	".pyi":    {"pyright-langserver", "--stdio"},
+	".rs":     {"rust-analyzer"},
+	".c":      {"clangd"},
+	".h":      {"clangd"},
+	".cpp":    {"clangd"},
+	".cc":     {"clangd"},
+	".cxx":    {"clangd"},
+	".hpp":    {"clangd"},
+	".java":   {"jdtls"},
+	".kt":     {"kotlin-language-server"},
+	".kts":    {"kotlin-language-server"},
+	".rb":     {"ruby-lsp"},
+	".php":    {"intelephense", "--stdio"},
+	".zig":    {"zls"},
+	".lua":    {"lua-language-server"},
+	".ex":     {"elixir-ls"},
+	".exs":    {"elixir-ls"},
+	".hs":     {"haskell-language-server-wrapper", "--lsp"},
+	".swift":  {"sourcekit-lsp"},
+	".ml":     {"ocamllsp"},
+	".mli":    {"ocamllsp"},
+	".scala":  {"metals"},
+	".clj":    {"clojure-lsp"},
+	".cljs":   {"clojure-lsp"},
+	".cljc":   {"clojure-lsp"},
+	".dart":   {"dart", "language-server"},
+	".gleam":  {"gleam", "lsp"},
+	".nix":    {"nixd"},
+	".tf":     {"terraform-ls", "serve"},
+	".sh":     {"bash-language-server", "start"},
+	".bash":   {"bash-language-server", "start"},
+	".yaml":   {"yaml-language-server", "--stdio"},
+	".yml":    {"yaml-language-server", "--stdio"},
+	".json":   {"vscode-json-language-server", "--stdio"},
+	".css":    {"vscode-css-language-server", "--stdio"},
+	".scss":   {"vscode-css-language-server", "--stdio"},
+	".less":   {"vscode-css-language-server", "--stdio"},
+	".html":   {"vscode-html-language-server", "--stdio"},
+	".svelte": {"svelteserver", "--stdio"},
+	".vue":    {"vue-language-server", "--stdio"},
+	".astro":  {"astro-ls", "--stdio"},
 }
 
 // languageIDs maps a file extension to the LSP languageId used in didOpen.
 var languageIDs = map[string]string{
-	".go":  "go",
-	".ts":  "typescript",
-	".tsx": "typescriptreact",
-	".js":  "javascript",
-	".jsx": "javascriptreact",
-	".py":  "python",
-	".rs":  "rust",
+	".go":     "go",
+	".ts":     "typescript",
+	".tsx":    "typescriptreact",
+	".mts":    "typescript",
+	".cts":    "typescript",
+	".js":     "javascript",
+	".jsx":    "javascriptreact",
+	".mjs":    "javascript",
+	".cjs":    "javascript",
+	".py":     "python",
+	".pyi":    "python",
+	".rs":     "rust",
+	".c":      "c",
+	".h":      "c",
+	".cpp":    "cpp",
+	".cc":     "cpp",
+	".cxx":    "cpp",
+	".hpp":    "cpp",
+	".java":   "java",
+	".kt":     "kotlin",
+	".kts":    "kotlin",
+	".rb":     "ruby",
+	".php":    "php",
+	".zig":    "zig",
+	".lua":    "lua",
+	".ex":     "elixir",
+	".exs":    "elixir",
+	".hs":     "haskell",
+	".swift":  "swift",
+	".ml":     "ocaml",
+	".mli":    "ocaml",
+	".scala":  "scala",
+	".clj":    "clojure",
+	".cljs":   "clojure",
+	".cljc":   "clojure",
+	".dart":   "dart",
+	".gleam":  "gleam",
+	".nix":    "nix",
+	".tf":     "terraform",
+	".sh":     "shellscript",
+	".bash":   "shellscript",
+	".yaml":   "yaml",
+	".yml":    "yaml",
+	".json":   "json",
+	".css":    "css",
+	".scss":   "scss",
+	".less":   "less",
+	".html":   "html",
+	".svelte": "svelte",
+	".vue":    "vue",
+	".astro":  "astro",
+}
+
+// coreServerBinaries are the tier-1 servers for the languages agents hit most;
+// `zero doctor` treats their absence as warn-worthy. The long tail of servers
+// configured above for breadth is reported informationally only — warning about
+// a missing zls on a machine with no Zig code would be permanent noise.
+var coreServerBinaries = []string{
+	"gopls",
+	"typescript-language-server",
+	"pyright-langserver",
+	"rust-analyzer",
+}
+
+// CoreServerBinaries returns the tier-1 server binaries (sorted copy).
+func CoreServerBinaries() []string {
+	binaries := append([]string(nil), coreServerBinaries...)
+	sort.Strings(binaries)
+	return binaries
 }
 
 // ServerBinaries returns the unique set of language-server binaries ZERO may

internal/modelregistry/catalog.go

@@ -63,6 +63,10 @@ func DefaultModelEntries() []ModelEntry {
 		googleModel("gemini-2.5-flash-lite", "Gemini 2.5 Flash-Lite", "gemini-2.5-flash-lite", ModelStatusActive, []string{"google:gemini-2.5-flash-lite", "gemini-flash-lite"}, ContextLimits{ContextWindow: 1_048_576, MaxOutputTokens: 65_536}, ModelCost{InputPerMillion: 0.1, CachedInputPerMillion: 0.01, OutputPerMillion: 0.4}, []ModelCapability{ModelCapabilityVision, ModelCapabilityJSONMode, ModelCapabilityReasoning, ModelCapabilityLongContext}, standardReasoningEfforts(), "Google low-cost Flash model for background routing and summaries."),
 	}
 	decorateModelDepth(entries)
+	// Overlay volatile facts (context limits, base pricing) from a cached
+	// models.dev snapshot when one is present and fresh — see modelsdev.go.
+	// Identity fields (ids, aliases, patterns, deprecations) stay curated.
+	entries = applyModelsDevOverrides(entries, cachedModelsDevProviders())
 	return cloneModelEntries(entries)
 }
 

internal/modelregistry/modelsdev.go

@@ -0,0 +1,269 @@
+package modelregistry
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"time"
+)
+
+// Live models.dev overlay for the curated catalog. The hand-maintained
+// DefaultModelEntries list is the source of truth for identity — ids, aliases,
+// match patterns, deprecations, escalation targets — but its VOLATILE facts
+// (context window, max output tokens, per-million pricing) go stale between
+// releases. When a cached snapshot of https://models.dev/api.json is present,
+// those fields are refreshed from it at registry construction; everything else
+// stays curated. The overlay never adds models (an auto-added entry would lack
+// aliases, match patterns, and provider wiring) and never touches the network
+// on the registry hot path — fetching happens only in the explicit background
+// refresh, cached to disk with a TTL.
+
+const (
+	modelsDevDefaultURL = "https://models.dev/api.json"
+	// modelsDevRefreshAfter is how old the cache may get before a background
+	// refresh re-fetches it.
+	modelsDevRefreshAfter = 24 * time.Hour
+	// modelsDevMaxAge is the oldest cache still applied as an overlay. Beyond
+	// this the curated catalog (updated with the binary) is likely fresher than
+	// the snapshot, so a stale file is ignored rather than trusted.
+	modelsDevMaxAge      = 7 * 24 * time.Hour
+	modelsDevFetchLimit  = 32 << 20 // 32MiB guard on the response body
+	modelsDevFetchWindow = 15 * time.Second
+)
+
+// modelsDevModel is the subset of a models.dev model record the overlay uses.
+type modelsDevModel struct {
+	Limit struct {
+		Context int `json:"context"`
+		Output  int `json:"output"`
+	} `json:"limit"`
+	Cost struct {
+		Input      float64 `json:"input"`
+		Output     float64 `json:"output"`
+		CacheRead  float64 `json:"cache_read"`
+		CacheWrite float64 `json:"cache_write"`
+	} `json:"cost"`
+}
+
+// modelsDevProvider matches one provider object in api.json.
+type modelsDevProvider struct {
+	Models map[string]modelsDevModel `json:"models"`
+}
+
+// parseModelsDev decodes an api.json document into provider-slug -> api-model
+// -> record. api.json is a top-level object keyed by provider id.
+func parseModelsDev(data []byte) (map[string]map[string]modelsDevModel, error) {
+	var doc map[string]modelsDevProvider
+	if err := json.Unmarshal(data, &doc); err != nil {
+		return nil, fmt.Errorf("modelregistry: parse models.dev: %w", err)
+	}
+	providers := make(map[string]map[string]modelsDevModel, len(doc))
+	for slug, provider := range doc {
+		if len(provider.Models) == 0 {
+			continue
+		}
+		providers[slug] = provider.Models
+	}
+	if len(providers) == 0 {
+		return nil, fmt.Errorf("modelregistry: models.dev document has no providers")
+	}
+	return providers, nil
+}
+
+// modelsDevSlugs maps an entry's provider kind to the models.dev provider ids
+// worth checking. Only first-party slugs are used: router entries on models.dev
+// carry generic or stale numbers.
+func modelsDevSlugs(kind ProviderKind) []string {
+	switch kind {
+	case ProviderAnthropic:
+		return []string{"anthropic"}
+	case ProviderOpenAI:
+		return []string{"openai"}
+	case ProviderGoogle:
+		return []string{"google", "google-vertex"}
+	default:
+		return nil
+	}
+}
+
+// applyModelsDevOverrides refreshes each curated entry's context limits and
+// base pricing from the snapshot, when the snapshot knows the model. Tiered
+// pricing is never touched: models.dev has no tier data, and mixing a live
+// base rate with curated tiers would misprice the tier boundaries.
+func applyModelsDevOverrides(entries []ModelEntry, providers map[string]map[string]modelsDevModel) []ModelEntry {
+	if len(providers) == 0 {
+		return entries
+	}
+	for i := range entries {
+		entry := &entries[i]
+		var record modelsDevModel
+		found := false
+		for _, slug := range modelsDevSlugs(entry.Provider) {
+			if models, ok := providers[slug]; ok {
+				if candidate, ok := models[strings.TrimSpace(entry.APIModel)]; ok {
+					record = candidate
+					found = true
+					break
+				}
+			}
+		}
+		if !found {
+			continue
+		}
+		if record.Limit.Context > 0 {
+			entry.ContextLimits.ContextWindow = record.Limit.Context
+		}
+		if record.Limit.Output > 0 {
+			entry.ContextLimits.MaxOutputTokens = record.Limit.Output
+		}
+		if len(entry.Cost.Tiers) == 0 && record.Cost.Input > 0 && record.Cost.Output > 0 {
+			entry.Cost.InputPerMillion = record.Cost.Input
+			entry.Cost.OutputPerMillion = record.Cost.Output
+			if record.Cost.CacheRead > 0 {
+				entry.Cost.CachedInputPerMillion = record.Cost.CacheRead
+			}
+			if record.Cost.CacheWrite > 0 {
+				entry.Cost.CacheWritePerMillion = record.Cost.CacheWrite
+			}
+			entry.Cost.Source = "models.dev/api.json (cached)"
+		}
+	}
+	return entries
+}
+
+// modelsDevCachePath returns the on-disk cache location. ZERO_MODELS_CACHE_PATH
+// overrides it (used by tests and unusual setups).
+func modelsDevCachePath() (string, error) {
+	if override := strings.TrimSpace(os.Getenv("ZERO_MODELS_CACHE_PATH")); override != "" {
+		return override, nil
+	}
+	base, err := os.UserCacheDir()
+	if err != nil {
+		return "", err
+	}
+	return filepath.Join(base, "zero", "modelsdev.json"), nil
+}
+
+var (
+	modelsDevOnce    sync.Once
+	modelsDevCached  map[string]map[string]modelsDevModel
+	modelsDevEnabled atomic.Bool
+)
+
+// EnableModelsDevOverlay opts the process into applying the cached models.dev
+// snapshot on top of the curated catalog. The CLI entrypoint calls it; library
+// consumers and tests that never do get the curated catalog byte-identical to
+// before, so hermetic tests can't be perturbed by a cache file on the machine.
+// ZERO_DISABLE_MODELS_FETCH disables both the overlay and the fetch.
+func EnableModelsDevOverlay() {
+	modelsDevEnabled.Store(true)
+}
+
+// cachedModelsDevProviders loads the cached snapshot once per process. Not
+// enabled, missing, stale (> modelsDevMaxAge), or malformed all yield nil and
+// the curated catalog is used untouched. Read once deliberately:
+// DefaultRegistry is called on hot paths (pickers, cost views) and must not
+// re-stat the file every time; a background refresh benefits the NEXT process.
+func cachedModelsDevProviders() map[string]map[string]modelsDevModel {
+	if !modelsDevEnabled.Load() || strings.TrimSpace(os.Getenv("ZERO_DISABLE_MODELS_FETCH")) != "" {
+		return nil
+	}
+	modelsDevOnce.Do(func() {
+		path, err := modelsDevCachePath()
+		if err != nil {
+			return
+		}
+		info, err := os.Stat(path)
+		if err != nil || time.Since(info.ModTime()) > modelsDevMaxAge {
+			return
+		}
+		data, err := os.ReadFile(path)
+		if err != nil {
+			return
+		}
+		providers, err := parseModelsDev(data)
+		if err != nil {
+			return
+		}
+		modelsDevCached = providers
+	})
+	return modelsDevCached
+}
+
+// resetModelsDevCacheForTest clears the process-level cache memoization and
+// disables the overlay.
+func resetModelsDevCacheForTest() {
+	modelsDevOnce = sync.Once{}
+	modelsDevCached = nil
+	modelsDevEnabled.Store(false)
+}
+
+// RefreshModelsDevCache fetches models.dev/api.json into the on-disk cache
+// when the cache is missing or older than modelsDevRefreshAfter. It is safe to
+// call fire-and-forget from startup (use a goroutine); it never affects the
+// current process's registry (see cachedModelsDevProviders). Disabled entirely
+// by ZERO_DISABLE_MODELS_FETCH. The URL can be overridden with ZERO_MODELS_URL.
+func RefreshModelsDevCache(ctx context.Context) error {
+	if strings.TrimSpace(os.Getenv("ZERO_DISABLE_MODELS_FETCH")) != "" {
+		return nil
+	}
+	path, err := modelsDevCachePath()
+	if err != nil {
+		return err
+	}
+	if info, err := os.Stat(path); err == nil && time.Since(info.ModTime()) < modelsDevRefreshAfter {
+		return nil
+	}
+
+	url := strings.TrimSpace(os.Getenv("ZERO_MODELS_URL"))
+	if url == "" {
+		url = modelsDevDefaultURL
+	}
+	fetchCtx, cancel := context.WithTimeout(ctx, modelsDevFetchWindow)
+	defer cancel()
+	request, err := http.NewRequestWithContext(fetchCtx, http.MethodGet, url, nil)
+	if err != nil {
+		return err
+	}
+	request.Header.Set("User-Agent", "zero-models-refresh/0.1")
+	response, err := http.DefaultClient.Do(request)
+	if err != nil {
+		return err
+	}
+	defer response.Body.Close()
+	if response.StatusCode != http.StatusOK {
+		return fmt.Errorf("modelregistry: models.dev fetch: HTTP %d", response.StatusCode)
+	}
+	data, err := io.ReadAll(io.LimitReader(response.Body, modelsDevFetchLimit))
+	if err != nil {
+		return err
+	}
+	// Validate before persisting: a bad body must never clobber a good cache.
+	if _, err := parseModelsDev(data); err != nil {
+		return err
+	}
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		return err
+	}
+	temp, err := os.CreateTemp(filepath.Dir(path), "modelsdev-*.json")
+	if err != nil {
+		return err
+	}
+	if _, err := temp.Write(data); err != nil {
+		_ = temp.Close()
+		_ = os.Remove(temp.Name())
+		return err
+	}
+	if err := temp.Close(); err != nil {
+		_ = os.Remove(temp.Name())
+		return err
+	}
+	return os.Rename(temp.Name(), path)
+}

internal/modelregistry/modelsdev_test.go

@@ -0,0 +1,228 @@
+package modelregistry
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+)
+
+const sampleModelsDev = `{
+  "anthropic": {
+    "id": "anthropic",
+    "models": {
+      "claude-sonnet-4-5-20250929": {
+        "limit": {"context": 1000000, "output": 64000},
+        "cost": {"input": 3.5, "output": 17.5, "cache_read": 0.35, "cache_write": 4.4}
+      }
+    }
+  },
+  "google": {
+    "id": "google",
+    "models": {
+      "gemini-2.5-pro": {
+        "limit": {"context": 2097152, "output": 65536},
+        "cost": {"input": 9.99, "output": 9.99}
+      }
+    }
+  }
+}`
+
+func TestParseModelsDev(t *testing.T) {
+	providers, err := parseModelsDev([]byte(sampleModelsDev))
+	if err != nil {
+		t.Fatal(err)
+	}
+	record, ok := providers["anthropic"]["claude-sonnet-4-5-20250929"]
+	if !ok {
+		t.Fatal("expected anthropic sonnet record")
+	}
+	if record.Limit.Context != 1_000_000 || record.Cost.Input != 3.5 {
+		t.Fatalf("unexpected record: %+v", record)
+	}
+	if _, err := parseModelsDev([]byte(`{}`)); err == nil {
+		t.Fatal("empty document must be rejected")
+	}
+	if _, err := parseModelsDev([]byte(`not json`)); err == nil {
+		t.Fatal("malformed document must be rejected")
+	}
+}
+
+func TestApplyModelsDevOverrides(t *testing.T) {
+	// Point the cache at a non-existent file so DefaultModelEntries returns the
+	// pure curated catalog, then apply the sample snapshot explicitly.
+	t.Setenv("ZERO_MODELS_CACHE_PATH", filepath.Join(t.TempDir(), "absent.json"))
+	resetModelsDevCacheForTest()
+	t.Cleanup(resetModelsDevCacheForTest)
+
+	providers, err := parseModelsDev([]byte(sampleModelsDev))
+	if err != nil {
+		t.Fatal(err)
+	}
+	entries := applyModelsDevOverrides(DefaultModelEntries(), providers)
+
+	var sonnet, geminiPro, opus ModelEntry
+	for _, entry := range entries {
+		switch entry.ID {
+		case "claude-sonnet-4.5":
+			sonnet = entry
+		case "gemini-2.5-pro":
+			geminiPro = entry
+		case "claude-opus-4.1":
+			opus = entry
+		}
+	}
+
+	// Known model: limits and base pricing refreshed from the snapshot.
+	if sonnet.ContextLimits.ContextWindow != 1_000_000 || sonnet.ContextLimits.MaxOutputTokens != 64_000 {
+		t.Fatalf("sonnet limits not overridden: %+v", sonnet.ContextLimits)
+	}
+	if sonnet.Cost.InputPerMillion != 3.5 || sonnet.Cost.OutputPerMillion != 17.5 || sonnet.Cost.CacheWritePerMillion != 4.4 {
+		t.Fatalf("sonnet cost not overridden: %+v", sonnet.Cost)
+	}
+	if sonnet.Cost.Source != "models.dev/api.json (cached)" {
+		t.Fatalf("sonnet cost source not marked: %q", sonnet.Cost.Source)
+	}
+
+	// Tiered pricing is curated: limits refresh, cost must NOT (gemini-2.5-pro
+	// has curated tiers and the snapshot's flat 9.99 would misprice them).
+	if geminiPro.ContextLimits.ContextWindow != 2_097_152 {
+		t.Fatalf("gemini limits not overridden: %+v", geminiPro.ContextLimits)
+	}
+	if geminiPro.Cost.InputPerMillion == 9.99 || len(geminiPro.Cost.Tiers) == 0 {
+		t.Fatalf("tiered cost must stay curated: %+v", geminiPro.Cost)
+	}
+
+	// Model absent from the snapshot: untouched.
+	if opus.ContextLimits.ContextWindow != 200_000 || opus.Cost.InputPerMillion != 15 {
+		t.Fatalf("opus must be untouched: %+v %+v", opus.ContextLimits, opus.Cost)
+	}
+}
+
+func TestRefreshModelsDevCacheFetchesAndCaches(t *testing.T) {
+	hits := 0
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		hits++
+		_, _ = w.Write([]byte(sampleModelsDev))
+	}))
+	defer server.Close()
+
+	cachePath := filepath.Join(t.TempDir(), "modelsdev.json")
+	t.Setenv("ZERO_MODELS_CACHE_PATH", cachePath)
+	t.Setenv("ZERO_MODELS_URL", server.URL)
+	t.Setenv("ZERO_DISABLE_MODELS_FETCH", "")
+
+	if err := RefreshModelsDevCache(t.Context()); err != nil {
+		t.Fatal(err)
+	}
+	if hits != 1 {
+		t.Fatalf("expected 1 fetch, got %d", hits)
+	}
+	if _, err := os.Stat(cachePath); err != nil {
+		t.Fatalf("cache file missing: %v", err)
+	}
+	// Fresh cache: second call must not re-fetch.
+	if err := RefreshModelsDevCache(t.Context()); err != nil {
+		t.Fatal(err)
+	}
+	if hits != 1 {
+		t.Fatalf("fresh cache must skip the fetch, got %d hits", hits)
+	}
+}
+
+func TestRefreshModelsDevCacheRejectsBadBodyWithoutClobbering(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		_, _ = w.Write([]byte("not json"))
+	}))
+	defer server.Close()
+
+	cachePath := filepath.Join(t.TempDir(), "modelsdev.json")
+	if err := os.WriteFile(cachePath, []byte(sampleModelsDev), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	stale := time.Now().Add(-48 * time.Hour)
+	if err := os.Chtimes(cachePath, stale, stale); err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("ZERO_MODELS_CACHE_PATH", cachePath)
+	t.Setenv("ZERO_MODELS_URL", server.URL)
+
+	if err := RefreshModelsDevCache(t.Context()); err == nil {
+		t.Fatal("bad body must return an error")
+	}
+	content, err := os.ReadFile(cachePath)
+	if err != nil || string(content) != sampleModelsDev {
+		t.Fatal("bad fetch must not clobber the existing cache")
+	}
+}
+
+func TestRefreshModelsDevCacheDisabledByEnv(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		t.Fatal("fetch must not happen when disabled")
+	}))
+	defer server.Close()
+	t.Setenv("ZERO_MODELS_CACHE_PATH", filepath.Join(t.TempDir(), "modelsdev.json"))
+	t.Setenv("ZERO_MODELS_URL", server.URL)
+	t.Setenv("ZERO_DISABLE_MODELS_FETCH", "1")
+	if err := RefreshModelsDevCache(t.Context()); err != nil {
+		t.Fatal(err)
+	}
+}
+
+func TestCachedModelsDevProvidersIgnoresStaleCache(t *testing.T) {
+	cachePath := filepath.Join(t.TempDir(), "modelsdev.json")
+	if err := os.WriteFile(cachePath, []byte(sampleModelsDev), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	stale := time.Now().Add(-modelsDevMaxAge - time.Hour)
+	if err := os.Chtimes(cachePath, stale, stale); err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("ZERO_MODELS_CACHE_PATH", cachePath)
+	resetModelsDevCacheForTest()
+	t.Cleanup(resetModelsDevCacheForTest)
+	EnableModelsDevOverlay()
+
+	if providers := cachedModelsDevProviders(); providers != nil {
+		t.Fatal("stale cache must be ignored")
+	}
+}
+
+func TestCachedModelsDevProvidersRequiresOptIn(t *testing.T) {
+	cachePath := filepath.Join(t.TempDir(), "modelsdev.json")
+	if err := os.WriteFile(cachePath, []byte(sampleModelsDev), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("ZERO_MODELS_CACHE_PATH", cachePath)
+	resetModelsDevCacheForTest()
+	t.Cleanup(resetModelsDevCacheForTest)
+
+	// Without EnableModelsDevOverlay a fresh, valid cache must still be ignored:
+	// library consumers and hermetic tests get the pure curated catalog.
+	if providers := cachedModelsDevProviders(); providers != nil {
+		t.Fatal("overlay must be opt-in")
+	}
+}
+
+func TestDefaultModelEntriesAppliesFreshCache(t *testing.T) {
+	cachePath := filepath.Join(t.TempDir(), "modelsdev.json")
+	if err := os.WriteFile(cachePath, []byte(sampleModelsDev), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("ZERO_MODELS_CACHE_PATH", cachePath)
+	resetModelsDevCacheForTest()
+	t.Cleanup(resetModelsDevCacheForTest)
+	EnableModelsDevOverlay()
+
+	for _, entry := range DefaultModelEntries() {
+		if entry.ID == "claude-sonnet-4.5" {
+			if entry.ContextLimits.ContextWindow != 1_000_000 {
+				t.Fatalf("fresh cache must overlay the registry: %+v", entry.ContextLimits)
+			}
+			return
+		}
+	}
+	t.Fatal("claude-sonnet-4.5 not found")
+}

internal/oauth/encrypt.go

@@ -114,7 +114,12 @@ func createSecretFile(path string) ([]byte, error) {
 			}
 			return writeNewSecretFile(path)
 		}
-		if !errors.Is(err, os.ErrExist) {
+		// On Windows a concurrent holder's os.Remove leaves the lock file in a
+		// "delete pending" state, so an O_EXCL create races it with
+		// ERROR_ACCESS_DENIED (os.ErrPermission) rather than ErrExist. Treat that
+		// as contention and retry too — mirroring acquireFileLock in lock.go —
+		// otherwise concurrent secret creation spuriously fails on Windows.
+		if !errors.Is(err, os.ErrExist) && !errors.Is(err, os.ErrPermission) {
 			return nil, fmt.Errorf("oauth: create token secret lock: %w", err)
 		}
 		if data, rerr := readSecretFileRetry(path); rerr == nil {

internal/providercatalog/catalog.go

@@ -34,8 +34,6 @@ const (
 
 var ErrUnknownProvider = errors.New("unknown provider")
 
-const AIMLAPIID = "aimlapi"
-
 type Descriptor struct {
 	ID                  string
 	Name                string
@@ -101,7 +99,6 @@ var descriptors = []Descriptor{
 	localOpenAI("ollama", "Ollama Local", "http://localhost:11434/v1", "llama3.1", "ollama local"),
 	localOpenAI("lmstudio", "LM Studio", "http://localhost:1234/v1", "local-model", "lm-studio", "lm studio"),
 	oauthProvider(openAICompat("openrouter", "OpenRouter", "https://openrouter.ai/api/v1", "openai/gpt-4.1", []string{"OPENROUTER_API_KEY"}), true, false),
-	openAICompat(AIMLAPIID, "AI/ML API", "https://api.aimlapi.com/v1", "openai/gpt-5-chat", []string{"AIMLAPI_API_KEY"}, "aiml api", "ai/ml api"),
 	// Hugging Face Inference Providers — OpenAI-compatible router at
 	// https://router.huggingface.co/v1 exposes hundreds of OSS models. OAuth
 	// requires a one-time app registration at huggingface.co/settings/applications/new

internal/providercatalog/catalog_test.go

@@ -16,7 +16,6 @@ var expectedCatalogIDs = []string{
 	"ollama",
 	"lmstudio",
 	"openrouter",
-	"aimlapi",
 	"huggingface",
 	"chatgpt",
 	"groq",
@@ -103,28 +102,6 @@ func TestRecommendedProviderEndpoint(t *testing.T) {
 	}
 }
 
-func TestAIMLAPIDescriptor(t *testing.T) {
-	descriptor, err := Require("aimlapi")
-	if err != nil {
-		t.Fatalf("Require(aimlapi) error = %v", err)
-	}
-	if descriptor.Name != "AI/ML API" {
-		t.Fatalf("Name = %q, want AI/ML API", descriptor.Name)
-	}
-	if descriptor.DefaultBaseURL != "https://api.aimlapi.com/v1" {
-		t.Fatalf("DefaultBaseURL = %q, want AI/ML API endpoint", descriptor.DefaultBaseURL)
-	}
-	if descriptor.DefaultModel != "openai/gpt-5-chat" {
-		t.Fatalf("DefaultModel = %q, want openai/gpt-5-chat", descriptor.DefaultModel)
-	}
-	if descriptor.Transport != TransportOpenAICompatible {
-		t.Fatalf("Transport = %q, want %q", descriptor.Transport, TransportOpenAICompatible)
-	}
-	if !reflect.DeepEqual(descriptor.AuthEnvVars, []string{"AIMLAPI_API_KEY"}) {
-		t.Fatalf("AuthEnvVars = %#v, want AIMLAPI_API_KEY", descriptor.AuthEnvVars)
-	}
-}
-
 func TestLongCatDescriptor(t *testing.T) {
 	descriptor, err := Require("longcat")
 	if err != nil {
@@ -315,7 +292,7 @@ func TestListByTransportPreservesCatalogOrder(t *testing.T) {
 		TransportBedrock:         {"bedrock"},
 		TransportVertex:          {"vertex"},
 		TransportAnthropicCompat: {"minimax", "minimaxi-cn", "custom-anthropic-compatible"},
-		TransportOpenAICompat:    {"gitlawb-opengateway", "ollama-cloud", "ollama", "lmstudio", "openrouter", "aimlapi", "huggingface", "chatgpt", "groq", "deepseek", "together", "dashscope", "moonshot", "longcat", "nvidia-nim", "mistral", "github", "xai", "venice", "xiaomi-mimo", "bankr", "zai", "zai-cn", "kilocode", "opencode", "opencode-go", "atomic-chat", "chatgpt-proxy", "custom-openai-compatible"},
+		TransportOpenAICompat:    {"gitlawb-opengateway", "ollama-cloud", "ollama", "lmstudio", "openrouter", "huggingface", "chatgpt", "groq", "deepseek", "together", "dashscope", "moonshot", "longcat", "nvidia-nim", "mistral", "github", "xai", "venice", "xiaomi-mimo", "bankr", "zai", "zai-cn", "kilocode", "opencode", "opencode-go", "atomic-chat", "chatgpt-proxy", "custom-openai-compatible"},
 	}
 
 	for transport, wantIDs := range cases {

internal/providerhealth/connectivity_client_test.go

@@ -169,8 +169,8 @@ func TestValidateEndpointAllowsLoopbackOnlyForLocalProvider(t *testing.T) {
 		if err := validateEndpoint(ctx, ep, r, true); err != nil {
 			t.Errorf("local provider endpoint %q should be allowed with allowLoopback=true, got %v", ep, err)
 		}
-		if !endpointIsLoopback(ep) {
-			t.Errorf("endpointIsLoopback(%q) = false, want true", ep)
+		if !endpointIsLoopbackOrPrivate(ep) {
+			t.Errorf("endpointIsLoopbackOrPrivate(%q) = false, want true", ep)
 		}
 	}
 
@@ -179,14 +179,27 @@ func TestValidateEndpointAllowsLoopbackOnlyForLocalProvider(t *testing.T) {
 		t.Error("loopback must remain blocked when allowLoopback=false (redirects / non-local)")
 	}
 
-	// allowLoopback relaxes loopback ONLY — other special ranges (cloud metadata,
-	// link-local, private) stay blocked even for a local provider config.
-	for _, ep := range []string{"http://169.254.169.254/latest/meta-data", "http://10.0.0.5:8080/v1"} {
+	// allowLoopbackOrPrivate relaxes loopback AND private-network ranges — but
+	// link-local and other special ranges (cloud metadata, documentation) stay
+	// blocked even for a local provider config.
+	for _, ep := range []string{"http://169.254.169.254/latest/meta-data", "http://192.0.2.1:8080/v1"} {
 		if err := validateEndpoint(ctx, ep, staticResolver{addr: netip.MustParseAddr("169.254.169.254")}, true); err == nil {
-			t.Errorf("non-loopback special address %q must stay blocked even with allowLoopback=true", ep)
+			t.Errorf("special-use address %q must stay blocked even with flag=true", ep)
 		}
-		if endpointIsLoopback(ep) {
-			t.Errorf("endpointIsLoopback(%q) = true, want false", ep)
+		if endpointIsLoopbackOrPrivate(ep) {
+			t.Errorf("endpointIsLoopbackOrPrivate(%q) = true, want false", ep)
+		}
+	}
+
+	// Private-network addresses (192.168.x.y / 10.x.y.z / 172.16.x.y) are
+	// now allowed when the flag is true — the user is pointing at their own
+	// LAN box (e.g. llama.cpp on another local machine).
+	for _, ep := range []string{"http://192.168.1.100:8080/v1", "http://10.0.0.5:8080/v1", "http://172.16.0.10:8080/v1"} {
+		if err := validateEndpoint(ctx, ep, staticResolver{addr: netip.MustParseAddr("192.168.1.100")}, true); err != nil {
+			t.Errorf("private-network address %q should be allowed with flag=true, got %v", ep, err)
+		}
+		if !endpointIsLoopbackOrPrivate(ep) {
+			t.Errorf("endpointIsLoopbackOrPrivate(%q) = false, want true", ep)
 		}
 	}
 }

internal/providerhealth/providerhealth.go

@@ -269,7 +269,7 @@ func connectivityCheck(ctx context.Context, profile config.ProviderProfile, kind
 	requestCtx, cancel := context.WithTimeout(ctx, timeout)
 	defer cancel()
 
-	request, allowLoopback, err := healthRequest(requestCtx, profile, kind, options)
+	request, allowLoopbackOrPrivate, err := healthRequest(requestCtx, profile, kind, options)
 	if err != nil {
 		category := CategoryConfig
 		var safety endpointSafetyError
@@ -280,7 +280,7 @@ func connectivityCheck(ctx context.Context, profile config.ProviderProfile, kind
 	}
 	client := options.HTTPClient
 	if client == nil {
-		client = newConnectivityClient(timeout, options.Resolver, sensitiveAuthHeaderNames(profile, kind), allowLoopback)
+		client = newConnectivityClient(timeout, options.Resolver, sensitiveAuthHeaderNames(profile, kind), allowLoopbackOrPrivate)
 	}
 	response, err := client.Do(request)
 	if err != nil {
@@ -332,8 +332,8 @@ func healthRequest(ctx context.Context, profile config.ProviderProfile, kind con
 	// Loopback is permitted only when the user's OWN configured base_url is loopback
 	// (a local provider like Ollama/LM Studio) — never for a redirect target. This is
 	// the flag returned to the caller so the dial path applies the same policy. (AUDIT-H1)
-	allowLoopback := endpointIsLoopback(endpoint)
-	if err := validateEndpoint(ctx, endpoint, options.Resolver, allowLoopback); err != nil {
+	allowLoopbackOrPrivate := endpointIsLoopbackOrPrivate(endpoint)
+	if err := validateEndpoint(ctx, endpoint, options.Resolver, allowLoopbackOrPrivate); err != nil {
 		return nil, false, err
 	}
 	request, err := http.NewRequestWithContext(ctx, http.MethodGet, endpoint, nil)
@@ -344,7 +344,7 @@ func healthRequest(ctx context.Context, profile config.ProviderProfile, kind con
 		request.Header.Set("User-Agent", options.UserAgent)
 	}
 	applyAuth(request, profile, kind)
-	return request, allowLoopback, nil
+	return request, allowLoopbackOrPrivate, nil
 }
 
 func resolvedBaseURL(profile config.ProviderProfile, kind config.ProviderKind) (string, error) {
@@ -367,11 +367,11 @@ func resolvedBaseURL(profile config.ProviderProfile, kind config.ProviderKind) (
 }
 
 // validateEndpoint enforces the SSRF policy for the provider connectivity probe.
-// allowLoopback permits loopback hosts (localhost / 127.0.0.0/8 / ::1) ONLY — it is
-// set true exclusively for the user's own explicitly-configured local provider
-// base_url (e.g. Ollama/LM Studio), never for redirect targets, and it relaxes
-// loopback alone (other private/special ranges, incl. cloud metadata, stay blocked).
-func validateEndpoint(ctx context.Context, endpoint string, resolver Resolver, allowLoopback bool) error {
+// allowLoopbackOrPrivate permits loopback hosts (localhost / 127.0.0.0/8 / ::1)
+// and private-network hosts (192.168.0.0/16 / 10.0.0.0/8 / 172.16.0.0/12) — it
+// is set true exclusively for the user's own explicitly-configured base_url
+// (e.g. Ollama/LM Studio on the LAN), never for redirect targets.
+func validateEndpoint(ctx context.Context, endpoint string, resolver Resolver, allowLoopbackOrPrivate bool) error {
 	parsed, err := url.Parse(endpoint)
 	if err != nil {
 		return err
@@ -385,13 +385,13 @@ func validateEndpoint(ctx context.Context, endpoint string, resolver Resolver, a
 	}
 	normalized := strings.TrimSuffix(strings.ToLower(host), ".")
 	if normalized == "localhost" || strings.HasSuffix(normalized, ".localhost") {
-		if allowLoopback {
+		if allowLoopbackOrPrivate {
 			return nil // user-configured local provider; loopback is intentional
 		}
 		return endpointSafetyError{message: "provider connectivity URL is unsafe: localhost hosts are blocked"}
 	}
 	if addr, err := netip.ParseAddr(normalized); err == nil {
-		if reason := blockedAddrReason(addr); reason != "" && !(allowLoopback && addr.IsLoopback()) {
+		if reason := blockedAddrReason(addr); reason != "" && !(allowLoopbackOrPrivate && (addr.IsLoopback() || addr.IsPrivate())) {
 			return endpointSafetyError{message: "provider connectivity URL is unsafe: " + reason}
 		}
 		return nil
@@ -407,17 +407,19 @@ func validateEndpoint(ctx context.Context, endpoint string, resolver Resolver, a
 		return endpointSafetyError{message: "provider connectivity host resolved to no addresses"}
 	}
 	for _, addr := range addrs {
-		if reason := blockedAddrReason(addr); reason != "" && !(allowLoopback && addr.IsLoopback()) {
+		if reason := blockedAddrReason(addr); reason != "" && !(allowLoopbackOrPrivate && (addr.IsLoopback() || addr.IsPrivate())) {
 			return endpointSafetyError{message: "provider connectivity URL is unsafe: " + reason}
 		}
 	}
 	return nil
 }
 
-// endpointIsLoopback reports whether the configured endpoint's host is a loopback
-// host the user typed themselves (localhost or a 127.0.0.0/8 / ::1 literal). Used to
-// decide whether the connectivity probe may reach the user's own local provider.
-func endpointIsLoopback(endpoint string) bool {
+// endpointIsLoopbackOrPrivate reports whether the configured endpoint's host is a loopback
+// or private-network host the user typed themselves — a local provider on the LAN
+// (e.g. 192.168.x.y / 10.x.y.z) or localhost. Used to decide whether the
+// connectivity probe may relax address-blocking and reach the user's own
+// explicitly-configured provider.
+func endpointIsLoopbackOrPrivate(endpoint string) bool {
 	parsed, err := url.Parse(endpoint)
 	if err != nil {
 		return false
@@ -427,7 +429,7 @@ func endpointIsLoopback(endpoint string) bool {
 		return true
 	}
 	if addr, err := netip.ParseAddr(host); err == nil {
-		return addr.IsLoopback()
+		return addr.IsLoopback() || addr.IsPrivate()
 	}
 	return false
 }
@@ -446,7 +448,7 @@ const maxConnectivityRedirects = 5
 //     time and then dials the validated IP literal, closing the TOCTOU window
 //     between the pre-flight validateEndpoint check and the actual connection
 //     (DNS rebinding).
-func newConnectivityClient(timeout time.Duration, resolver Resolver, sensitiveHeaders []string, allowLoopback bool) *http.Client {
+func newConnectivityClient(timeout time.Duration, resolver Resolver, sensitiveHeaders []string, allowLoopbackOrPrivate bool) *http.Client {
 	if resolver == nil {
 		resolver = defaultResolver{}
 	}
@@ -456,7 +458,7 @@ func newConnectivityClient(timeout time.Duration, resolver Resolver, sensitiveHe
 	} else {
 		transport = &http.Transport{}
 	}
-	transport.DialContext = safeDialContext(resolver, allowLoopback)
+	transport.DialContext = safeDialContext(resolver, allowLoopbackOrPrivate)
 	return &http.Client{
 		Timeout:   timeout,
 		Transport: transport,
@@ -476,8 +478,8 @@ func newConnectivityClient(timeout time.Duration, resolver Resolver, sensitiveHe
 				}
 			}
 			// Redirects are NEVER allowed to reach loopback, even for a local provider:
-			// allowLoopback covers only the user's own configured base_url, not a 3xx
-			// target (which could be attacker-controlled SSRF). (AUDIT-H1)
+			// allowLoopbackOrPrivate covers only the user's own configured base_url,
+			// not a 3xx target (which could be attacker-controlled SSRF). (AUDIT-H1)
 			return validateEndpoint(req.Context(), req.URL.String(), resolver, false)
 		},
 	}
@@ -519,7 +521,7 @@ func sensitiveAuthHeaderNames(profile config.ProviderProfile, kind config.Provid
 // safeDialContext returns a dial function that resolves the target host, refuses
 // the connection if any resolved address is blocked, then dials the validated IP
 // literal so the kernel cannot re-resolve to a different address after the check.
-func safeDialContext(resolver Resolver, allowLoopback bool) func(context.Context, string, string) (net.Conn, error) {
+func safeDialContext(resolver Resolver, allowLoopbackOrPrivate bool) func(context.Context, string, string) (net.Conn, error) {
 	dialer := &net.Dialer{Timeout: 10 * time.Second, KeepAlive: 30 * time.Second}
 	return func(ctx context.Context, network, address string) (net.Conn, error) {
 		host, port, err := net.SplitHostPort(address)
@@ -527,7 +529,7 @@ func safeDialContext(resolver Resolver, allowLoopback bool) func(context.Context
 			return nil, err
 		}
 		if addr, parseErr := netip.ParseAddr(host); parseErr == nil {
-			if reason := blockedAddrReason(addr); reason != "" && !(allowLoopback && addr.IsLoopback()) {
+			if reason := blockedAddrReason(addr); reason != "" && !(allowLoopbackOrPrivate && (addr.IsLoopback() || addr.IsPrivate())) {
 				return nil, endpointSafetyError{message: "provider connectivity URL is unsafe: " + reason}
 			}
 			return dialer.DialContext(ctx, network, address)
@@ -540,7 +542,7 @@ func safeDialContext(resolver Resolver, allowLoopback bool) func(context.Context
 			return nil, endpointSafetyError{message: "provider connectivity host resolved to no addresses"}
 		}
 		for _, addr := range addrs {
-			if reason := blockedAddrReason(addr); reason != "" && !(allowLoopback && addr.IsLoopback()) {
+			if reason := blockedAddrReason(addr); reason != "" && !(allowLoopbackOrPrivate && (addr.IsLoopback() || addr.IsPrivate())) {
 				return nil, endpointSafetyError{message: "provider connectivity URL is unsafe: " + reason}
 			}
 		}

internal/providermodeldiscovery/discovery.go

@@ -195,7 +195,7 @@ func discoverOpenAIModels(ctx context.Context, profile config.ProviderProfile, o
 		AuthHeader:        profile.AuthHeader,
 		AuthScheme:        profile.AuthScheme,
 		AuthHeaderValue:   profile.AuthHeaderValue,
-		CustomHeaders:     providerio.CopyHeaders(profile.CustomHeaders),
+		CustomHeaders:     profile.CustomHeaders,
 	}, nil)
 }
 

internal/providermodeldiscovery/discovery_test.go

@@ -52,44 +52,6 @@ func TestDiscoverOpenAICompatibleModelsFetchesModelsEndpoint(t *testing.T) {
 	}
 }
 
-func TestDiscoverAIMLAPIModelsSendsAuthAndCustomHeadersWithoutAttribution(t *testing.T) {
-	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		for header, want := range map[string]string{
-			"Authorization": "Bearer test-key",
-			"X-Trace":       "test",
-		} {
-			if got := r.Header.Get(header); got != want {
-				t.Errorf("%s = %q, want %q", header, got, want)
-			}
-		}
-		// No first-party referral/attribution headers are injected for catalog
-		// presets; aimlapi rides through CopyHeaders like every other provider.
-		for _, header := range []string{
-			"X-AIMLAPI-Partner-ID",
-			"X-AIMLAPI-Integration-Repo",
-			"X-AIMLAPI-Integration-Version",
-		} {
-			if got := r.Header.Get(header); got != "" {
-				t.Errorf("%s = %q, want no attribution header", header, got)
-			}
-		}
-		w.Header().Set("Content-Type", "application/json")
-		_, _ = w.Write([]byte(`{"data":[{"id":"openai/gpt-5-chat"}]}`))
-	}))
-	defer server.Close()
-
-	_, err := Discover(context.Background(), config.ProviderProfile{
-		CatalogID:     "aimlapi",
-		ProviderKind:  config.ProviderKindOpenAICompatible,
-		BaseURL:       server.URL + "/v1",
-		APIKey:        "test-key",
-		CustomHeaders: map[string]string{"X-Trace": "test"},
-	}, Options{HTTPClient: server.Client()})
-	if err != nil {
-		t.Fatalf("Discover returned error: %v", err)
-	}
-}
-
 func TestDiscoverOpenAICompatibleModelsHonorsAuthHeaderValue(t *testing.T) {
 	// A profile can authenticate via a raw auth-header value instead of APIKey;
 	// discovery must send it rather than probe unauthenticated.

internal/providers/anthropic/cache_breakpoints_test.go

@@ -0,0 +1,96 @@
+package anthropic
+
+import (
+	"context"
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+// The final two messages must carry a cache_control breakpoint on their last
+// cacheable block so the conversation transcript — not just system prompt and
+// tools — is a prompt-cache hit on the next turn. Earlier messages must stay
+// unmarked (Anthropic caps breakpoints at 4 per request: system, tools, and
+// these two).
+func TestAnthropicRequestMarksLastTwoMessagesForCaching(t *testing.T) {
+	var gotBody map[string]any
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if err := json.NewDecoder(r.Body).Decode(&gotBody); err != nil {
+			t.Fatalf("decode request body: %v", err)
+		}
+		writeSSEEvent(w, "message_stop", `{"type":"message_stop"}`)
+	}))
+	defer server.Close()
+
+	provider, err := New(Options{BaseURL: server.URL + "/", Model: "claude-test"})
+	if err != nil {
+		t.Fatalf("New returned error: %v", err)
+	}
+	stream, err := provider.StreamCompletion(context.Background(), zeroruntime.CompletionRequest{
+		Messages: []zeroruntime.Message{
+			{Role: zeroruntime.MessageRoleUser, Content: "first"},
+			{
+				Role:      zeroruntime.MessageRoleAssistant,
+				Content:   "calling a tool",
+				ToolCalls: []zeroruntime.ToolCall{{ID: "toolu_1", Name: "grep", Arguments: `{}`}},
+			},
+			{Role: zeroruntime.MessageRoleTool, Content: "result", ToolCallID: "toolu_1"},
+		},
+	})
+	if err != nil {
+		t.Fatalf("StreamCompletion returned error: %v", err)
+	}
+	drain(stream)
+
+	messages := gotBody["messages"].([]any)
+	if len(messages) != 3 {
+		t.Fatalf("expected 3 wire messages, got %d: %#v", len(messages), messages)
+	}
+	lastBlockHasCache := func(message any) bool {
+		blocks, ok := message.(map[string]any)["content"].([]any)
+		if !ok || len(blocks) == 0 {
+			return false
+		}
+		last := blocks[len(blocks)-1].(map[string]any)
+		control, ok := last["cache_control"].(map[string]any)
+		return ok && control["type"] == "ephemeral"
+	}
+	if lastBlockHasCache(messages[0]) {
+		t.Fatalf("first message must not carry a breakpoint: %#v", messages[0])
+	}
+	if !lastBlockHasCache(messages[1]) {
+		t.Fatalf("second-to-last message must carry a breakpoint: %#v", messages[1])
+	}
+	if !lastBlockHasCache(messages[2]) {
+		t.Fatalf("last message must carry a breakpoint: %#v", messages[2])
+	}
+}
+
+// A string-content message must be converted to a block array so it can carry
+// the breakpoint, and thinking blocks must never carry cache_control (the API
+// rejects them) — the marker goes on the last cacheable block instead.
+func TestApplyMessageCacheBreakpointsSkipsThinkingBlocks(t *testing.T) {
+	messages := []anthropicMessage{
+		{Role: "user", Content: "plain string"},
+		{Role: "assistant", Content: []map[string]any{
+			{"type": "text", "text": "answer"},
+			{"type": "thinking", "thinking": "hmm", "signature": "sig"},
+		}},
+	}
+	applyMessageCacheBreakpoints(messages)
+
+	userBlocks := contentBlocks(messages[0].Content)
+	if len(userBlocks) != 1 || userBlocks[0]["cache_control"] == nil {
+		t.Fatalf("string content must become a marked block: %#v", messages[0].Content)
+	}
+	assistantBlocks := contentBlocks(messages[1].Content)
+	if assistantBlocks[1]["cache_control"] != nil {
+		t.Fatalf("thinking block must not be marked: %#v", assistantBlocks[1])
+	}
+	if assistantBlocks[0]["cache_control"] == nil {
+		t.Fatalf("text block must carry the breakpoint instead: %#v", assistantBlocks[0])
+	}
+}

internal/providers/anthropic/provider.go

@@ -405,9 +405,35 @@ func (provider *Provider) anthropicRequest(request zeroruntime.CompletionRequest
 		}
 		mapped.Tools[len(mapped.Tools)-1].CacheControl = &cacheControl{Type: cacheEphemeral}
 	}
+	applyMessageCacheBreakpoints(mapped.Messages)
 	return mapped, nil
 }
 
+// applyMessageCacheBreakpoints marks the last content block of the final two
+// messages with cache_control so the conversation transcript is cached
+// turn-over-turn, not just the system prompt and tool definitions. Two message
+// breakpoints (not one) keep the previous turn's prefix a cache hit while the
+// newest suffix is being written. Anthropic allows at most 4 breakpoints per
+// request: system + tools use two above, these use the remaining two. Thinking
+// blocks cannot carry cache_control, so the marker goes on the last block that
+// can.
+func applyMessageCacheBreakpoints(messages []anthropicMessage) {
+	marked := 0
+	for i := len(messages) - 1; i >= 0 && marked < 2; i-- {
+		blocks := contentBlocks(messages[i].Content)
+		for j := len(blocks) - 1; j >= 0; j-- {
+			blockType, _ := blocks[j]["type"].(string)
+			if blockType == "thinking" || blockType == "redacted_thinking" {
+				continue
+			}
+			blocks[j]["cache_control"] = map[string]any{"type": cacheEphemeral}
+			messages[i].Content = blocks
+			marked++
+			break
+		}
+	}
+}
+
 func mapMessages(messages []zeroruntime.Message) (string, []anthropicMessage, error) {
 	systemParts := []string{}
 	mapped := []anthropicMessage{}

internal/providers/factory.go

@@ -24,7 +24,7 @@ type Options struct {
 	ModelRegistry *modelregistry.Registry
 	// OAuthResolver, when set, lets the provider authenticate model calls with an
 	// OAuth bearer token (preferred over the API key). nil => API-key auth only.
-	// Applied to the OpenAI and Anthropic providers.
+	// Applied to the OpenAI, Anthropic, and Google (Gemini) providers.
 	OAuthResolver providerio.TokenResolver
 	// OAuthLoginKey is the credential-store key OAuthResolver bound to (empty when
 	// there is no OAuth login). It is the SAME selection the bearer resolver made,
@@ -61,7 +61,7 @@ func New(profile config.ProviderProfile, options Options) (zeroruntime.Provider,
 			AuthHeader:      profile.AuthHeader,
 			AuthScheme:      profile.AuthScheme,
 			AuthHeaderValue: profile.AuthHeaderValue,
-			CustomHeaders:   providerio.CopyHeaders(profile.CustomHeaders),
+			CustomHeaders:   profile.CustomHeaders,
 			OAuthResolver:   options.OAuthResolver,
 			MaxTokens:       resolved.maxOutputTokens,
 			HTTPClient:      options.HTTPClient,
@@ -91,6 +91,7 @@ func New(profile config.ProviderProfile, options Options) (zeroruntime.Provider,
 			AuthScheme:      profile.AuthScheme,
 			AuthHeaderValue: profile.AuthHeaderValue,
 			CustomHeaders:   profile.CustomHeaders,
+			OAuthResolver:   options.OAuthResolver,
 			MaxTokens:       resolved.maxOutputTokens,
 			HTTPClient:      options.HTTPClient,
 			UserAgent:       options.UserAgent,

internal/providers/factory_test.go

@@ -95,55 +95,6 @@ func TestNewThreadsCustomProviderHeaders(t *testing.T) {
 	}
 }
 
-func TestNewAIMLAPIProviderSendsEndpointAndAuthWithoutAttribution(t *testing.T) {
-	transport := &captureTransport{responseBody: "data: [DONE]\n\n"}
-	provider, err := New(config.ProviderProfile{
-		Name:          "aimlapi",
-		CatalogID:     "aimlapi",
-		ProviderKind:  config.ProviderKindOpenAICompatible,
-		APIKey:        "aimlapi-test-key",
-		Model:         "openai/gpt-5-chat",
-		CustomHeaders: map[string]string{"X-Trace": "test"},
-	}, Options{HTTPClient: &http.Client{Transport: transport}})
-	if err != nil {
-		t.Fatalf("New() error = %v", err)
-	}
-	stream, err := provider.StreamCompletion(context.Background(), zeroruntime.CompletionRequest{
-		Messages: []zeroruntime.Message{{Role: zeroruntime.MessageRoleUser, Content: "hello"}},
-	})
-	if err != nil {
-		t.Fatalf("StreamCompletion() error = %v", err)
-	}
-	for range stream {
-	}
-
-	if transport.request == nil {
-		t.Fatal("HTTP client was not used")
-	}
-	if got := transport.request.URL.String(); got != "https://api.aimlapi.com/v1/chat/completions" {
-		t.Fatalf("request URL = %q, want AI/ML API endpoint", got)
-	}
-	for header, want := range map[string]string{
-		"Authorization": "Bearer aimlapi-test-key",
-		"X-Trace":       "test",
-	} {
-		if got := transport.request.Header.Get(header); got != want {
-			t.Fatalf("%s = %q, want %q", header, got, want)
-		}
-	}
-	// No first-party referral/attribution headers are injected for catalog
-	// presets; aimlapi rides through CopyHeaders like every other provider.
-	for _, header := range []string{
-		"X-AIMLAPI-Partner-ID",
-		"X-AIMLAPI-Integration-Repo",
-		"X-AIMLAPI-Integration-Version",
-	} {
-		if got := transport.request.Header.Get(header); got != "" {
-			t.Fatalf("%s = %q, want no attribution header", header, got)
-		}
-	}
-}
-
 func TestNewSupportsOpenAIProviderKind(t *testing.T) {
 	provider, err := New(config.ProviderProfile{
 		Name:         "openai",

internal/providers/gemini/provider.go

@@ -49,6 +49,10 @@ type Options struct {
 	CustomHeaders   map[string]string
 	HTTPClient      *http.Client
 	UserAgent       string
+	// OAuthResolver, when set, supplies an OAuth bearer credential per request and
+	// is retried once with a forced token refresh after an upstream 401 (matching
+	// the OpenAI and Anthropic providers). Nil falls back to plain API-key auth.
+	OAuthResolver providerio.TokenResolver
 	// StreamIdleTimeout aborts the stream if no data arrives for this long.
 	// When unset, Zero uses providerio.ResolveStreamIdleTimeout — the
 	// ZERO_STREAM_IDLE_TIMEOUT override or providerio.DefaultStreamIdleTimeout.
@@ -67,6 +71,7 @@ type Provider struct {
 	customHeaders     map[string]string
 	httpClient        *http.Client
 	userAgent         string
+	oauthResolver     providerio.TokenResolver
 	streamIdleTimeout time.Duration
 }
 
@@ -95,6 +100,7 @@ func New(options Options) (*Provider, error) {
 		customHeaders:     providerio.CopyHeaders(options.CustomHeaders),
 		httpClient:        providerio.HTTPClient(options.HTTPClient),
 		userAgent:         options.UserAgent,
+		oauthResolver:     options.OAuthResolver,
 		streamIdleTimeout: providerio.ResolveStreamIdleTimeout(options.StreamIdleTimeout),
 	}, nil
 }
@@ -127,20 +133,22 @@ func (provider *Provider) stream(ctx context.Context, body []byte, events chan<-
 	streamCtx, cancelStream := context.WithCancel(ctx)
 	defer cancelStream()
 
-	response, err := providerio.SendWithRetry(streamCtx, provider.httpClient, http.MethodPost, provider.streamURL(), body, func(request *http.Request) {
-		request.Header.Set("Content-Type", "application/json")
-		if provider.userAgent != "" {
-			request.Header.Set("User-Agent", provider.userAgent)
-		}
-		providerio.ApplyAuthHeaders(request, providerio.AuthHeaders{
+	response, err := providerio.SendWithAuthRetry(streamCtx, provider.httpClient, http.MethodPost, provider.streamURL(), body,
+		providerio.AuthHeaders{
 			APIKey:            provider.apiKey,
 			DefaultAuthHeader: "x-goog-api-key",
 			AuthHeader:        provider.authHeader,
 			AuthScheme:        provider.authScheme,
 			AuthHeaderValue:   provider.authHeaderValue,
 			CustomHeaders:     provider.customHeaders,
-		})
-	}, 0)
+		},
+		provider.oauthResolver,
+		func(request *http.Request) {
+			request.Header.Set("Content-Type", "application/json")
+			if provider.userAgent != "" {
+				request.Header.Set("User-Agent", provider.userAgent)
+			}
+		}, 0)
 	if err != nil {
 		providerio.SendEvent(ctx, events, zeroruntime.StreamEvent{Type: zeroruntime.StreamEventError, Error: provider.redact("provider stream error: " + err.Error())})
 		return

internal/providers/gemini/provider_test.go

@@ -370,6 +370,57 @@ func TestStreamCompletionClassifiesHTTPAndPromptBlockErrors(t *testing.T) {
 	}
 }
 
+// A 401 with an OAuth resolver is retried once with a force-refreshed token; the
+// replayed request carries the refreshed bearer and succeeds.
+func TestStreamCompletionRetries401WithRefreshedToken(t *testing.T) {
+	var attempts int
+	var secondAuth string
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		attempts++
+		if attempts == 1 {
+			http.Error(w, `{"error":{"message":"token expired"}}`, http.StatusUnauthorized)
+			return
+		}
+		secondAuth = r.Header.Get("Authorization")
+		writeSSE(w, `{}`)
+	}))
+	defer server.Close()
+
+	var forceRefreshOnRetry bool
+	resolver := func(ctx context.Context, forceRefresh bool) (string, string, bool, error) {
+		if forceRefresh {
+			forceRefreshOnRetry = true
+			return "Authorization", "Bearer refreshed", true, nil
+		}
+		return "Authorization", "Bearer stale", true, nil
+	}
+
+	provider, err := New(Options{
+		APIKey:        "sk-google",
+		BaseURL:       server.URL,
+		Model:         "gemini-test",
+		OAuthResolver: resolver,
+	})
+	if err != nil {
+		t.Fatalf("New returned error: %v", err)
+	}
+	stream, err := provider.StreamCompletion(context.Background(), validRequest())
+	if err != nil {
+		t.Fatalf("StreamCompletion returned error: %v", err)
+	}
+	drain(stream)
+
+	if attempts != 2 {
+		t.Fatalf("attempts = %d, want 2 (initial 401 + one refreshed retry)", attempts)
+	}
+	if !forceRefreshOnRetry {
+		t.Fatalf("resolver was not called with forceRefresh on the retry")
+	}
+	if secondAuth != "Bearer refreshed" {
+		t.Fatalf("retry Authorization = %q, want refreshed bearer", secondAuth)
+	}
+}
+
 func TestStreamCompletionEmitsStreamErrorObject(t *testing.T) {
 	provider := newTestProviderWithKey(t, "sk-google", func(w http.ResponseWriter, r *http.Request) {
 		writeSSE(w, `{"error":{"code":429,"message":"stream failed sk-google","status":"RESOURCE_EXHAUSTED"}}`)

internal/providers/providerio/retry.go

@@ -23,12 +23,18 @@ import (
 // (billable) work. Only the INITIAL request is ever in scope; once the response
 // body starts streaming it is never re-issued.
 
-const defaultMaxRetryAttempts = 3
+const defaultMaxRetryAttempts = 6
 
 // maxBackoff caps a single backoff wait so a hostile or buggy Retry-After can't
 // stall the agent for minutes.
 const maxBackoff = 30 * time.Second
 
+// retryBackoffBase is the first wait when the server supplied no Retry-After.
+// Rate-limit windows are measured in seconds, not milliseconds: retrying a 429
+// after 400ms almost always burns the attempt while still limited, so the
+// schedule is 2s, 4s, 8s, 16s, then maxBackoff. A var so tests can shrink it.
+var retryBackoffBase = 2 * time.Second
+
 // SendWithRetry issues an HTTP request, retrying ONLY the safe-to-replay server
 // responses (429 and 503, see ShouldRetryStatus) up to maxAttempts — backing off
 // between tries and honoring a server Retry-After header and context
@@ -110,17 +116,11 @@ func ShouldRetryStatus(code int) bool {
 }
 
 // Backoff waits before retry attempt N (1-based), returning false if the context
-// is cancelled during the wait. The wait is attempt*400ms unless the server
-// supplied a (positive) Retry-After, and is capped at maxBackoff.
+// is cancelled during the wait. A server-supplied (positive) Retry-After wins;
+// otherwise the wait doubles from retryBackoffBase per attempt. Either way the
+// wait is capped at maxBackoff.
 func Backoff(ctx context.Context, attempt int, retryAfter time.Duration) bool {
-	wait := time.Duration(attempt) * 400 * time.Millisecond
-	if retryAfter > 0 {
-		wait = retryAfter
-	}
-	if wait > maxBackoff {
-		wait = maxBackoff
-	}
-	timer := time.NewTimer(wait)
+	timer := time.NewTimer(backoffWait(attempt, retryAfter))
 	defer timer.Stop()
 	select {
 	case <-ctx.Done():
@@ -130,6 +130,27 @@ func Backoff(ctx context.Context, attempt int, retryAfter time.Duration) bool {
 	}
 }
 
+// backoffWait computes the wait before retry attempt N (1-based): Retry-After
+// when supplied, else exponential from retryBackoffBase, both capped at
+// maxBackoff. The exponent is clamped so a large attempt count cannot overflow.
+func backoffWait(attempt int, retryAfter time.Duration) time.Duration {
+	wait := retryAfter
+	if wait <= 0 {
+		exponent := attempt - 1
+		if exponent > 5 {
+			exponent = 5
+		}
+		if exponent < 0 {
+			exponent = 0
+		}
+		wait = retryBackoffBase * time.Duration(1<<exponent)
+	}
+	if wait > maxBackoff {
+		wait = maxBackoff
+	}
+	return wait
+}
+
 // RetryAfter parses a response's Retry-After header (delay-seconds or an HTTP
 // date) into a positive duration, or 0 when absent/unparseable. The result is
 // capped at maxBackoff by Backoff.

internal/providers/providerio/retry_test.go

@@ -94,7 +94,40 @@ func TestBackoffWaitsThenReturnsTrue(t *testing.T) {
 	}
 }
 
+// shrinkBackoff makes retry waits negligible for the duration of a test.
+func shrinkBackoff(t *testing.T) {
+	t.Helper()
+	saved := retryBackoffBase
+	retryBackoffBase = time.Millisecond
+	t.Cleanup(func() { retryBackoffBase = saved })
+}
+
+func TestBackoffWaitSchedule(t *testing.T) {
+	// Without Retry-After the wait doubles per attempt from 2s and caps at 30s;
+	// a supplied Retry-After wins but is capped too.
+	cases := []struct {
+		attempt    int
+		retryAfter time.Duration
+		want       time.Duration
+	}{
+		{1, 0, 2 * time.Second},
+		{2, 0, 4 * time.Second},
+		{3, 0, 8 * time.Second},
+		{4, 0, 16 * time.Second},
+		{5, 0, 30 * time.Second},  // 32s capped
+		{50, 0, 30 * time.Second}, // clamped exponent, no overflow
+		{1, 7 * time.Second, 7 * time.Second},
+		{1, 5 * time.Minute, 30 * time.Second}, // hostile Retry-After capped
+	}
+	for _, c := range cases {
+		if got := backoffWait(c.attempt, c.retryAfter); got != c.want {
+			t.Errorf("backoffWait(%d, %v) = %v, want %v", c.attempt, c.retryAfter, got, c.want)
+		}
+	}
+}
+
 func TestSendWithRetryRetriesThenSucceeds(t *testing.T) {
+	shrinkBackoff(t)
 	var hits int32
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		if atomic.AddInt32(&hits, 1) == 1 {
@@ -140,6 +173,7 @@ func TestSendWithRetryReturnsNonRetryableImmediately(t *testing.T) {
 }
 
 func TestSendWithRetryReturnsLastResponseAfterMaxAttempts(t *testing.T) {
+	shrinkBackoff(t)
 	var hits int32
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		atomic.AddInt32(&hits, 1)

internal/sandbox/manager_test.go

@@ -72,6 +72,9 @@ func TestPermissionProfileFromPolicyIncludesDefaultTempWriteRoots(t *testing.T)
 	if !writeRootsContain(profile.FileSystem.WriteRoots, tmpdir) {
 		t.Fatalf("write roots = %#v, want temp root %q", profile.FileSystem.WriteRoots, tmpdir)
 	}
+	// /tmp is a default temp write root on POSIX only (see
+	// defaultTempWriteRootCandidatesForGOOS); on Windows the bare path resolves
+	// against the current drive, so a stray C:\tmp must not turn this on.
 	if runtime.GOOS != "windows" && pathExists("/tmp") && !writeRootsContain(profile.FileSystem.WriteRoots, "/tmp") {
 		t.Fatalf("write roots = %#v, want /tmp", profile.FileSystem.WriteRoots)
 	}

internal/sandbox/scope_test.go

@@ -129,6 +129,9 @@ func TestNewScopeNormalizesAndValidatesExtraRoots(t *testing.T) {
 	if !stringSliceContains(roots, normalizeWorkspaceRootBestEffort(extra)) {
 		t.Fatalf("Roots()=%v want extra root %q", roots, normalizeWorkspaceRootBestEffort(extra))
 	}
+	// /tmp is a default temp write root on POSIX only (see
+	// defaultTempWriteRootCandidatesForGOOS); on Windows the bare path resolves
+	// against the current drive, so a stray C:\tmp must not turn this on.
 	if runtime.GOOS != "windows" && pathExists("/tmp") && !stringSliceContains(roots, normalizeWorkspaceRootBestEffort("/tmp")) {
 		t.Fatalf("Roots()=%v want default /tmp write root", roots)
 	}

internal/sandbox/windows_process_windows.go

@@ -87,11 +87,20 @@ func windowsCommandLine(args []string) (string, error) {
 	if len(args) == 0 {
 		return "", errorsNewWindowsProcess("missing command")
 	}
-	out := make([]string, 0, len(args))
 	for _, arg := range args {
 		if strings.ContainsRune(arg, 0) {
 			return "", errorsNewWindowsProcess("command argument contains NUL")
 		}
+	}
+	// Zero's cmd.exe shell invocation needs its final element (the shell
+	// command text) reaching cmd.exe raw, not escaped like a normal argv
+	// element: see WindowsShellCommandLine for why. Every other command this
+	// runner might launch is escaped normally below.
+	if commandLine, ok := windowsShellCommandLineFromArgs(args); ok {
+		return commandLine, nil
+	}
+	out := make([]string, 0, len(args))
+	for _, arg := range args {
 		out = append(out, syscall.EscapeArg(arg))
 	}
 	return strings.Join(out, " "), nil

internal/sandbox/windows_runner.go

@@ -105,6 +105,56 @@ const (
 	WindowsSandboxLevelDisabled   WindowsSandboxLevel = "disabled"
 )
 
+// WindowsShellArgs returns cmd.exe's argv (after the executable name) for
+// running commandText as a shell command on Windows: no wrapping quotes and
+// no /S, so commandText reaches cmd.exe's own /C remainder parsing exactly as
+// written. See WindowsShellCommandLine for why that matters. Used both as
+// CommandSpec.Args for the plain (unwrapped) path and, via
+// windowsShellCommandLineFromArgs, to recognize this same shape once it has
+// round-tripped through the sandboxed runner's CLI argv.
+func WindowsShellArgs(commandText string) []string {
+	return []string{"/d", "/c", commandText}
+}
+
+// WindowsShellCommandLine builds the raw command line a Windows CreateProcess
+// call should receive to run commandText as a shell command: "cmd.exe /d /c "
+// followed by commandText completely unescaped.
+//
+// cmd.exe's own /C remainder parsing is not CommandLineToArgvW-compatible: it
+// applies its own quote-stripping heuristic (documented in `cmd /?`) whenever
+// the remainder starts with a literal quote, and that heuristic does not
+// understand backslash-escaped inner quotes the way a normal argv consumer
+// would. A command like `python -c "print(15 / 3)"` passed through the
+// standard exec.Cmd Args mechanism gets wrapped in an outer pair of quotes
+// (because it contains spaces) with its own quotes escaped as \" — exactly
+// the shape that heuristic corrupts, stripping the outer quotes but leaving
+// the literal backslashes behind. Passing commandText through raw, exactly as
+// the model wrote it, means cmd.exe parses it the same way it would if typed
+// directly at an interactive prompt: a leading quoted executable path like
+// `"C:\Program Files\foo.exe" arg` is preserved correctly by cmd.exe's own
+// quote-preserving special case, and a command with no leading quote at all,
+// like the python example above, is never touched by the stripping heuristic
+// in the first place.
+func WindowsShellCommandLine(commandText string) string {
+	return "cmd.exe /d /c " + commandText
+}
+
+// windowsShellCommandLineFromArgs reports whether args is exactly
+// ["cmd.exe"]+WindowsShellArgs(text) for some text and, if so, returns
+// WindowsShellCommandLine's result for that text. The sandboxed runner path
+// only has the deserialized argv (WindowsSandboxCommandConfig.Command) to
+// work from, not the original commandText, so it recovers the raw command
+// line by recognizing the shape instead.
+func windowsShellCommandLineFromArgs(args []string) (string, bool) {
+	if len(args) != 4 {
+		return "", false
+	}
+	if !strings.EqualFold(args[0], "cmd.exe") || args[1] != "/d" || args[2] != "/c" {
+		return "", false
+	}
+	return WindowsShellCommandLine(args[3]), true
+}
+
 type WindowsSandboxCommandArgsOptions struct {
 	SandboxHome       string
 	CommandCWD        string

internal/sandbox/windows_runner_test.go

@@ -147,3 +147,56 @@ func TestRunWindowsSandboxCommandRunnerRejectsInvalidArgs(t *testing.T) {
 		t.Fatalf("stderr = %q, want runner name", stderr.String())
 	}
 }
+
+func TestWindowsShellArgsHasNoSAndKeepsCommandTextWhole(t *testing.T) {
+	args := WindowsShellArgs(`python -c "print(15 / 3)"`)
+	want := []string{"/d", "/c", `python -c "print(15 / 3)"`}
+	if len(args) != len(want) {
+		t.Fatalf("args = %#v, want %#v", args, want)
+	}
+	for index := range want {
+		if args[index] != want[index] {
+			t.Fatalf("args = %#v, want %#v", args, want)
+		}
+	}
+}
+
+func TestWindowsShellCommandLineLeavesCommandTextUnescaped(t *testing.T) {
+	// The whole point of this function: commandText's own quotes must reach
+	// cmd.exe exactly as written, not re-quoted or backslash-escaped the way
+	// a normal argv-escaping function (like syscall.EscapeArg) would treat an
+	// argument containing spaces and embedded quotes.
+	commandText := `python -c "print(15 / 3)"`
+	got := WindowsShellCommandLine(commandText)
+	want := `cmd.exe /d /c python -c "print(15 / 3)"`
+	if got != want {
+		t.Fatalf("WindowsShellCommandLine = %q, want %q", got, want)
+	}
+}
+
+func TestWindowsShellCommandLineFromArgsRecognizesTheShellShape(t *testing.T) {
+	commandLine, ok := windowsShellCommandLineFromArgs([]string{"cmd.exe", "/d", "/c", `echo "hi"`})
+	if !ok {
+		t.Fatal("windowsShellCommandLineFromArgs did not recognize the shell shape")
+	}
+	if want := `cmd.exe /d /c echo "hi"`; commandLine != want {
+		t.Fatalf("commandLine = %q, want %q", commandLine, want)
+	}
+}
+
+func TestWindowsShellCommandLineFromArgsRejectsOtherShapes(t *testing.T) {
+	cases := [][]string{
+		nil,
+		{"cmd.exe"},
+		{"cmd.exe", "/d", "/c"},
+		{"cmd.exe", "/d", "/c", "echo hi", "extra"},
+		{"git.exe", "/d", "/c", "echo hi"},
+		{"cmd.exe", "/s", "/c", "echo hi"},
+		{"cmd.exe", "/d", "/k", "echo hi"},
+	}
+	for _, args := range cases {
+		if _, ok := windowsShellCommandLineFromArgs(args); ok {
+			t.Fatalf("windowsShellCommandLineFromArgs(%#v) matched, want no match", args)
+		}
+	}
+}

internal/sandbox/windows_unelevated.go

@@ -61,8 +61,10 @@ func buildWindowsUnelevatedAppliedPlan(config WindowsSandboxCommandConfig) (Wind
 
 // loadWindowsUnelevatedSetupMarker reads the marker; a missing file is the
 // common first-run state and returns an empty marker, not an error. A marker
-// with an unexpected schema version is treated as empty so the runner re-applies
-// (and rewrites) rather than failing.
+// with an unexpected schema version or a corrupt/truncated body is likewise
+// treated as empty so the runner re-applies (and rewrites) rather than failing:
+// the marker only memoizes idempotent ACL applies, so resetting it can only
+// cost redundant work, never skip enforcement.
 func loadWindowsUnelevatedSetupMarker(sandboxHome string) (WindowsUnelevatedSetupMarker, error) {
 	sandboxHome = strings.TrimSpace(sandboxHome)
 	if sandboxHome == "" {
@@ -78,7 +80,7 @@ func loadWindowsUnelevatedSetupMarker(sandboxHome string) (WindowsUnelevatedSetu
 	}
 	var marker WindowsUnelevatedSetupMarker
 	if err := json.Unmarshal(bytes, &marker); err != nil {
-		return WindowsUnelevatedSetupMarker{}, fmt.Errorf("parse windows unelevated setup marker %s: %w", path, err)
+		return WindowsUnelevatedSetupMarker{SchemaVersion: windowsUnelevatedSetupMarkerSchemaVersion}, nil
 	}
 	if marker.SchemaVersion != windowsUnelevatedSetupMarkerSchemaVersion {
 		return WindowsUnelevatedSetupMarker{SchemaVersion: windowsUnelevatedSetupMarkerSchemaVersion}, nil

internal/sandbox/windows_unelevated_test.go

@@ -138,6 +138,38 @@ func TestWindowsUnelevatedSetupMarkerEvictsOldestPastCap(t *testing.T) {
 	}
 }
 
+func TestLoadWindowsUnelevatedSetupMarkerSelfHealsCorruptFile(t *testing.T) {
+	home := filepath.Join(t.TempDir(), ".zero-sandbox")
+	if err := os.MkdirAll(home, 0o700); err != nil {
+		t.Fatalf("mkdir sandbox home: %v", err)
+	}
+	path := WindowsUnelevatedSetupMarkerPath(home)
+	// A truncated body must reset like an unknown schema, not brick every
+	// unelevated command until the file is hand-deleted.
+	if err := os.WriteFile(path, []byte(`{"schemaVersion":1,"appliedPlans":[{"aclPlanHash":"h`), 0o600); err != nil {
+		t.Fatalf("write corrupt marker: %v", err)
+	}
+	marker, err := loadWindowsUnelevatedSetupMarker(home)
+	if err != nil {
+		t.Fatalf("corrupt marker must self-heal, got error: %v", err)
+	}
+	if len(marker.AppliedPlans) != 0 || marker.SchemaVersion != windowsUnelevatedSetupMarkerSchemaVersion {
+		t.Fatalf("corrupt marker must reset to empty current-schema marker, got %+v", marker)
+	}
+	// The healed marker must accept new recordings over the corrupt file.
+	applied := WindowsUnelevatedAppliedPlan{ACLPlanHash: "hash-heal", ACLPlanEntries: 2}
+	if err := recordWindowsUnelevatedAppliedPlan(home, applied); err != nil {
+		t.Fatalf("record over corrupt marker: %v", err)
+	}
+	marker, err = loadWindowsUnelevatedSetupMarker(home)
+	if err != nil {
+		t.Fatalf("reload healed marker: %v", err)
+	}
+	if !marker.contains(applied) {
+		t.Fatalf("healed marker = %+v, want the recorded plan", marker)
+	}
+}
+
 func TestLoadWindowsUnelevatedSetupMarkerResetsUnknownSchema(t *testing.T) {
 	home := filepath.Join(t.TempDir(), ".zero-sandbox")
 	if err := os.MkdirAll(home, 0o700); err != nil {

internal/sessions/exec_session.go

@@ -160,10 +160,7 @@ func FormatExecPrompt(prompt string, prepared PreparedExec) string {
 	if prepared.Mode == ModeNew || len(prepared.ContextEvents) == 0 {
 		return prompt
 	}
-	events := prepared.ContextEvents
-	if len(events) > 20 {
-		events = events[len(events)-20:]
-	}
+	events := promptContextEvents(prepared.ContextEvents)
 
 	lines := []string{}
 	for _, event := range events {
@@ -187,6 +184,25 @@ func FormatExecPrompt(prompt string, prepared PreparedExec) string {
 	}, "\n")
 }
 
+func promptContextEvents(events []Event) []Event {
+	const maxPromptContextEvents = 80
+
+	filtered := make([]Event, 0, len(events))
+	for _, event := range events {
+		switch event.Type {
+		case EventMessage, EventCompaction, EventSessionFork, EventSessionChild, EventSpecialistStart, EventSpecialistStop, EventError:
+			filtered = append(filtered, event)
+		}
+	}
+	if len(filtered) == 0 {
+		filtered = append(filtered, events...)
+	}
+	if len(filtered) > maxPromptContextEvents {
+		filtered = filtered[len(filtered)-maxPromptContextEvents:]
+	}
+	return filtered
+}
+
 func forkTitle(title string) string {
 	if title == "" {
 		return ""

internal/sessions/replay.go

@@ -71,7 +71,7 @@ type CompactionPayload struct {
 	Truncated                bool       `json:"truncated,omitempty"`
 }
 
-const defaultCompactionPreserveLast = 8
+const defaultCompactionPreserveLast = 6
 const defaultCompactionMaxPromptChars = 8000
 
 func (store *Store) PlanRewind(sessionID string, options RewindOptions) (RewindPlan, error) {

internal/sessions/store_test.go

@@ -555,6 +555,91 @@ func TestPrepareExecSessionResolvesResumeAndFork(t *testing.T) {
 	}
 }
 
+func TestFormatExecPromptKeepsConversationMessagesWhenNoisyEventsFollow(t *testing.T) {
+	events := []Event{
+		{Sequence: 1, Type: EventMessage, Payload: json.RawMessage(`{"role":"user","content":"first user request"}`)},
+		{Sequence: 2, Type: EventMessage, Payload: json.RawMessage(`{"role":"assistant","content":"first assistant answer"}`)},
+	}
+	for sequence := 3; sequence <= 45; sequence++ {
+		events = append(events, Event{Sequence: sequence, Type: EventToolResult, Payload: json.RawMessage(`{"name":"read_file","output":"noisy tool result"}`)})
+	}
+	events = append(events, Event{Sequence: 46, Type: EventMessage, Payload: json.RawMessage(`{"role":"user","content":"latest user request"}`)})
+
+	prompt := FormatExecPrompt("continue", PreparedExec{
+		Mode:          ModeResume,
+		Session:       Metadata{SessionID: "session-with-noise"},
+		ContextEvents: events,
+	})
+
+	for _, want := range []string{"first user request", "first assistant answer", "latest user request", "Current user request:", "continue"} {
+		if !strings.Contains(prompt, want) {
+			t.Fatalf("expected prompt to contain %q, got %q", want, prompt)
+		}
+	}
+	if strings.Contains(prompt, "noisy tool result") {
+		t.Fatalf("expected prompt to omit noisy non-conversation events, got %q", prompt)
+	}
+}
+
+func TestFormatExecPromptTruncatesConversationMessagesAfterFilteringNoise(t *testing.T) {
+	events := make([]Event, 0, 100)
+	for sequence := 1; sequence <= 85; sequence++ {
+		events = append(events, Event{
+			Sequence: sequence,
+			Type:     EventMessage,
+			Payload:  json.RawMessage(fmt.Sprintf(`{"role":"user","content":"conversation message %03d"}`, sequence)),
+		})
+		if sequence%10 == 0 {
+			events = append(events, Event{
+				Sequence: 1000 + sequence,
+				Type:     EventToolResult,
+				Payload:  json.RawMessage(fmt.Sprintf(`{"name":"read_file","output":"noisy tool result %03d"}`, sequence)),
+			})
+		}
+	}
+	events = append(events, Event{
+		Sequence: 2000,
+		Type:     EventToolResult,
+		Payload:  json.RawMessage(`{"name":"bash","output":"trailing noisy tool result"}`),
+	})
+
+	contextEvents := promptContextEvents(events)
+	if len(contextEvents) != 80 {
+		t.Fatalf("expected prompt context to keep 80 events, got %d", len(contextEvents))
+	}
+	if contextEvents[0].Sequence != 6 || contextEvents[len(contextEvents)-1].Sequence != 85 {
+		t.Fatalf("expected prompt context to keep message sequences 6-85, got first=%d last=%d", contextEvents[0].Sequence, contextEvents[len(contextEvents)-1].Sequence)
+	}
+
+	prompt := FormatExecPrompt("continue", PreparedExec{
+		Mode:          ModeResume,
+		Session:       Metadata{SessionID: "long-session-with-noise"},
+		ContextEvents: events,
+	})
+
+	if got, want := strings.Count(prompt, "- #"), len(contextEvents); got != want {
+		t.Fatalf("expected prompt to contain %d context event lines, got %d in %q", want, got, prompt)
+	}
+	for _, want := range []string{"conversation message 006", "conversation message 085", "Current user request:", "continue"} {
+		if !strings.Contains(prompt, want) {
+			t.Fatalf("expected prompt to contain %q, got %q", want, prompt)
+		}
+	}
+	for _, unwanted := range []string{
+		"conversation message 001",
+		"conversation message 002",
+		"conversation message 003",
+		"conversation message 004",
+		"conversation message 005",
+		"noisy tool result",
+		"trailing noisy tool result",
+	} {
+		if strings.Contains(prompt, unwanted) {
+			t.Fatalf("expected prompt to omit %q, got %q", unwanted, prompt)
+		}
+	}
+}
+
 func TestPrepareExecPersistsSpecialistMetadataForNewSession(t *testing.T) {
 	store := NewStore(StoreOptions{RootDir: t.TempDir(), Now: fixedClock("2026-06-04T14:30:00Z")})
 

internal/specialist/exec.go

@@ -25,6 +25,7 @@ import (
 const (
 	sessionTagSpecialist     = "specialist"
 	promptFileThresholdBytes = 4 * 1024
+	maxSpecialistDepth       = 8 // hard cap to prevent infinite recursion/resource exhaustion via Task
 )
 
 const SessionTagSpecialist = sessionTagSpecialist
@@ -218,6 +219,13 @@ func (executor Executor) Run(ctx context.Context, params TaskParameters, options
 	if options.CurrentDepth < 0 {
 		return ExecResult{}, fmt.Errorf("current depth cannot be negative")
 	}
+	// >= (not >): this Run call spawns a CHILD at options.CurrentDepth+1, so a
+	// parent already AT the cap must still be rejected here rather than being
+	// allowed to launch one more level before the child's own next call trips
+	// the guard.
+	if options.CurrentDepth >= maxSpecialistDepth {
+		return ExecResult{}, fmt.Errorf("spawning a specialist at depth %d would exceed maximum nesting depth %d", options.CurrentDepth+1, maxSpecialistDepth)
+	}
 	if strings.TrimSpace(params.Prompt) == "" {
 		return ExecResult{}, fmt.Errorf("specialist prompt is required")
 	}

internal/specialist/exec_test.go

@@ -1,6 +1,7 @@
 package specialist
 
 import (
+	"context"
 	"os"
 	"path/filepath"
 	"reflect"
@@ -336,6 +337,39 @@ func TestBuildArgsRejectsInvalidInputs(t *testing.T) {
 	}
 }
 
+func TestRunRejectsDepthExceedingMax(t *testing.T) {
+	_, err := (Executor{}).Run(context.Background(), TaskParameters{
+		Prompt: "hi",
+	}, TaskRunOptions{CurrentDepth: maxSpecialistDepth + 1})
+	if err == nil || !strings.Contains(err.Error(), "depth") {
+		t.Fatalf("Run error = %v, want depth error", err)
+	}
+}
+
+// TestRunRejectsDepthAtMax covers the boundary: a parent already AT the cap
+// must be rejected too, since this Run call would launch a child one level
+// past it (--depth CurrentDepth+1). Only checking ">" here would let that
+// child start before the guard ever caught it. The guard sits before the
+// fresh/resume branch in Run, so both call shapes must be proven to reject
+// here rather than falling through to runFresh/runResume.
+func TestRunRejectsDepthAtMax(t *testing.T) {
+	tests := []struct {
+		name   string
+		params TaskParameters
+	}{
+		{name: "fresh", params: TaskParameters{Prompt: "hi"}},
+		{name: "resume", params: TaskParameters{Prompt: "hi", Resume: "child_session"}},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			_, err := (Executor{}).Run(context.Background(), tc.params, TaskRunOptions{CurrentDepth: maxSpecialistDepth})
+			if err == nil || !strings.Contains(err.Error(), "depth") {
+				t.Fatalf("Run error = %v, want depth error", err)
+			}
+		})
+	}
+}
+
 func TestBuildArgsRejectsInvalidSessionIDs(t *testing.T) {
 	_, err := (Executor{NewSessionID: func() (string, error) { return "../escape", nil }}).BuildArgs(BuildArgsInput{
 		Prompt: "hi",

internal/tools/ask_user.go

@@ -272,15 +272,36 @@ func coerceAskUserOptions(optionsValue, descsValue any) (labels []string, descri
 // FormatAskUserAnswers renders question/answer pairs into a clear, model-readable
 // block. Missing answers are surfaced explicitly so the model never silently
 // treats an unanswered question as answered.
+//
+// It distinguishes two shapes of "empty" the model would otherwise conflate: a
+// wholesale dismissal (the user closed the prompt without answering ANYTHING) is
+// flagged up front as a skip, so the model doesn't invent a default; a single
+// blank field amid other answers is marked "(left blank)".
 func FormatAskUserAnswers(questions []AskUserQuestion, answers []string) string {
-	lines := make([]string, 0, len(questions)*3)
+	anyAnswered := false
+	for index := range questions {
+		if index < len(answers) && strings.TrimSpace(answers[index]) != "" {
+			anyAnswered = true
+			break
+		}
+	}
+
+	lines := make([]string, 0, len(questions)*3+2)
+	if len(questions) > 0 && !anyAnswered {
+		lines = append(lines, "[note] The user dismissed this prompt without answering. Treat this as a skip: do not assume a default — ask again more specifically, or proceed only if the intent is already unambiguous.")
+		lines = append(lines, "")
+	}
 	for index, question := range questions {
 		answer := ""
 		if index < len(answers) {
 			answer = strings.TrimSpace(answers[index])
 		}
 		if answer == "" {
-			answer = "(no answer provided)"
+			if anyAnswered {
+				answer = "(left blank)"
+			} else {
+				answer = "(skipped)"
+			}
 		}
 		lines = append(lines, fmt.Sprintf("%d. [question] %s", index+1, question.Question))
 		lines = append(lines, "[answer] "+answer)

internal/tools/ask_user_format_test.go

@@ -0,0 +1,34 @@
+package tools
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestFormatAskUserAnswersDistinguishesSkipFromBlank(t *testing.T) {
+	questions := []AskUserQuestion{
+		{Question: "Which database?"},
+		{Question: "Migrate existing data?"},
+	}
+
+	// Wholesale dismissal (nothing answered): flagged as a skip up front.
+	dismissed := FormatAskUserAnswers(questions, []string{"", ""})
+	if !strings.Contains(dismissed, "dismissed this prompt") || !strings.Contains(dismissed, "Treat this as a skip") {
+		t.Fatalf("all-empty answers must be flagged as a skip, got:\n%s", dismissed)
+	}
+	if strings.Contains(dismissed, "left blank") {
+		t.Fatalf("a full dismissal should read as skipped, not left-blank:\n%s", dismissed)
+	}
+	if !strings.Contains(dismissed, "(skipped)") {
+		t.Fatalf("dismissed questions should render (skipped):\n%s", dismissed)
+	}
+
+	// Partial answer: the blank one is "left blank", no wholesale-skip note.
+	partial := FormatAskUserAnswers(questions, []string{"postgres", ""})
+	if strings.Contains(partial, "dismissed this prompt") {
+		t.Fatalf("a partial answer is not a dismissal:\n%s", partial)
+	}
+	if !strings.Contains(partial, "postgres") || !strings.Contains(partial, "(left blank)") {
+		t.Fatalf("partial answers should show the answer and mark the blank one left-blank:\n%s", partial)
+	}
+}

internal/tools/bash.go

@@ -1,7 +1,6 @@
 package tools
 
 import (
-	"bytes"
 	"context"
 	"errors"
 	"fmt"
@@ -120,14 +119,17 @@ func (tool bashTool) run(ctx context.Context, args map[string]any, engine *zeroS
 			Meta:   meta,
 		}
 	}
-	// Release any plan-scoped resources once the command has finished running.
 	defer plan.Cleanup()
 	addSandboxMeta(meta, plan)
 
-	var stdout bytes.Buffer
-	var stderr bytes.Buffer
-	command.Stdout = &stdout
-	command.Stderr = &stderr
+	// Bound the capture so a command with runaway output (`cat huge.log`, `yes`)
+	// can't grow Zero's memory before truncation: only the head+tail each stream
+	// will ever surface to the model are retained, the middle is discarded as it
+	// streams, and the true size is counted for the truncation marker.
+	stdout := newBoundedBuffer(bashOutputBudgetBytes, bashOutputBudgetBytes)
+	stderr := newBoundedBuffer(bashOutputBudgetBytes, bashOutputBudgetBytes)
+	command.Stdout = stdout
+	command.Stderr = stderr
 
 	// Kill the shell as a process group on timeout and bound the post-kill I/O
 	// wait, so a backgrounded child cannot outlive the command or hang Run().
@@ -139,7 +141,12 @@ func (tool bashTool) run(ctx context.Context, args map[string]any, engine *zeroS
 	err = command.Run()
 	exitCode := commandExitCode(err)
 	meta["exit_code"] = strconv.Itoa(exitCode)
-	stderrText := appendSandboxBlocks(stderr.String(), monitor.Stop())
+	stdoutText := stdout.retained()
+	stderrRetained := stderr.retained()
+	stderrText := appendSandboxBlocks(stderrRetained, monitor.Stop())
+	// Sandbox blocks are extra model-visible stderr bytes appended after capture;
+	// count them toward the true total so the budget/marker stay accurate.
+	stderrTotal := stderr.total + (len(stderrText) - len(stderrRetained))
 
 	if errors.Is(commandCtx.Err(), context.DeadlineExceeded) {
 		return Result{
@@ -156,26 +163,31 @@ func (tool bashTool) run(ctx context.Context, args map[string]any, engine *zeroS
 				Meta:   meta,
 			}
 		}
-		markLikelySandboxDenial(meta, plan, exitCode, stdout.String(), stderrText)
+		markLikelySandboxDenial(meta, plan, exitCode, stdoutText, stderrText)
+		outText, errText, truncated := budgetBashCapture(stdoutText, stdout.total, stderrText, stderrTotal, meta)
 		return Result{
-			Status: StatusError,
-			Output: formatBashOutputWithShellHint(commandText, stdout.String(), stderrText, exitCode, meta),
-			Meta:   meta,
+			Status:    StatusError,
+			Output:    formatBashOutputWithShellHint(commandText, outText, errText, exitCode, meta),
+			Truncated: truncated,
+			Meta:      meta,
 		}
 	}
 
-	markLikelySandboxDenial(meta, plan, exitCode, stdout.String(), stderrText)
+	markLikelySandboxDenial(meta, plan, exitCode, stdoutText, stderrText)
+	outText, errText, truncated := budgetBashCapture(stdoutText, stdout.total, stderrText, stderrTotal, meta)
 	if meta[SandboxLikelyDeniedMeta] == "true" {
 		return Result{
-			Status: StatusError,
-			Output: formatBashOutputWithShellHint(commandText, stdout.String(), stderrText, exitCode, meta),
-			Meta:   meta,
+			Status:    StatusError,
+			Output:    formatBashOutputWithShellHint(commandText, outText, errText, exitCode, meta),
+			Truncated: truncated,
+			Meta:      meta,
 		}
 	}
 	return Result{
-		Status: StatusOK,
-		Output: formatBashOutput(stdout.String(), stderrText, exitCode),
-		Meta:   meta,
+		Status:    StatusOK,
+		Output:    formatBashOutput(outText, errText, exitCode),
+		Truncated: truncated,
+		Meta:      meta,
 	}
 }
 
@@ -221,6 +233,14 @@ func shellIssueBlockResult(issue shellIssue) Result {
 	}
 }
 
+// buildBashCommand returns the exec.Cmd and the sandbox plan for running
+// commandText. On Windows, when the command is not wrapped by the sandbox
+// engine (plan.Wrapped == false), it also overrides the child's raw command
+// line so commandText reaches cmd.exe unescaped; see
+// zeroSandbox.WindowsShellCommandLine for why that matters. The wrapped case
+// gets the same treatment inside the sandboxed runner process itself
+// (internal/sandbox/windows_process_windows.go), since that command line is
+// built there, not here.
 func buildBashCommand(ctx context.Context, commandText string, absoluteCwd string, engine *zeroSandbox.Engine) (*exec.Cmd, zeroSandbox.CommandPlan, error) {
 	spec := zeroSandbox.CommandSpec{
 		Name: shellExecutable(),
@@ -228,7 +248,11 @@ func buildBashCommand(ctx context.Context, commandText string, absoluteCwd strin
 		Dir:  absoluteCwd,
 	}
 	if engine != nil {
-		return engine.CommandContext(ctx, spec)
+		command, plan, err := engine.CommandContext(ctx, spec)
+		if err == nil {
+			applyWindowsShellCommandLine(command, commandText, plan.Wrapped)
+		}
+		return command, plan, err
 	}
 	plan := zeroSandbox.CommandPlan{
 		Backend: zeroSandbox.Backend{
@@ -242,6 +266,7 @@ func buildBashCommand(ctx context.Context, commandText string, absoluteCwd strin
 	}
 	command := exec.CommandContext(ctx, spec.Name, spec.Args...)
 	command.Dir = spec.Dir
+	applyWindowsShellCommandLine(command, commandText, plan.Wrapped)
 	return command, plan, nil
 }
 
@@ -302,7 +327,7 @@ func shellExecutable() string {
 
 func shellArguments(command string) []string {
 	if runtime.GOOS == "windows" {
-		return []string{"/d", "/s", "/c", command}
+		return zeroSandbox.WindowsShellArgs(command)
 	}
 	return []string{"-c", command}
 }
@@ -345,6 +370,124 @@ func formatBashOutput(stdout string, stderr string, exitCode int) string {
 	return strings.Join(parts, "\n")
 }
 
+// bashOutputBudgetBytes caps each of stdout/stderr shown to the model. bash is the
+// one tool that can emit unbounded output (`cat large.log`, `find /`, verbose test
+// runs); every other read/search tool already budgets its output. Head+tail
+// truncation keeps both the start and the end of an oversized stream, since
+// build/test failures usually surface at the tail.
+const bashOutputBudgetBytes = 96 * 1024
+
+// budgetBashOutput truncates stdout and stderr to bashOutputBudgetBytes each,
+// keeping the head and tail of anything larger, and records raw/emitted byte
+// counts plus a truncated flag in meta (mirroring outputBudgetMeta's shape for
+// the read/search tools). Detection that needs the full output (sandbox-denial
+// scanning) must run on the raw strings before this is applied.
+func budgetBashOutput(stdout string, stderr string, meta map[string]string) (string, string, bool) {
+	return budgetBashCapture(stdout, len(stdout), stderr, len(stderr), meta)
+}
+
+// budgetBashCapture is budgetBashOutput for the streaming-capture path: outTotal
+// and errTotal are the true byte counts (from boundedBuffer.total), which may
+// exceed the retained strings when the middle was dropped during capture. Meta's
+// raw_bytes therefore reflects everything the command produced, not just what was
+// kept in memory.
+func budgetBashCapture(out string, outTotal int, errStr string, errTotal int, meta map[string]string) (string, string, bool) {
+	outText, outRaw, outTrunc := truncateHeadTailWithTotal(out, outTotal, bashOutputBudgetBytes)
+	errText, errRaw, errTrunc := truncateHeadTailWithTotal(errStr, errTotal, bashOutputBudgetBytes)
+	truncated := outTrunc || errTrunc
+	if meta != nil {
+		emitted := len(outText) + len(errText)
+		meta["raw_bytes"] = strconv.Itoa(outRaw + errRaw)
+		meta["emitted_bytes"] = strconv.Itoa(emitted)
+		meta["estimated_tokens"] = strconv.Itoa(estimatedTokensFromBytes(emitted))
+		if truncated {
+			meta["truncated"] = "true"
+		}
+	}
+	return outText, errText, truncated
+}
+
+// boundedBuffer is an io.Writer that retains at most headCap bytes from the start
+// and tailCap bytes from the end of a stream while counting the total written, so
+// a command emitting unbounded output (`cat huge.log`, `yes`) cannot grow Zero's
+// memory: the middle is discarded as it arrives instead of buffered whole and then
+// truncated. total records the full size for the truncation marker even though the
+// middle is never held. Not safe for concurrent writes; exec drives Stdout and
+// Stderr from separate goroutines, so each stream gets its own buffer.
+type boundedBuffer struct {
+	head    []byte
+	headCap int
+	tail    []byte
+	tailCap int
+	total   int
+}
+
+func newBoundedBuffer(headCap, tailCap int) *boundedBuffer {
+	return &boundedBuffer{headCap: headCap, tailCap: tailCap}
+}
+
+func (b *boundedBuffer) Write(p []byte) (int, error) {
+	n := len(p)
+	b.total += n
+	// Fill the head until it reaches headCap; the head is written once and frozen.
+	if len(b.head) < b.headCap {
+		take := b.headCap - len(b.head)
+		if take > len(p) {
+			take = len(p)
+		}
+		b.head = append(b.head, p[:take]...)
+		p = p[take:]
+	}
+	// Anything past the head feeds a rolling tail that keeps only the last tailCap
+	// bytes. Compact once it grows past 2×tailCap so the append stays amortized O(1)
+	// while memory never exceeds ~2×tailCap.
+	if len(p) > 0 && b.tailCap > 0 {
+		b.tail = append(b.tail, p...)
+		if len(b.tail) > 2*b.tailCap {
+			b.tail = append(b.tail[:0], b.tail[len(b.tail)-b.tailCap:]...)
+		}
+	}
+	return n, nil
+}
+
+// retained returns the kept head+tail bytes (marker-less) as a string. The junction
+// between head and tail lands in the middle, which the display budget trims away;
+// callers that need the true size read total separately.
+func (b *boundedBuffer) retained() string {
+	if len(b.tail) > b.tailCap {
+		// Not yet compacted since the last overflow; expose only the last tailCap.
+		return string(b.head) + string(b.tail[len(b.tail)-b.tailCap:])
+	}
+	return string(b.head) + string(b.tail)
+}
+
+// truncateHeadTail keeps the first and last halves of value when it exceeds
+// maxBytes, dropping the middle behind a marker. Returns the possibly-truncated
+// text, the raw byte length, and whether truncation happened.
+func truncateHeadTail(value string, maxBytes int) (string, int, bool) {
+	// value holds the whole stream, so its length is the true raw size.
+	return truncateHeadTailWithTotal(value, len(value), maxBytes)
+}
+
+// truncateHeadTailWithTotal head+tail-truncates value to maxBytes, using total —
+// the full original byte count — for the "N bytes omitted" marker and the raw
+// count. total may exceed len(value) when the middle was already discarded during
+// bounded capture (boundedBuffer): value then holds only the retained head+tail,
+// and this trims it to the display budget while still reporting the true total.
+func truncateHeadTailWithTotal(value string, total, maxBytes int) (string, int, bool) {
+	if maxBytes <= 0 || total <= maxBytes {
+		return value, total, false
+	}
+	marker := fmt.Sprintf("\n[zero] output truncated: %d bytes omitted from the middle — redirect to a file and read_file a range for the full text\n", total-maxBytes)
+	budget := maxBytes - len(marker)
+	if budget < 0 {
+		budget = 0
+	}
+	head := budget / 2
+	tail := budget - head
+	return utf8Prefix(value, head) + marker + utf8Suffix(value, tail), total, true
+}
+
 func formatBashOutputWithShellHint(command string, stdout string, stderr string, exitCode int, meta map[string]string) string {
 	output := formatBashOutput(stdout, stderr, exitCode)
 	if issue := detectShellOutputIssue(command, stdout+"\n"+stderr, runtime.GOOS); issue != nil {

internal/tools/bash_budget_test.go

@@ -0,0 +1,127 @@
+package tools
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+	"testing"
+)
+
+// Small output passes through untouched and records raw==emitted, no truncated flag.
+func TestBudgetBashOutputSmallPassesThrough(t *testing.T) {
+	meta := map[string]string{}
+	out, errStr, truncated := budgetBashOutput("hello\n", "warn\n", meta)
+	if out != "hello\n" || errStr != "warn\n" {
+		t.Fatalf("small output altered: out=%q err=%q", out, errStr)
+	}
+	if truncated {
+		t.Fatalf("small output must report truncated=false")
+	}
+	if meta["truncated"] == "true" {
+		t.Fatalf("small output must not be flagged truncated: %v", meta)
+	}
+	if meta["raw_bytes"] != strconv.Itoa(len("hello\n")+len("warn\n")) {
+		t.Fatalf("raw_bytes wrong: %v", meta)
+	}
+}
+
+// Oversized stdout is truncated head+tail: both the first and last lines survive,
+// the middle is dropped behind a marker, and meta is flagged.
+func TestBudgetBashOutputTruncatesHeadAndTail(t *testing.T) {
+	head := "FIRST_LINE_MARKER\n"
+	tail := "\nLAST_LINE_MARKER"
+	big := head + strings.Repeat("x", bashOutputBudgetBytes) + tail
+
+	meta := map[string]string{}
+	out, _, truncated := budgetBashOutput(big, "", meta)
+
+	if !truncated {
+		t.Fatalf("oversized output must report truncated=true")
+	}
+	if !strings.Contains(out, "FIRST_LINE_MARKER") {
+		t.Fatalf("head lost after truncation")
+	}
+	if !strings.Contains(out, "LAST_LINE_MARKER") {
+		t.Fatalf("tail lost after truncation (failures live at the tail)")
+	}
+	if !strings.Contains(out, "output truncated") {
+		t.Fatalf("expected a truncation marker, got:\n%s", out[:200])
+	}
+	if len(out) > bashOutputBudgetBytes {
+		t.Fatalf("emitted %d bytes exceeds budget %d", len(out), bashOutputBudgetBytes)
+	}
+	if meta["truncated"] != "true" {
+		t.Fatalf("expected truncated=true, got %v", meta)
+	}
+	if meta["raw_bytes"] != strconv.Itoa(len(big)) {
+		t.Fatalf("raw_bytes = %s, want %d", meta["raw_bytes"], len(big))
+	}
+	if got, _ := strconv.Atoi(meta["emitted_bytes"]); got != len(out) {
+		t.Fatalf("emitted_bytes = %s, want %d", meta["emitted_bytes"], len(out))
+	}
+}
+
+// boundedBuffer must keep memory bounded to head+tail while counting every byte,
+// so a runaway command can't OOM Zero before truncation runs. It keeps the very
+// first bytes (head) and the very last bytes (tail), discarding the middle.
+func TestBoundedBufferKeepsHeadAndTailBounded(t *testing.T) {
+	b := newBoundedBuffer(8, 8)
+	total := 0
+	for i := 0; i < 1000; i++ {
+		chunk := []byte(fmt.Sprintf("%05d.", i)) // 6 bytes each
+		n, err := b.Write(chunk)
+		if err != nil || n != len(chunk) {
+			t.Fatalf("Write = (%d, %v), want (%d, nil)", n, err, len(chunk))
+		}
+		total += len(chunk)
+	}
+	if b.total != total {
+		t.Fatalf("total = %d, want %d (every byte must be counted)", b.total, total)
+	}
+	// Retained memory is bounded regardless of how much streamed through.
+	if len(b.head) > 8 {
+		t.Fatalf("head grew past its cap: %d", len(b.head))
+	}
+	if len(b.tail) > 2*8 {
+		t.Fatalf("tail grew past 2×cap (compaction failed): %d", len(b.tail))
+	}
+	r := b.retained()
+	if len(r) > 8+8 {
+		t.Fatalf("retained %d bytes, want <= headCap+tailCap", len(r))
+	}
+	if !strings.HasPrefix(r, "00000.") {
+		t.Fatalf("retained head lost, got prefix %q", r[:min(6, len(r))])
+	}
+	if !strings.HasSuffix(r, "00999.") {
+		t.Fatalf("retained tail lost, got %q", r)
+	}
+}
+
+// budgetBashCapture reports the TRUE total (not the retained size) in the marker
+// and raw_bytes, even though only a bounded head+tail was ever held in memory.
+func TestBudgetBashCaptureReportsTrueTotal(t *testing.T) {
+	// Retained head+tail as boundedBuffer would hand over; the real command produced
+	// far more than was kept.
+	retained := "HEAD_START" + strings.Repeat("y", bashOutputBudgetBytes) + "TAIL_END"
+	total := 10 * bashOutputBudgetBytes
+
+	meta := map[string]string{}
+	out, _, truncated := budgetBashCapture(retained, total, "", 0, meta)
+
+	if !truncated {
+		t.Fatal("oversized capture must report truncated=true")
+	}
+	if !strings.Contains(out, "HEAD_START") || !strings.Contains(out, "TAIL_END") {
+		t.Fatalf("head/tail lost after budgeting:\n%s", out[:min(120, len(out))])
+	}
+	if len(out) > bashOutputBudgetBytes {
+		t.Fatalf("emitted %d bytes exceeds budget %d", len(out), bashOutputBudgetBytes)
+	}
+	if meta["raw_bytes"] != strconv.Itoa(total) {
+		t.Fatalf("raw_bytes = %s, want the true total %d", meta["raw_bytes"], total)
+	}
+	// The marker must cite the true omitted count (total-budget), not retained-budget.
+	if !strings.Contains(out, strconv.Itoa(total-bashOutputBudgetBytes)) {
+		t.Fatalf("marker should cite the true omitted byte count:\n%s", out[:min(200, len(out))])
+	}
+}

internal/tools/bash_proc_unix.go

@@ -42,3 +42,7 @@ func hardenProcessLifetime(command *exec.Cmd) {
 		return nil
 	}
 }
+
+// applyWindowsShellCommandLine is a no-op outside Windows; there is no
+// cmd.exe command-line-parsing quirk to work around here.
+func applyWindowsShellCommandLine(command *exec.Cmd, commandText string, wrapped bool) {}

internal/tools/bash_proc_windows.go

@@ -5,7 +5,10 @@ package tools
 import (
 	"os/exec"
 	"strconv"
+	"syscall"
 	"time"
+
+	zeroSandbox "github.com/Gitlawb/zero/internal/sandbox"
 )
 
 // bashWaitDelay bounds how long Wait blocks for the I/O pipes to drain after the
@@ -28,3 +31,17 @@ func hardenProcessLifetime(command *exec.Cmd) {
 		return nil
 	}
 }
+
+// applyWindowsShellCommandLine overrides command's raw child command line so
+// commandText reaches cmd.exe unescaped instead of auto-quoted the way
+// exec.Cmd would normally encode a single Args element. Skipped when wrapped
+// is true: the sandbox engine then routes execution through a separate
+// zero-windows-command-runner process, which builds its own child command
+// line from scratch (internal/sandbox/windows_process_windows.go) rather than
+// inheriting whatever this outer exec.Cmd is configured with.
+func applyWindowsShellCommandLine(command *exec.Cmd, commandText string, wrapped bool) {
+	if wrapped {
+		return
+	}
+	command.SysProcAttr = &syscall.SysProcAttr{CmdLine: zeroSandbox.WindowsShellCommandLine(commandText)}
+}

internal/tools/bash_tool_test.go

@@ -8,6 +8,7 @@ import (
 	"os"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"strings"
 	"testing"
 	"time"
@@ -17,6 +18,18 @@ import (
 
 func TestMain(m *testing.M) {
 	if len(os.Args) >= 3 && os.Args[1] == "--zero-bash-helper" {
+		if os.Args[2] == "echo-arg" {
+			// Echoes back exactly the one argument it received, to verify a
+			// quoted, space-and-slash-containing argument (the same shape as
+			// `python -c "print(15 / 3)"`) survives the shell invocation
+			// intact instead of being truncated/corrupted.
+			if len(os.Args) < 4 {
+				fmt.Fprintln(os.Stderr, "echo-arg requires exactly one argument")
+				os.Exit(1)
+			}
+			fmt.Println(os.Args[3])
+			return
+		}
 		runBashToolHelper(os.Args[2])
 		return
 	}
@@ -161,6 +174,35 @@ func TestDetectShellCommandIssueRequiresActualLSCommand(t *testing.T) {
 	}
 }
 
+func TestDetectShellCommandIssueFlagsPipedPosixUtilities(t *testing.T) {
+	for _, command := range []string{
+		`git log --oneline -- pinokio_compat.py | head`,
+		`git blame -L 40,46 pinokio_compat.py --format=%author | grep summary`,
+		`cat file.txt | wc -l`,
+		`some_command | tail -n 20`,
+	} {
+		if issue := detectShellCommandIssue(command, "windows"); issue == nil {
+			t.Fatalf("expected POSIX utility pipeline to be flagged for %q", command)
+		}
+	}
+}
+
+func TestDetectShellCommandIssueAllowsUnrelatedCommands(t *testing.T) {
+	for _, command := range []string{
+		`git log --oneline`,
+		`dir /b`,
+		`findstr /r "summary" file.txt`,
+		`echo header text`,
+		`tail-log.sh`,
+		`grep-cli --version`,
+		`sed.exe --help`,
+	} {
+		if issue := detectShellCommandIssue(command, "windows"); issue != nil {
+			t.Fatalf("expected unrelated command to pass for %q, got %#v", command, issue)
+		}
+	}
+}
+
 func TestDetectShellOutputIssueAddsWindowsSyntaxHint(t *testing.T) {
 	issue := detectShellOutputIssue(`cd /d/tmp/zero-pr-158 && ls -la`, "The syntax of the command is incorrect.", "windows")
 	if issue == nil {
@@ -211,6 +253,43 @@ func TestBashToolRunsCommandInWorkspace(t *testing.T) {
 	}
 }
 
+// A command with runaway output must not be buffered whole in memory: the capture
+// is bounded to head+tail, yet raw_bytes still reports the true (much larger) size.
+// If capture were unbounded this would balloon Zero's memory before truncation.
+func TestBashToolBoundsRunawayOutputCapture(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("uses a POSIX shell pipeline (yes | head)")
+	}
+	const produced = 500000 // ~5× the 96 KiB budget
+
+	result := NewBashTool(t.TempDir()).Run(context.Background(), map[string]any{
+		"command": fmt.Sprintf("yes ABCDEFGH | head -c %d", produced),
+	})
+
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok status, got %s: %s", result.Status, result.Output)
+	}
+	if !result.Truncated || result.Meta["truncated"] != "true" {
+		t.Fatalf("runaway output should be flagged truncated, meta=%v", result.Meta)
+	}
+	if !strings.Contains(result.Output, "output truncated") {
+		t.Fatalf("expected a truncation marker in output")
+	}
+	// The emitted (model-visible) text is capped near the budget...
+	if len(result.Output) > bashOutputBudgetBytes+1024 {
+		t.Fatalf("emitted %d bytes far exceeds budget %d", len(result.Output), bashOutputBudgetBytes)
+	}
+	// ...but raw_bytes reflects the full stream, proving the total was counted while
+	// only a bounded slice was ever held (raw >> the ~2×budget retention cap).
+	raw, err := strconv.Atoi(result.Meta["raw_bytes"])
+	if err != nil {
+		t.Fatalf("raw_bytes not an int: %q", result.Meta["raw_bytes"])
+	}
+	if raw < produced {
+		t.Fatalf("raw_bytes = %d, want >= %d (full stream counted)", raw, produced)
+	}
+}
+
 func TestBashToolUsesRequestedCwd(t *testing.T) {
 	root := t.TempDir()
 	nested := filepath.Join(root, "nested")
@@ -506,6 +585,68 @@ func TestBashToolAllowsNonInteractiveCommand(t *testing.T) {
 	}
 }
 
+// TestBashToolPreservesEmbeddedQuotesOnWindows pins a real, previously-broken
+// case: passing commandText as a normal exec.Cmd argument makes Go wrap it in
+// an outer pair of quotes (it contains spaces) with its own quotes escaped as
+// \", and cmd.exe's /C remainder parsing strips the first and last literal
+// quote character in that remainder without undoing the backslash-escaping -
+// corrupting a command whose own text contains embedded double quotes,
+// exactly the shape of `python -c "print(15 / 3)"`, `git commit -m
+// "message"`, `node -e "..."`. Before the fix, the helper below received a
+// truncated, mis-quoted argument (starting with a stray literal `"`) instead
+// of the text between the quotes.
+func TestBashToolPreservesEmbeddedQuotesOnWindows(t *testing.T) {
+	if runtime.GOOS != "windows" {
+		t.Skip("cmd.exe /S/C quote-remainder parsing is Windows-specific")
+	}
+	root := t.TempDir()
+	executable := shellQuote(os.Args[0])
+	// Space and a slash: the two characters whose position in the original
+	// bug report's SyntaxError ("print(15, truncated right before the /")
+	// showed exactly where the corruption happened.
+	commandText := executable + ` --zero-bash-helper echo-arg "hello / world"`
+
+	result := NewBashTool(root).Run(context.Background(), map[string]any{
+		"command": commandText,
+	})
+
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok status, got %s: %s", result.Status, result.Output)
+	}
+	if !strings.Contains(result.Output, "hello / world") {
+		t.Fatalf("expected the embedded-quote argument to survive intact, got %q", result.Output)
+	}
+}
+
+// TestBashToolRunsCommandLineForLoopSyntax pins cmd.exe command-line syntax,
+// not batch-file syntax: a `for %i in (...) do ...` loop with a single
+// percent sign is valid typed directly at a cmd.exe prompt (or via
+// `cmd /C "..."`), but requires %%i inside an actual .bat/.cmd FILE, because
+// batch files perform an extra pass of percent-substitution before FOR ever
+// runs, consuming the single percent so the loop variable never binds. An
+// earlier version of this fix ran commandText from a temporary .cmd file
+// instead of passing it straight through to cmd.exe's /C remainder, which
+// silently broke single-percent syntax like this.
+func TestBashToolRunsCommandLineForLoopSyntax(t *testing.T) {
+	if runtime.GOOS != "windows" {
+		t.Skip("cmd.exe command-line vs batch-file parsing is Windows-specific")
+	}
+	root := t.TempDir()
+
+	result := NewBashTool(root).Run(context.Background(), map[string]any{
+		"command": "for %i in (1 2 3) do echo %i",
+	})
+
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok status, got %s: %s", result.Status, result.Output)
+	}
+	for _, want := range []string{"1", "2", "3"} {
+		if !strings.Contains(result.Output, want) {
+			t.Fatalf("expected the for-loop to expand %%i to %q, got %q", want, result.Output)
+		}
+	}
+}
+
 func helperCommand(name string) string {
 	executable := shellQuote(os.Args[0])
 	return executable + " --zero-bash-helper " + name

internal/tools/edit_file.go

@@ -2,6 +2,7 @@ package tools
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"os"
 	"strings"
@@ -44,7 +45,7 @@ func (tool editFileTool) Run(ctx context.Context, args map[string]any) Result {
 	return tool.RunWithOptions(ctx, args, RunOptions{})
 }
 
-func (tool editFileTool) RunWithOptions(_ context.Context, args map[string]any, options RunOptions) Result {
+func (tool editFileTool) RunWithOptions(ctx context.Context, args map[string]any, options RunOptions) Result {
 	requestedPath, err := aliasedStringArg(args, []string{"path", "file", "file_path", "filename"}, "", true, false)
 	if err != nil {
 		return errorResult("Error: Invalid arguments for edit_file: " + err.Error())
@@ -98,6 +99,35 @@ func (tool editFileTool) RunWithOptions(_ context.Context, args map[string]any,
 		}
 	}
 
+	// Fuzzy fallback: when the exact string (and its CRLF translation) is absent,
+	// run a cascade of tolerant matchers (trimmed lines, block anchors, collapsed
+	// whitespace, indentation drift, escape normalization) to locate the span the
+	// model intended. Only a span that occurs literally in the file is accepted,
+	// so the replacement applied below is still exact.
+	if occurrences == 0 {
+		findOld, findNew := oldString, newString
+		if strings.Contains(content, "\r\n") && !strings.Contains(findOld, "\r\n") {
+			findOld = strings.ReplaceAll(findOld, "\n", "\r\n")
+			findNew = strings.ReplaceAll(findNew, "\n", "\r\n")
+		}
+		search, ferr := fuzzyEditMatch(content, findOld, replaceAll)
+		switch {
+		case ferr == nil:
+			oldString = search
+			// The model's new_string was written at old_string's (mismatched)
+			// indentation; re-shape it to the span actually being replaced so a
+			// tolerant match never strips indentation or a trailing CR.
+			newString = adaptReplacementToSpan(search, findOld, findNew)
+			occurrences = strings.Count(content, search)
+		case errors.Is(ferr, errEditFuzzyAmbiguous):
+			return errorResult("Error: old_string matches multiple locations in " + relativePath + " even after fuzzy matching. Provide more surrounding context to make the match unique, or pass replace_all: true.")
+		case errors.Is(ferr, errEditFuzzyNotFound):
+			// Fall through to the exact-match error below.
+		default:
+			return errorResult("Error editing " + relativePath + ": " + ferr.Error())
+		}
+	}
+
 	if occurrences == 0 {
 		return errorResult("Error: Could not find the exact string to replace in " + relativePath + ". The old_string must match the file byte-for-byte.")
 	}
@@ -120,6 +150,10 @@ func (tool editFileTool) RunWithOptions(_ context.Context, args map[string]any,
 	if err := os.WriteFile(absolutePath, []byte(updated), 0o644); err != nil {
 		return errorResult("Error writing " + relativePath + ": " + err.Error())
 	}
+	// Optional format-on-write (ZERO_FORMAT_ON_WRITE). Must run BEFORE the
+	// FileTracker re-baseline: recording pre-format content would make the very
+	// next edit look like an external modification and trip the conflict guard.
+	updated = maybeFormatWrittenFile(ctx, absolutePath, updated)
 	// Re-baseline to the content we just wrote so subsequent edits in this session
 	// compare against the current on-disk state, not the pre-edit version.
 	newInfo, _ := os.Stat(absolutePath)
@@ -130,6 +164,7 @@ func (tool editFileTool) RunWithOptions(_ context.Context, args map[string]any,
 		suffix = "s"
 	}
 	summary := fmt.Sprintf("Successfully edited %s (replaced %d occurrence%s).", relativePath, replacedCount, suffix)
+	summary += inlineDiagnostics(ctx, options, absolutePath, relativePath)
 	result := okResult(summary)
 	result.ChangedFiles = []string{relativePath}
 	// Card-only preview (Display.Preview): the model's Output stays the one-line

internal/tools/edit_replacers.go

@@ -0,0 +1,554 @@
+package tools
+
+// Fuzzy fallback matching for edit_file. When the model's old_string fails to
+// match byte-for-byte (drifted indentation, collapsed whitespace, escaped
+// characters, a slightly-misremembered middle line), these replacers propose
+// candidate spans that plausibly correspond to what the model intended. Only a
+// candidate that occurs literally in the file is accepted, so the replacement
+// itself is always exact even when the match was tolerant.
+//
+// Strategy cascade ported from opencode's edit tool, whose replacers were in
+// turn distilled from Cline's diff-apply evals and gemini-cli's edit corrector.
+
+import (
+	"errors"
+	"regexp"
+	"sort"
+	"strings"
+)
+
+// editReplacer proposes candidate spans for find inside content. Candidates
+// are re-validated by fuzzyEditMatch before use.
+type editReplacer func(content, find string) []string
+
+var (
+	errEditFuzzyNotFound  = errors.New("no fuzzy match for old_string")
+	errEditFuzzyAmbiguous = errors.New("fuzzy match for old_string is ambiguous")
+)
+
+// Minimum average middle-line similarity for block-anchor matches.
+const editAnchorSimilarityThreshold = 0.65
+
+// fuzzyEditMatch runs the replacer cascade and returns the exact span of
+// content to replace. When replaceAll is false the span must be unique in
+// content; an ambiguous candidate is skipped in favor of later candidates and
+// only reported if nothing unique is found. A span wildly larger than
+// old_string is refused outright rather than risking a destructive edit.
+func fuzzyEditMatch(content, find string, replaceAll bool) (string, error) {
+	replacers := []editReplacer{
+		lineTrimmedReplacer,
+		blockAnchorReplacer,
+		whitespaceNormalizedReplacer,
+		indentationFlexibleReplacer,
+		escapeNormalizedReplacer,
+		trimmedBoundaryReplacer,
+		contextAwareReplacer,
+	}
+	found := false
+	for _, replacer := range replacers {
+		// Collect the replacer's DISTINCT candidate spans that literally occur in
+		// content. Two or more distinct spans from one strategy (e.g. duplicate
+		// blocks at different indentation, each occurring once) mean the model's
+		// intent is genuinely ambiguous — silently editing the first would be a
+		// wrong-span write, so it is rejected instead.
+		var candidates []string
+		seen := map[string]bool{}
+		for _, search := range replacer(content, find) {
+			if search == "" || seen[search] {
+				continue
+			}
+			seen[search] = true
+			if strings.Index(content, search) < 0 {
+				continue
+			}
+			candidates = append(candidates, search)
+		}
+		if len(candidates) == 0 {
+			continue
+		}
+		found = true
+		if !replaceAll && len(candidates) > 1 {
+			return "", errEditFuzzyAmbiguous
+		}
+		search := candidates[0]
+		if isDisproportionateEditMatch(search, find) {
+			return "", errors.New("refusing replacement because the matched span is much larger than old_string; re-read the file and provide the full exact old_string for the intended replacement")
+		}
+		if replaceAll {
+			return search, nil
+		}
+		if strings.Index(content, search) == strings.LastIndex(content, search) {
+			return search, nil
+		}
+		// The single candidate occurs at multiple positions; a later, stricter
+		// strategy may still resolve a unique span, so keep cascading.
+	}
+	if !found {
+		return "", errEditFuzzyNotFound
+	}
+	return "", errEditFuzzyAmbiguous
+}
+
+// isDisproportionateEditMatch guards against anchor-style replacers matching a
+// span far larger than the text the model asked to replace (e.g. first/last
+// line anchors bridging hundreds of unrelated lines).
+func isDisproportionateEditMatch(search, find string) bool {
+	findLines := strings.Count(find, "\n") + 1
+	searchLines := strings.Count(search, "\n") + 1
+	limit := findLines + 3
+	if findLines*2 > limit {
+		limit = findLines * 2
+	}
+	if searchLines >= limit {
+		return true
+	}
+	if findLines == 1 {
+		return false
+	}
+	searchTrimmed := len(strings.TrimSpace(search))
+	findTrimmed := len(strings.TrimSpace(find))
+	byteLimit := findTrimmed + 500
+	if findTrimmed*4 > byteLimit {
+		byteLimit = findTrimmed * 4
+	}
+	return searchTrimmed > byteLimit
+}
+
+// splitFindLines splits find into lines, dropping the trailing empty line a
+// trailing newline produces so windows align with real content lines.
+func splitFindLines(find string) []string {
+	lines := strings.Split(find, "\n")
+	if len(lines) > 0 && lines[len(lines)-1] == "" {
+		lines = lines[:len(lines)-1]
+	}
+	return lines
+}
+
+// lineTrimmedReplacer matches when every line equals the corresponding
+// old_string line after trimming surrounding whitespace.
+func lineTrimmedReplacer(content, find string) []string {
+	contentLines := strings.Split(content, "\n")
+	findLines := splitFindLines(find)
+	if len(findLines) == 0 {
+		return nil
+	}
+	var candidates []string
+	for i := 0; i+len(findLines) <= len(contentLines); i++ {
+		matches := true
+		for j := range findLines {
+			if strings.TrimSpace(contentLines[i+j]) != strings.TrimSpace(findLines[j]) {
+				matches = false
+				break
+			}
+		}
+		if matches {
+			candidates = append(candidates, strings.Join(contentLines[i:i+len(findLines)], "\n"))
+		}
+	}
+	return candidates
+}
+
+// blockAnchorReplacer anchors on the first and last lines (trimmed) and
+// accepts the block when the middle lines average >= the similarity threshold
+// (Levenshtein-based), tolerating a slightly misremembered interior.
+func blockAnchorReplacer(content, find string) []string {
+	findLines := splitFindLines(find)
+	if len(findLines) < 3 {
+		return nil
+	}
+	contentLines := strings.Split(content, "\n")
+	firstAnchor := strings.TrimSpace(findLines[0])
+	lastAnchor := strings.TrimSpace(findLines[len(findLines)-1])
+	searchBlockSize := len(findLines)
+	maxLineDelta := searchBlockSize / 4
+	if maxLineDelta < 1 {
+		maxLineDelta = 1
+	}
+
+	type span struct{ start, end int }
+	var candidates []span
+	for i := 0; i < len(contentLines); i++ {
+		if strings.TrimSpace(contentLines[i]) != firstAnchor {
+			continue
+		}
+		// Only the first occurrence of the last anchor after this start counts,
+		// mirroring the reference implementation: a farther-away closing line is
+		// assumed to close a different block.
+		for j := i + 2; j < len(contentLines); j++ {
+			if strings.TrimSpace(contentLines[j]) != lastAnchor {
+				continue
+			}
+			actualBlockSize := j - i + 1
+			delta := actualBlockSize - searchBlockSize
+			if delta < 0 {
+				delta = -delta
+			}
+			if delta <= maxLineDelta {
+				candidates = append(candidates, span{start: i, end: j})
+			}
+			break
+		}
+	}
+	if len(candidates) == 0 {
+		return nil
+	}
+
+	middleSimilarity := func(candidate span) float64 {
+		actualBlockSize := candidate.end - candidate.start + 1
+		linesToCheck := searchBlockSize - 2
+		if actualBlockSize-2 < linesToCheck {
+			linesToCheck = actualBlockSize - 2
+		}
+		if linesToCheck <= 0 {
+			return 1.0
+		}
+		similarity := 0.0
+		for j := 1; j < searchBlockSize-1 && j < actualBlockSize-1; j++ {
+			contentLine := strings.TrimSpace(contentLines[candidate.start+j])
+			findLine := strings.TrimSpace(findLines[j])
+			maxLen := len(contentLine)
+			if len(findLine) > maxLen {
+				maxLen = len(findLine)
+			}
+			if maxLen == 0 {
+				continue
+			}
+			distance := levenshtein(contentLine, findLine)
+			similarity += 1 - float64(distance)/float64(maxLen)
+		}
+		return similarity / float64(linesToCheck)
+	}
+
+	// Return EVERY candidate that clears the similarity threshold, best first.
+	// Picking only the best would hide competing blocks from fuzzyEditMatch's
+	// distinct-candidate ambiguity check and could silently edit the wrong
+	// block; with all qualifying spans surfaced, one clear winner still
+	// resolves (single candidate) while two plausible blocks are rejected as
+	// ambiguous. replaceAll consumers take the first (most similar) span.
+	type scoredSpan struct {
+		span       span
+		similarity float64
+	}
+	var qualifying []scoredSpan
+	for _, candidate := range candidates {
+		if s := middleSimilarity(candidate); s >= editAnchorSimilarityThreshold {
+			qualifying = append(qualifying, scoredSpan{span: candidate, similarity: s})
+		}
+	}
+	if len(qualifying) == 0 {
+		return nil
+	}
+	sort.SliceStable(qualifying, func(i, j int) bool {
+		return qualifying[i].similarity > qualifying[j].similarity
+	})
+	spans := make([]string, 0, len(qualifying))
+	for _, scored := range qualifying {
+		spans = append(spans, strings.Join(contentLines[scored.span.start:scored.span.end+1], "\n"))
+	}
+	return spans
+}
+
+var editWhitespaceRun = regexp.MustCompile(`\s+`)
+
+func normalizeEditWhitespace(text string) string {
+	return strings.TrimSpace(editWhitespaceRun.ReplaceAllString(text, " "))
+}
+
+// whitespaceNormalizedReplacer matches after collapsing all whitespace runs to
+// a single space: full lines, sub-line spans (via a word-boundary regex), and
+// multi-line windows.
+func whitespaceNormalizedReplacer(content, find string) []string {
+	normalizedFind := normalizeEditWhitespace(find)
+	if normalizedFind == "" {
+		return nil
+	}
+	contentLines := strings.Split(content, "\n")
+	var candidates []string
+	var subLinePattern *regexp.Regexp
+	for _, line := range contentLines {
+		normalizedLine := normalizeEditWhitespace(line)
+		if normalizedLine == normalizedFind {
+			candidates = append(candidates, line)
+			continue
+		}
+		if !strings.Contains(normalizedLine, normalizedFind) {
+			continue
+		}
+		if subLinePattern == nil {
+			words := strings.Fields(find)
+			quoted := make([]string, len(words))
+			for i, word := range words {
+				quoted[i] = regexp.QuoteMeta(word)
+			}
+			pattern, err := regexp.Compile(strings.Join(quoted, `\s+`))
+			if err != nil {
+				continue
+			}
+			subLinePattern = pattern
+		}
+		if match := subLinePattern.FindString(line); match != "" {
+			candidates = append(candidates, match)
+		}
+	}
+
+	findLines := strings.Split(find, "\n")
+	if len(findLines) > 1 {
+		for i := 0; i+len(findLines) <= len(contentLines); i++ {
+			block := strings.Join(contentLines[i:i+len(findLines)], "\n")
+			if normalizeEditWhitespace(block) == normalizedFind {
+				candidates = append(candidates, block)
+			}
+		}
+	}
+	return candidates
+}
+
+// stripCommonIndentation removes the minimum leading-whitespace width shared
+// by all non-empty lines, so blocks match regardless of their nesting depth.
+func stripCommonIndentation(text string) string {
+	lines := strings.Split(text, "\n")
+	minIndent := -1
+	for _, line := range lines {
+		if strings.TrimSpace(line) == "" {
+			continue
+		}
+		indent := len(line) - len(strings.TrimLeft(line, " \t"))
+		if minIndent < 0 || indent < minIndent {
+			minIndent = indent
+		}
+	}
+	if minIndent <= 0 {
+		return text
+	}
+	stripped := make([]string, len(lines))
+	for i, line := range lines {
+		if strings.TrimSpace(line) == "" {
+			stripped[i] = line
+			continue
+		}
+		stripped[i] = line[minIndent:]
+	}
+	return strings.Join(stripped, "\n")
+}
+
+func indentationFlexibleReplacer(content, find string) []string {
+	normalizedFind := stripCommonIndentation(find)
+	contentLines := strings.Split(content, "\n")
+	findLines := strings.Split(find, "\n")
+	var candidates []string
+	for i := 0; i+len(findLines) <= len(contentLines); i++ {
+		block := strings.Join(contentLines[i:i+len(findLines)], "\n")
+		if stripCommonIndentation(block) == normalizedFind {
+			candidates = append(candidates, block)
+		}
+	}
+	return candidates
+}
+
+var editEscapeSequence = regexp.MustCompile("\\\\(n|t|r|'|\"|`|\\\\|\\n|\\$)")
+
+// unescapeEditString undoes one level of string escaping (\n, \t, \", \\, a
+// backslash-newline continuation, \$) — the model sometimes reproduces file
+// content as it appeared inside a quoted string literal.
+func unescapeEditString(text string) string {
+	return editEscapeSequence.ReplaceAllStringFunc(text, func(match string) string {
+		switch match[1:] {
+		case "n":
+			return "\n"
+		case "t":
+			return "\t"
+		case "r":
+			return "\r"
+		case "\n":
+			return "\n"
+		default:
+			// ', ", `, \, $ all unescape to themselves.
+			return match[1:]
+		}
+	})
+}
+
+func escapeNormalizedReplacer(content, find string) []string {
+	unescapedFind := unescapeEditString(find)
+	var candidates []string
+	if strings.Contains(content, unescapedFind) {
+		candidates = append(candidates, unescapedFind)
+	}
+	contentLines := strings.Split(content, "\n")
+	findLines := strings.Split(unescapedFind, "\n")
+	for i := 0; i+len(findLines) <= len(contentLines); i++ {
+		block := strings.Join(contentLines[i:i+len(findLines)], "\n")
+		if unescapeEditString(block) == unescapedFind {
+			candidates = append(candidates, block)
+		}
+	}
+	return candidates
+}
+
+// trimmedBoundaryReplacer tolerates stray leading/trailing whitespace (often
+// blank lines) around an otherwise-exact old_string.
+func trimmedBoundaryReplacer(content, find string) []string {
+	trimmedFind := strings.TrimSpace(find)
+	if trimmedFind == find || trimmedFind == "" {
+		return nil
+	}
+	var candidates []string
+	if strings.Contains(content, trimmedFind) {
+		candidates = append(candidates, trimmedFind)
+	}
+	contentLines := strings.Split(content, "\n")
+	findLines := strings.Split(find, "\n")
+	for i := 0; i+len(findLines) <= len(contentLines); i++ {
+		block := strings.Join(contentLines[i:i+len(findLines)], "\n")
+		if strings.TrimSpace(block) == trimmedFind {
+			candidates = append(candidates, block)
+		}
+	}
+	return candidates
+}
+
+// contextAwareReplacer anchors on the first and last lines and accepts an
+// equal-length block when at least half of its non-empty middle lines match
+// after trimming — a cheaper, stricter cousin of blockAnchorReplacer.
+func contextAwareReplacer(content, find string) []string {
+	findLines := splitFindLines(find)
+	if len(findLines) < 3 {
+		return nil
+	}
+	contentLines := strings.Split(content, "\n")
+	firstAnchor := strings.TrimSpace(findLines[0])
+	lastAnchor := strings.TrimSpace(findLines[len(findLines)-1])
+	var candidates []string
+	for i := 0; i < len(contentLines); i++ {
+		if strings.TrimSpace(contentLines[i]) != firstAnchor {
+			continue
+		}
+		for j := i + 2; j < len(contentLines); j++ {
+			if strings.TrimSpace(contentLines[j]) != lastAnchor {
+				continue
+			}
+			if j-i+1 == len(findLines) {
+				matching, nonEmpty := 0, 0
+				for k := 1; k < len(findLines)-1; k++ {
+					blockLine := strings.TrimSpace(contentLines[i+k])
+					findLine := strings.TrimSpace(findLines[k])
+					if blockLine == "" && findLine == "" {
+						continue
+					}
+					nonEmpty++
+					if blockLine == findLine {
+						matching++
+					}
+				}
+				if nonEmpty == 0 || float64(matching)/float64(nonEmpty) >= 0.5 {
+					candidates = append(candidates, strings.Join(contentLines[i:j+1], "\n"))
+				}
+			}
+			break
+		}
+	}
+	return candidates
+}
+
+// adaptReplacementToSpan re-shapes the model's replacement to the span a
+// tolerant matcher resolved. When old_string only matched after normalization,
+// new_string was written at old_string's (wrong) shape, so applying it raw
+// would strip the file's indentation or drop a trailing CR:
+//
+//  1. Uniform re-indent: when every span line equals delta + the corresponding
+//     find line's indentation (the line-trimmed / indentation-flexible shapes),
+//     the same delta is prepended to every non-blank replacement line. Any
+//     line that breaks the uniform-delta relationship disables the shift —
+//     block-anchor matches with a drifted interior are left untouched.
+//  2. Trailing CR: a span from a CRLF file ends mid-line at "\r" (candidates
+//     are built by joining lines split on "\n"); the replacement gets the same
+//     trailing "\r" so the file's CRLF pairs stay intact.
+func adaptReplacementToSpan(span, find, replacement string) string {
+	if delta, ok := uniformIndentDelta(span, find); ok && delta != "" {
+		lines := strings.Split(replacement, "\n")
+		for i, line := range lines {
+			if strings.TrimSpace(line) == "" {
+				continue
+			}
+			lines[i] = delta + line
+		}
+		replacement = strings.Join(lines, "\n")
+	}
+	if strings.HasSuffix(span, "\r") && !strings.HasSuffix(replacement, "\r") {
+		replacement += "\r"
+	}
+	return replacement
+}
+
+// uniformIndentDelta returns the indentation prefix that, prepended to every
+// non-blank find line, yields the corresponding span line's indentation. ok is
+// false when line counts differ, any line pair disagrees on the delta, or the
+// span is not simply a uniformly deeper-indented copy of find.
+func uniformIndentDelta(span, find string) (string, bool) {
+	spanLines := strings.Split(span, "\n")
+	findLines := splitFindLines(find)
+	if len(spanLines) != len(findLines) {
+		return "", false
+	}
+	leadingWhitespace := func(line string) string {
+		return line[:len(line)-len(strings.TrimLeft(line, " \t"))]
+	}
+	delta := ""
+	haveDelta := false
+	for i := range findLines {
+		spanLine := strings.TrimSuffix(spanLines[i], "\r")
+		findLine := strings.TrimSuffix(findLines[i], "\r")
+		if strings.TrimSpace(spanLine) == "" && strings.TrimSpace(findLine) == "" {
+			continue
+		}
+		spanIndent := leadingWhitespace(spanLine)
+		findIndent := leadingWhitespace(findLine)
+		if !strings.HasSuffix(spanIndent, findIndent) {
+			return "", false
+		}
+		lineDelta := spanIndent[:len(spanIndent)-len(findIndent)]
+		if !haveDelta {
+			delta = lineDelta
+			haveDelta = true
+			continue
+		}
+		if lineDelta != delta {
+			return "", false
+		}
+	}
+	return delta, haveDelta
+}
+
+// levenshtein computes edit distance with a two-row rolling matrix.
+func levenshtein(a, b string) int {
+	if a == "" {
+		return len(b)
+	}
+	if b == "" {
+		return len(a)
+	}
+	previous := make([]int, len(b)+1)
+	current := make([]int, len(b)+1)
+	for j := range previous {
+		previous[j] = j
+	}
+	for i := 1; i <= len(a); i++ {
+		current[0] = i
+		for j := 1; j <= len(b); j++ {
+			cost := 1
+			if a[i-1] == b[j-1] {
+				cost = 0
+			}
+			minimum := previous[j] + 1
+			if current[j-1]+1 < minimum {
+				minimum = current[j-1] + 1
+			}
+			if previous[j-1]+cost < minimum {
+				minimum = previous[j-1] + cost
+			}
+			current[j] = minimum
+		}
+		previous, current = current, previous
+	}
+	return previous[len(b)]
+}

internal/tools/edit_replacers_test.go

@@ -0,0 +1,369 @@
+package tools
+
+import (
+	"context"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func runEdit(t *testing.T, dir, initial string, args map[string]any) (Result, string) {
+	t.Helper()
+	path := filepath.Join(dir, "target.txt")
+	if err := os.WriteFile(path, []byte(initial), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	if args["path"] == nil {
+		args["path"] = "target.txt"
+	}
+	result := NewEditFileTool(dir).Run(context.Background(), args)
+	after, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return result, string(after)
+}
+
+func TestEditExactMatchStillPreferred(t *testing.T) {
+	// The fast path must keep byte-exact semantics: an exact match wins even
+	// when fuzzy strategies would also find candidates elsewhere.
+	result, after := runEdit(t, t.TempDir(), "alpha\nbeta\ngamma\n", map[string]any{
+		"old_string": "beta",
+		"new_string": "delta",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if after != "alpha\ndelta\ngamma\n" {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyLineTrimmed(t *testing.T) {
+	// Model reproduced the lines without the file's leading indentation; the
+	// line-trimmed strategy must find the real indented span and preserve the
+	// file's own indentation on the replaced region boundary.
+	initial := "func a() {\n\tx := 1\n\treturn x\n}\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "x := 1\nreturn x",
+		"new_string": "y := 2\nreturn y",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	// new_string was written at old_string's outdented shape; the span's
+	// indentation must be re-applied so the file keeps its tabs.
+	if !strings.Contains(after, "\ty := 2\n\treturn y\n}") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+	if strings.Contains(after, "x := 1") {
+		t.Fatalf("old span survived: %q", after)
+	}
+}
+
+func TestEditFuzzyWhitespaceNormalizedSingleLine(t *testing.T) {
+	// Whitespace runs collapsed: "x  :=   1" in the file, single spaces from
+	// the model.
+	initial := "start\nx  :=   1\nend\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "x := 1",
+		"new_string": "x := 2",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if !strings.Contains(after, "x := 2") || strings.Contains(after, ":=   1") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyBlockAnchorToleratesMiddleDrift(t *testing.T) {
+	// First and last lines anchor the block; one interior line differs slightly
+	// (comment text drifted). Levenshtein similarity keeps it above 0.65.
+	initial := strings.Join([]string{
+		"func handler(w http.ResponseWriter, r *http.Request) {",
+		"\t// write the response body to the client",
+		"\tw.WriteHeader(http.StatusOK)",
+		"\tfmt.Fprint(w, \"done\")",
+		"}",
+		"",
+	}, "\n")
+	find := strings.Join([]string{
+		"func handler(w http.ResponseWriter, r *http.Request) {",
+		"\t// write the response body to client",
+		"\tw.WriteHeader(http.StatusOK)",
+		"\tfmt.Fprint(w, \"done\")",
+		"}",
+	}, "\n")
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": find,
+		"new_string": "func handler(w http.ResponseWriter, r *http.Request) {\n\tw.WriteHeader(http.StatusNoContent)\n}",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if !strings.Contains(after, "StatusNoContent") || strings.Contains(after, "StatusOK") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyIndentationFlexible(t *testing.T) {
+	// The whole block sits one nesting level deeper in the file than in the
+	// model's old_string, with interior relative indentation preserved.
+	initial := "if ok {\n\t\tfor i := range xs {\n\t\t\tsum += xs[i]\n\t\t}\n}\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "for i := range xs {\n\tsum += xs[i]\n}",
+		"new_string": "sum += total(xs)",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	// The replacement lands at the span's real depth (two tabs), not at
+	// old_string's outdented depth.
+	if !strings.Contains(after, "\t\tsum += total(xs)") || strings.Contains(after, "range xs") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyCRLFLineTrimmedPreservesCarriageReturns(t *testing.T) {
+	// CRLF file + a multi-line outdented old_string: the resolved span carries
+	// tabs and trailing CRs; the replacement must inherit both so the file's
+	// CRLF pairs and indentation survive the edit.
+	initial := "func a() {\r\n\tx := 1\r\n\treturn x\r\n}\r\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "x := 1\nreturn x",
+		"new_string": "y := 2\nreturn y",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if !strings.Contains(after, "\ty := 2\r\n\treturn y\r\n}") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+	if strings.Contains(after, "\n\treturn y\n") {
+		t.Fatalf("bare LF introduced into CRLF file: %q", after)
+	}
+}
+
+func TestUniformIndentDelta(t *testing.T) {
+	cases := []struct {
+		span, find string
+		want       string
+		ok         bool
+	}{
+		{"\tx := 1\n\treturn x", "x := 1\nreturn x", "\t", true},
+		{"\t\tfor {\n\t\t\tgo()\n\t\t}", "for {\n\tgo()\n}", "\t\t", true},
+		{"x := 1", "x := 1", "", true},                          // no shift needed
+		{"\tx := 1\n  return x", "x := 1\nreturn x", "", false}, // mixed delta
+		{"\ta\n\tb\n\tc", "a\nb", "", false},                    // line-count mismatch
+		{"  a", "\ta", "", false},                               // find indent not a suffix
+	}
+	for _, c := range cases {
+		got, ok := uniformIndentDelta(c.span, c.find)
+		if ok != c.ok || (ok && got != c.want) {
+			t.Errorf("uniformIndentDelta(%q, %q) = %q,%v want %q,%v", c.span, c.find, got, ok, c.want, c.ok)
+		}
+	}
+}
+
+func TestAdaptReplacementLeavesNonUniformSpansAlone(t *testing.T) {
+	// Block-anchor style match with a drifted interior: no uniform delta, so
+	// the replacement must pass through untouched.
+	span := "\tfunc h() {\n\t\t// drifted comment\n\t}"
+	find := "func h() {\n// different comment\n}"
+	if got := adaptReplacementToSpan(span, find, "replacement()"); got != "replacement()" {
+		t.Fatalf("non-uniform span must not re-indent: %q", got)
+	}
+}
+
+func TestEditFuzzyEscapeNormalized(t *testing.T) {
+	// Model reproduced the line as it would appear inside a quoted string
+	// literal, with escaped quotes.
+	initial := "console.log(\"hello world\")\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": `console.log(\"hello world\")`,
+		"new_string": `console.log("goodbye")`,
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if !strings.Contains(after, `console.log("goodbye")`) {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyTrimmedBoundary(t *testing.T) {
+	// Stray whitespace around an otherwise-exact old_string; "  two  " is not a
+	// substring of the file, so only the fuzzy cascade can resolve it.
+	initial := "one\ntwo\nthree\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "  two  ",
+		"new_string": "2",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if after != "one\n2\nthree\n" {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestTrimmedBoundaryReplacerYieldsTrimmedSpan(t *testing.T) {
+	candidates := trimmedBoundaryReplacer("one\ntwo\nthree\n", "\ntwo\n")
+	if len(candidates) == 0 || candidates[0] != "two" {
+		t.Fatalf("expected trimmed candidate \"two\", got %v", candidates)
+	}
+	if trimmedBoundaryReplacer("anything", "already-trimmed") != nil {
+		t.Fatal("already-trimmed find must yield no candidates")
+	}
+}
+
+// Two identical blocks that only match old_string after indentation-tolerant
+// matching: exact search finds nothing, fuzzy finds both.
+const ambiguousBlocks = "\tif x {\n\t\tgo()\n\t}\nmid\n\tif x {\n\t\tgo()\n\t}\n"
+
+func TestEditFuzzyAmbiguousReportsError(t *testing.T) {
+	// No strategy can disambiguate identical blocks, so the tool must refuse
+	// rather than guess, and the file must be untouched.
+	result, after := runEdit(t, t.TempDir(), ambiguousBlocks, map[string]any{
+		"old_string": "if x {\n\tgo()\n}",
+		"new_string": "stop()",
+	})
+	if result.Status != StatusError || !strings.Contains(result.Output, "multiple locations") {
+		t.Fatalf("expected ambiguity error, got %q", result.Output)
+	}
+	if after != ambiguousBlocks {
+		t.Fatalf("file must be unchanged, got %q", after)
+	}
+}
+
+func TestEditFuzzyReplaceAll(t *testing.T) {
+	// replace_all applies the fuzzy-resolved span at every occurrence.
+	result, after := runEdit(t, t.TempDir(), ambiguousBlocks, map[string]any{
+		"old_string":  "if x {\n\tgo()\n}",
+		"new_string":  "stop()",
+		"replace_all": true,
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if strings.Count(after, "stop()") != 2 || strings.Contains(after, "go()") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestEditFuzzyNotFoundKeepsExactError(t *testing.T) {
+	result, after := runEdit(t, t.TempDir(), "alpha\n", map[string]any{
+		"old_string": "omega",
+		"new_string": "x",
+	})
+	if result.Status != StatusError || !strings.Contains(result.Output, "Could not find the exact string") {
+		t.Fatalf("expected not-found error, got %q", result.Output)
+	}
+	if after != "alpha\n" {
+		t.Fatalf("file must be unchanged, got %q", after)
+	}
+}
+
+func TestEditFuzzyCRLFFile(t *testing.T) {
+	// CRLF file + LF old_string with indentation drift: the CRLF translation
+	// feeds the cascade and the replacement preserves CRLF endings.
+	initial := "func a() {\r\n\tx := 1\r\n}\r\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "x := 1",
+		"new_string": "x := 2",
+	})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %q", result.Output)
+	}
+	if !strings.Contains(after, "\tx := 2\r\n") {
+		t.Fatalf("unexpected content: %q", after)
+	}
+}
+
+func TestIsDisproportionateEditMatch(t *testing.T) {
+	// A candidate span that dwarfs old_string must be refused: anchors bridging
+	// unrelated code would otherwise delete it all.
+	big := strings.Repeat("line\n", 10)
+	if !isDisproportionateEditMatch(big, "a\nb\nc") {
+		t.Fatal("10-line span for 3-line find must be disproportionate")
+	}
+	if isDisproportionateEditMatch("a\nb\nc\nd", "a\nb\nc") {
+		t.Fatal("one extra line within tolerance must be allowed")
+	}
+	if isDisproportionateEditMatch("single line", "single") {
+		t.Fatal("single-line finds are exempt from the byte-length guard")
+	}
+	if !isDisproportionateEditMatch("xx\n"+strings.Repeat("y", 900)+"\nzz", "xx\nab\nzz") {
+		t.Fatal("byte-length blowup must be disproportionate")
+	}
+}
+
+func TestLevenshtein(t *testing.T) {
+	cases := []struct {
+		a, b string
+		want int
+	}{
+		{"", "", 0},
+		{"abc", "", 3},
+		{"", "abc", 3},
+		{"kitten", "sitting", 3},
+		{"same", "same", 0},
+	}
+	for _, c := range cases {
+		if got := levenshtein(c.a, c.b); got != c.want {
+			t.Fatalf("levenshtein(%q,%q) = %d, want %d", c.a, c.b, got, c.want)
+		}
+	}
+}
+
+func TestEditFuzzyDistinctCandidatesAreAmbiguous(t *testing.T) {
+	// Two same-content blocks at DIFFERENT indentation: each resolved span is
+	// distinct and occurs exactly once, so the old literal-uniqueness check
+	// passed and silently edited the first block. Distinct-shaped candidates
+	// from one strategy must instead report ambiguity and leave the file alone.
+	initial := "\tif x {\n\t\tgo()\n\t}\nmid\n\t\tif x {\n\t\t\tgo()\n\t\t}\n"
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": "if x {\n\tgo()\n}",
+		"new_string": "stop()",
+	})
+	if result.Status != StatusError || !strings.Contains(result.Output, "multiple locations") {
+		t.Fatalf("distinct fuzzy candidates must be ambiguous, got %q", result.Output)
+	}
+	if after != initial {
+		t.Fatalf("file must be unchanged, got %q", after)
+	}
+}
+
+func TestEditFuzzyBlockAnchorTwoPlausibleBlocksAmbiguous(t *testing.T) {
+	// Two blocks share the same first/last anchor lines and BOTH interiors sit
+	// above the similarity threshold. Picking the "best" one would silently
+	// edit a block the model may not have meant; both must surface so the
+	// cascade reports ambiguity and the file stays untouched.
+	initial := strings.Join([]string{
+		"func setup(cfg Config) {",
+		"\tvalue := compute(alpha, beta, gamma)",
+		"}",
+		"",
+		"func setup(cfg Config) {",
+		"\tvalue := compute(alpha, delta, gamma)",
+		"}",
+		"",
+	}, "\n")
+	find := strings.Join([]string{
+		"func setup(cfg Config) {",
+		"\tvalue := compute(alpha, omega, gamma)",
+		"}",
+	}, "\n")
+	result, after := runEdit(t, t.TempDir(), initial, map[string]any{
+		"old_string": find,
+		"new_string": "func setup(cfg Config) {}",
+	})
+	if result.Status != StatusError || !strings.Contains(result.Output, "multiple locations") {
+		t.Fatalf("two plausible anchor blocks must be ambiguous, got %q", result.Output)
+	}
+	if after != initial {
+		t.Fatalf("file must be unchanged, got %q", after)
+	}
+}

internal/tools/exec_command.go

@@ -908,6 +908,15 @@ func formatExecCommandOutput(output string, sessionID int, exited bool, exitCode
 }
 
 func truncateExecOutput(output string, maxOutputTokens int) (string, bool) {
+	return truncateExecOutputSpill(output, maxOutputTokens, "exec_command")
+}
+
+// truncateExecOutputSpill keeps a head/tail window of the output within the
+// token budget and, on truncation, spills the full output to disk so the model
+// can grep/read the elided middle instead of re-running the command with a
+// bigger budget. The spill is best-effort: when it fails the notice simply
+// omits the file hint.
+func truncateExecOutputSpill(output string, maxOutputTokens int, toolName string) (string, bool) {
 	if maxOutputTokens <= 0 {
 		maxOutputTokens = defaultMaxOutputTokens
 	}
@@ -915,9 +924,13 @@ func truncateExecOutput(output string, maxOutputTokens int) (string, bool) {
 	if len(output) <= maxBytes {
 		return output, false
 	}
+	notice := "\n[zero] output truncated\n"
+	if spillPath := spillTruncatedOutput(toolName, output); spillPath != "" {
+		notice = "\n[zero] output truncated — full output saved to " + spillPath + " (grep or read_file it instead of re-running)\n"
+	}
 	head := maxBytes / 2
 	tail := maxBytes - head
-	return utf8Prefix(output, head) + "\n[zero] output truncated\n" + utf8Suffix(output, tail), true
+	return utf8Prefix(output, head) + notice + utf8Suffix(output, tail), true
 }
 
 func utf8Prefix(value string, maxBytes int) string {

internal/tools/format_on_write.go

@@ -0,0 +1,97 @@
+package tools
+
+import (
+	"context"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strings"
+	"time"
+)
+
+// Format-on-write for the mutating file tools. When enabled, a successful
+// edit_file/write_file runs the language's standard formatter on the file it
+// just wrote, so the model's output always lands in project-canonical style
+// and never fails a CI format check it cannot see. Off by default (set
+// ZERO_FORMAT_ON_WRITE=1): auto-reformatting changes bytes the model did not
+// write, which strict workflows may not want.
+//
+// Ordering matters: formatting runs BEFORE the FileTracker re-baseline, and
+// the caller records the POST-format content. Formatting after the baseline
+// would make the very next edit look like an external modification and trip
+// the conflict guard.
+
+// formatOnWriteTimeout bounds one formatter run; a wedged formatter must never
+// hang a tool call. On timeout the unformatted write stands.
+const formatOnWriteTimeout = 10 * time.Second
+
+// formatterCommands maps a file extension to the formatter argv; the file path
+// is appended as the final argument. Only in-place, config-respecting,
+// community-standard formatters — a missing binary silently skips formatting.
+var formatterCommands = map[string][]string{
+	".go":    {"gofmt", "-w"},
+	".rs":    {"rustfmt"},
+	".py":    {"ruff", "format", "--quiet"},
+	".ts":    {"prettier", "--log-level", "silent", "--write"},
+	".tsx":   {"prettier", "--log-level", "silent", "--write"},
+	".js":    {"prettier", "--log-level", "silent", "--write"},
+	".jsx":   {"prettier", "--log-level", "silent", "--write"},
+	".json":  {"prettier", "--log-level", "silent", "--write"},
+	".css":   {"prettier", "--log-level", "silent", "--write"},
+	".scss":  {"prettier", "--log-level", "silent", "--write"},
+	".html":  {"prettier", "--log-level", "silent", "--write"},
+	".md":    {"prettier", "--log-level", "silent", "--write"},
+	".yaml":  {"prettier", "--log-level", "silent", "--write"},
+	".yml":   {"prettier", "--log-level", "silent", "--write"},
+	".zig":   {"zig", "fmt"},
+	".dart":  {"dart", "format"},
+	".tf":    {"terraform", "fmt"},
+	".gleam": {"gleam", "format"},
+	".sh":    {"shfmt", "-w"},
+	".bash":  {"shfmt", "-w"},
+	".c":     {"clang-format", "-i"},
+	".h":     {"clang-format", "-i"},
+	".cpp":   {"clang-format", "-i"},
+	".hpp":   {"clang-format", "-i"},
+	".cc":    {"clang-format", "-i"},
+	".kt":    {"ktlint", "-F"},
+	".swift": {"swiftformat"},
+	".lua":   {"stylua"},
+}
+
+// formatOnWriteEnabled reports whether the opt-in env toggle is set.
+func formatOnWriteEnabled() bool {
+	value := strings.TrimSpace(os.Getenv("ZERO_FORMAT_ON_WRITE"))
+	return value != "" && value != "0" && !strings.EqualFold(value, "false")
+}
+
+// maybeFormatWrittenFile runs the configured formatter for absolutePath (when
+// enabled and on PATH) and returns the file's content afterwards. Best-effort
+// throughout: any failure — no formatter, formatter error, timeout, unreadable
+// result — returns writtenContent so the caller's state matches the last write
+// it performed itself.
+func maybeFormatWrittenFile(ctx context.Context, absolutePath string, writtenContent string) string {
+	if !formatOnWriteEnabled() {
+		return writtenContent
+	}
+	command, ok := formatterCommands[strings.ToLower(filepath.Ext(absolutePath))]
+	if !ok {
+		return writtenContent
+	}
+	if _, err := exec.LookPath(command[0]); err != nil {
+		return writtenContent
+	}
+	formatCtx, cancel := context.WithTimeout(ctx, formatOnWriteTimeout)
+	defer cancel()
+	arguments := append(append([]string(nil), command[1:]...), absolutePath)
+	formatter := exec.CommandContext(formatCtx, command[0], arguments...)
+	formatter.Dir = filepath.Dir(absolutePath)
+	if err := formatter.Run(); err != nil {
+		return writtenContent
+	}
+	formatted, err := os.ReadFile(absolutePath)
+	if err != nil {
+		return writtenContent
+	}
+	return string(formatted)
+}

internal/tools/format_on_write_test.go

@@ -0,0 +1,82 @@
+package tools
+
+import (
+	"context"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+// gofmt ships with the Go toolchain, so it is the one formatter guaranteed to
+// exist wherever these tests run.
+func requireGofmt(t *testing.T) {
+	t.Helper()
+	if _, err := exec.LookPath("gofmt"); err != nil {
+		t.Skip("gofmt not on PATH")
+	}
+}
+
+func TestFormatOnWriteDisabledByDefault(t *testing.T) {
+	requireGofmt(t)
+	t.Setenv("ZERO_FORMAT_ON_WRITE", "")
+	dir := t.TempDir()
+	ugly := "package a\n\nfunc  A( ) {   }\n"
+
+	result := NewWriteFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":    "a.go",
+		"content": ugly,
+	}, RunOptions{})
+	if result.Status != StatusOK {
+		t.Fatalf("write failed: %q", result.Output)
+	}
+	onDisk, err := os.ReadFile(filepath.Join(dir, "a.go"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(onDisk) != ugly {
+		t.Fatalf("formatting must be off by default, got %q", onDisk)
+	}
+}
+
+func TestFormatOnWriteFormatsAndKeepsTrackerConsistent(t *testing.T) {
+	requireGofmt(t)
+	t.Setenv("ZERO_FORMAT_ON_WRITE", "1")
+	dir := t.TempDir()
+	tracker := NewFileTracker()
+
+	write := NewWriteFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":    "a.go",
+		"content": "package a\n\nfunc  A( ) {   }\n",
+	}, RunOptions{FileTracker: tracker})
+	if write.Status != StatusOK {
+		t.Fatalf("write failed: %q", write.Output)
+	}
+	onDisk, err := os.ReadFile(filepath.Join(dir, "a.go"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !strings.Contains(string(onDisk), "func A() {") {
+		t.Fatalf("expected gofmt-formatted content, got %q", onDisk)
+	}
+
+	// The tracker must have been re-baselined to the POST-format content: a
+	// follow-up edit must not trip the external-modification conflict guard.
+	edit := NewEditFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":       "a.go",
+		"old_string": "func A() {",
+		"new_string": "func B() {",
+	}, RunOptions{FileTracker: tracker})
+	if edit.Status != StatusOK {
+		t.Fatalf("follow-up edit must not conflict after formatting: %q", edit.Output)
+	}
+}
+
+func TestFormatOnWriteSkipsUnknownExtensions(t *testing.T) {
+	t.Setenv("ZERO_FORMAT_ON_WRITE", "1")
+	content := maybeFormatWrittenFile(context.Background(), filepath.Join(t.TempDir(), "notes.xyz"), "raw   text")
+	if content != "raw   text" {
+		t.Fatalf("unknown extension must pass through: %q", content)
+	}
+}

internal/tools/glob.go

@@ -45,18 +45,18 @@ func NewScopedGlobTool(workspaceRoot string, scope PathScope) Tool {
 	}
 }
 
-func (tool globTool) Run(_ context.Context, args map[string]any) Result {
-	return tool.runWith(args, readExcluder{})
+func (tool globTool) Run(ctx context.Context, args map[string]any) Result {
+	return tool.runWith(ctx, args, readExcluder{})
 }
 
 // RunWithSandbox runs glob while skipping subtrees the sandbox policy denies
 // reads to (DenyRead). With no DenyRead configured the excluder is a no-op and
 // behavior is unchanged.
-func (tool globTool) RunWithSandbox(_ context.Context, args map[string]any, engine *sandbox.Engine) Result {
-	return tool.runWith(args, sandboxReadExcluder(engine))
+func (tool globTool) RunWithSandbox(ctx context.Context, args map[string]any, engine *sandbox.Engine) Result {
+	return tool.runWith(ctx, args, sandboxReadExcluder(engine))
 }
 
-func (tool globTool) runWith(args map[string]any, exclude readExcluder) Result {
+func (tool globTool) runWith(ctx context.Context, args map[string]any, exclude readExcluder) Result {
 	pattern, err := aliasedStringArg(args, []string{"pattern", "glob", "match", "query", "expression"}, "", true, false)
 	if err != nil {
 		return errorResult("Error: Invalid arguments for glob: " + err.Error())
@@ -88,8 +88,11 @@ func (tool globTool) runWith(args map[string]any, exclude readExcluder) Result {
 		return errorResult("Error running glob " + fmt.Sprintf("%q", pattern) + ": " + err.Error())
 	}
 
-	matches, err := scanGlob(root, displayRoot, matcher, includeDirs, exclude)
+	matches, err := scanGlob(ctx, root, displayRoot, matcher, includeDirs, exclude)
 	if err != nil {
+		if res, ok := searchCancelledResult("glob", err); ok {
+			return res
+		}
 		return errorResult("Error running glob " + fmt.Sprintf("%q", pattern) + ": " + err.Error())
 	}
 	if len(matches) == 0 {
@@ -126,9 +129,15 @@ func (tool globTool) runWith(args map[string]any, exclude readExcluder) Result {
 	}
 }
 
-func scanGlob(root string, displayRoot string, matcher *regexp.Regexp, includeDirs bool, exclude readExcluder) ([]string, error) {
+func scanGlob(ctx context.Context, root string, displayRoot string, matcher *regexp.Regexp, includeDirs bool, exclude readExcluder) ([]string, error) {
 	matches := []string{}
 	err := filepath.WalkDir(root, func(path string, entry fs.DirEntry, walkErr error) error {
+		// Checked first, ahead of walkErr: an unscoped scan over a large tree can
+		// run long enough that cancelling the run must stop the walk promptly
+		// rather than visiting every remaining entry to completion.
+		if err := ctx.Err(); err != nil {
+			return err
+		}
 		if walkErr != nil {
 			if path == root {
 				return walkErr

internal/tools/grep.go

@@ -55,19 +55,19 @@ func NewScopedGrepTool(workspaceRoot string, scope PathScope) Tool {
 	}
 }
 
-func (tool grepTool) Run(_ context.Context, args map[string]any) Result {
-	return tool.runWith(args, readExcluder{})
+func (tool grepTool) Run(ctx context.Context, args map[string]any) Result {
+	return tool.runWith(ctx, args, readExcluder{})
 }
 
 // RunWithSandbox runs the search while skipping subtrees the sandbox policy
 // denies reads to (DenyRead), so grep never surfaces content from a read-denied
 // path. With no DenyRead configured the excluder is a no-op and behavior is
 // unchanged.
-func (tool grepTool) RunWithSandbox(_ context.Context, args map[string]any, engine *sandbox.Engine) Result {
-	return tool.runWith(args, sandboxReadExcluder(engine))
+func (tool grepTool) RunWithSandbox(ctx context.Context, args map[string]any, engine *sandbox.Engine) Result {
+	return tool.runWith(ctx, args, sandboxReadExcluder(engine))
 }
 
-func (tool grepTool) runWith(args map[string]any, exclude readExcluder) Result {
+func (tool grepTool) runWith(ctx context.Context, args map[string]any, exclude readExcluder) Result {
 	pattern, err := aliasedStringArg(args, []string{"pattern", "query", "regex", "search", "expression"}, "", true, false)
 	if err != nil {
 		return errorResult("Error: Invalid arguments for grep: " + err.Error())
@@ -138,15 +138,24 @@ func (tool grepTool) runWith(args map[string]any, exclude readExcluder) Result {
 		}
 	}
 
-	files, err := grepFiles(resolvedRoot, target, globMatcher, exclude)
+	files, err := grepFiles(ctx, resolvedRoot, target, globMatcher, exclude)
 	if err != nil {
+		if res, ok := searchCancelledResult("grep", err); ok {
+			return res
+		}
 		return errorResult("Error running grep: " + err.Error())
 	}
 
 	// resolveScopedPath returns an absolute displayRoot when the target resolved
 	// to an extra (non-workspace) granted root; emit absolute match paths there so
 	// they survive a round-trip through read_file/edit_file (mirrors glob).
-	matches := collectGrepMatches(resolvedRoot, filepath.IsAbs(displayRoot), files, compiled)
+	matches, err := collectGrepMatches(ctx, resolvedRoot, filepath.IsAbs(displayRoot), files, compiled)
+	if err != nil {
+		if res, ok := searchCancelledResult("grep", err); ok {
+			return res
+		}
+		return errorResult("Error running grep: " + err.Error())
+	}
 	if len(matches) == 0 {
 		if outputMode == "count" {
 			return okResult("0 matches found")
@@ -261,7 +270,7 @@ func confineGrepFile(resolvedRoot string, path string) (string, string, bool) {
 	return filepath.ToSlash(relative), resolved, true
 }
 
-func grepFiles(resolvedRoot string, target string, globMatcher *regexp.Regexp, exclude readExcluder) ([]string, error) {
+func grepFiles(ctx context.Context, resolvedRoot string, target string, globMatcher *regexp.Regexp, exclude readExcluder) ([]string, error) {
 	info, err := os.Stat(target)
 	if err != nil {
 		return nil, err
@@ -288,6 +297,13 @@ func grepFiles(resolvedRoot string, target string, globMatcher *regexp.Regexp, e
 
 	files := []string{}
 	err = filepath.WalkDir(target, func(path string, entry os.DirEntry, walkErr error) error {
+		// Checked first, ahead of walkErr: an unscoped search over a large tree
+		// (e.g. a broad parent directory, not just the workspace) can run long
+		// enough that cancelling the run must stop the walk promptly rather than
+		// visiting every remaining entry to completion.
+		if err := ctx.Err(); err != nil {
+			return err
+		}
 		if walkErr != nil {
 			if path == target {
 				return walkErr
@@ -342,9 +358,16 @@ func grepFiles(resolvedRoot string, target string, globMatcher *regexp.Regexp, e
 	return files, nil
 }
 
-func collectGrepMatches(resolvedRoot string, absolutePaths bool, files []string, compiled *regexp.Regexp) []grepMatch {
+func collectGrepMatches(ctx context.Context, resolvedRoot string, absolutePaths bool, files []string, compiled *regexp.Regexp) ([]grepMatch, error) {
 	matches := []grepMatch{}
 	for _, file := range files {
+		// Checked per file, matching the walk's per-entry check in grepFiles: a
+		// broad search can already have collected many files before the run is
+		// cancelled, and reading/regex-scanning each one is real work that must
+		// stop promptly rather than running to completion.
+		if err := ctx.Err(); err != nil {
+			return matches, err
+		}
 		// Re-confine at read time (defense-in-depth) AND to compute the clean
 		// workspace-relative path used in output.
 		relative, resolvedPath, ok := confineGrepFile(resolvedRoot, file)
@@ -381,5 +404,5 @@ func collectGrepMatches(resolvedRoot string, absolutePaths bool, files []string,
 			})
 		}
 	}
-	return matches
+	return matches, nil
 }

internal/tools/inline_diagnostics.go

@@ -0,0 +1,18 @@
+package tools
+
+import "context"
+
+// inlineDiagnostics renders the post-write diagnostics block a mutating tool
+// appends to its output. Empty when no Diagnostics callback is wired, the file
+// is clean, or no language server is available — the tool output is then
+// byte-identical to the pre-diagnostics behavior.
+func inlineDiagnostics(ctx context.Context, options RunOptions, absolutePath, relativePath string) string {
+	if options.Diagnostics == nil {
+		return ""
+	}
+	block := options.Diagnostics(ctx, absolutePath)
+	if block == "" {
+		return ""
+	}
+	return "\n\nDiagnostics in " + relativePath + " after this change (fix any errors you introduced):\n" + block
+}

internal/tools/inline_diagnostics_test.go

@@ -0,0 +1,53 @@
+package tools
+
+import (
+	"context"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+// A wired Diagnostics callback must surface its block in the tool output for
+// both mutating tools, and a clean file (empty block) must leave the output
+// byte-identical to the pre-diagnostics behavior.
+func TestMutatingToolsAppendInlineDiagnostics(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(dir, "a.go"), []byte("package a\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	diagnostics := func(_ context.Context, absPath string) string {
+		return absPath + ":1:1: error: undefined: x"
+	}
+
+	editResult := NewEditFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":       "a.go",
+		"old_string": "package a",
+		"new_string": "package b",
+	}, RunOptions{Diagnostics: diagnostics})
+	if editResult.Status != StatusOK {
+		t.Fatalf("edit failed: %q", editResult.Output)
+	}
+	if !strings.Contains(editResult.Output, "Diagnostics in a.go after this change") || !strings.Contains(editResult.Output, "undefined: x") {
+		t.Fatalf("edit output missing diagnostics block: %q", editResult.Output)
+	}
+
+	writeResult := NewWriteFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":    "b.go",
+		"content": "package b\n",
+	}, RunOptions{Diagnostics: diagnostics})
+	if writeResult.Status != StatusOK {
+		t.Fatalf("write failed: %q", writeResult.Output)
+	}
+	if !strings.Contains(writeResult.Output, "Diagnostics in b.go after this change") {
+		t.Fatalf("write output missing diagnostics block: %q", writeResult.Output)
+	}
+
+	clean := NewWriteFileTool(dir).(optionsAwareTool).RunWithOptions(context.Background(), map[string]any{
+		"path":    "c.go",
+		"content": "package c\n",
+	}, RunOptions{Diagnostics: func(context.Context, string) string { return "" }})
+	if strings.Contains(clean.Output, "Diagnostics") {
+		t.Fatalf("clean file must not gain a diagnostics block: %q", clean.Output)
+	}
+}

internal/tools/registry.go

@@ -40,6 +40,12 @@ type RunOptions struct {
 	// tool-call progress in the specialist card. nil is a no-op (the default
 	// for every non-Task tool).
 	Progress func(streamjson.Event)
+	// Diagnostics, when set, returns a formatted language-diagnostics block for
+	// a file a mutating tool just wrote ("" when clean or no server available).
+	// edit_file/write_file append it to their output so the model sees an error
+	// it introduced in the same turn instead of waiting for a later verification
+	// pass. nil disables inline diagnostics.
+	Diagnostics func(ctx context.Context, absPath string) string
 }
 
 type sandboxAwareTool interface {