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.

internal/agent/compaction.go

@@ -325,6 +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 {
@@ -353,7 +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
 	}
@@ -370,7 +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
@@ -386,7 +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.
@@ -444,7 +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)
 			}
@@ -1051,6 +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,
@@ -2462,6 +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))
@@ -2505,7 +2562,7 @@ func partitionTools(registry *tools.Registry, permissionMode PermissionMode, opt
 			if tool.Name() == tools.ToolSearchToolName {
 				continue
 			}
-			definitions = append(definitions, runtimeToolDefinition(tool))
+			definitions = append(definitions, cachedRuntimeToolDefinition(defCache, tool))
 		}
 		sort.Slice(definitions, func(left int, right int) bool {
 			return definitions[left].Name < definitions[right].Name
@@ -2537,13 +2594,13 @@ func partitionTools(registry *tools.Registry, permissionMode PermissionMode, opt
 		}
 		if tools.IsDeferred(tool) {
 			if loaded[name] {
-				loadedTail = append(loadedTail, runtimeToolDefinition(tool))
+				loadedTail = append(loadedTail, cachedRuntimeToolDefinition(defCache, tool))
 			} else {
 				hiddenTools = append(hiddenTools, tool)
 			}
 			continue
 		}
-		eager = append(eager, runtimeToolDefinition(tool))
+		eager = append(eager, cachedRuntimeToolDefinition(defCache, tool))
 	}
 	sort.Slice(eager, func(left int, right int) bool {
 		return eager[left].Name < eager[right].Name
@@ -2577,6 +2634,24 @@ func partitionTools(registry *tools.Registry, permissionMode PermissionMode, opt
 	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 {

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_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/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/types.go

@@ -285,6 +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"
@@ -229,6 +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)
@@ -544,6 +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)

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
@@ -499,9 +505,10 @@ 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,
@@ -525,6 +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
@@ -700,23 +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 {
@@ -949,6 +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/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/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,
@@ -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/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/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"
@@ -123,10 +122,14 @@ func (tool bashTool) run(ctx context.Context, args map[string]any, engine *zeroS
 	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().
@@ -138,7 +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{
@@ -155,26 +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,
 	}
 }
 
@@ -357,6 +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_tool_test.go

@@ -8,6 +8,7 @@ import (
 	"os"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"strings"
 	"testing"
 	"time"
@@ -252,6 +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")

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/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 {

internal/tools/spill.go

@@ -0,0 +1,141 @@
+package tools
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/Gitlawb/zero/internal/redaction"
+)
+
+// Spill-to-disk for truncated tool output. When a command produces more than
+// its context budget, the tail was previously simply gone — the model's only
+// recourse was re-running the (possibly expensive or non-idempotent) command
+// with a bigger budget. Instead, the full output is written to a cache file
+// whose path is included in the truncation notice, so the model can grep or
+// read_file the remainder. Spilling is best-effort: on any error truncation
+// behaves exactly as before, just without the file hint.
+
+// spillRetention is how long spilled outputs are kept; files older than this
+// are opportunistically removed whenever a new spill happens.
+const spillRetention = 7 * 24 * time.Hour
+
+// spillRootPath returns the spill directory path without creating or checking
+// it. Used by the read-path resolver to recognize spill files as readable.
+func spillRootPath() string {
+	name := "zero-tool-output"
+	if uid := os.Getuid(); uid >= 0 {
+		name = fmt.Sprintf("zero-tool-output-%d", uid)
+	}
+	return filepath.Join(os.TempDir(), name)
+}
+
+// resolveSpillReadPath reports whether requestedPath is a file inside the
+// spill directory and, if so, returns its verified absolute path. Spill files
+// are zero-created, per-uid owned, and already secret-redacted, so letting the
+// scoped read tools open them is what makes the truncation notice's
+// "read_file it" recovery actually work. Symlinks are resolved and the result
+// must still be inside the spill dir, so a planted link cannot smuggle an
+// out-of-scope file through this gate.
+func resolveSpillReadPath(requestedPath string) (string, bool) {
+	if requestedPath == "" || !filepath.IsAbs(requestedPath) {
+		return "", false
+	}
+	root := spillRootPath()
+	cleaned := filepath.Clean(requestedPath)
+	if cleaned == root || !strings.HasPrefix(cleaned, root+string(filepath.Separator)) {
+		return "", false
+	}
+	resolved, err := filepath.EvalSymlinks(cleaned)
+	if err != nil {
+		return "", false
+	}
+	resolvedRoot, err := filepath.EvalSymlinks(root)
+	if err != nil {
+		return "", false
+	}
+	if !strings.HasPrefix(resolved, resolvedRoot+string(filepath.Separator)) {
+		return "", false
+	}
+	return resolved, true
+}
+
+// spillDir returns the per-user spill directory, creating it on first use.
+// Hardened for shared temp dirs (Linux /tmp): the name carries the uid so
+// users cannot collide, and a pre-existing path is only accepted when it is a
+// real directory (not a symlink that would redirect spills) owned by the
+// current user. Any doubt fails the spill — it is best-effort anyway.
+func spillDir() (string, error) {
+	dir := spillRootPath()
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return "", err
+	}
+	// MkdirAll follows symlinks and leaves an existing directory untouched, so
+	// verify what is actually at the path.
+	info, err := os.Lstat(dir)
+	if err != nil {
+		return "", err
+	}
+	if !info.IsDir() {
+		return "", fmt.Errorf("spill path %s is not a directory", dir)
+	}
+	if err := checkSpillDirOwner(info); err != nil {
+		return "", err
+	}
+	return dir, nil
+}
+
+// spillTruncatedOutput writes the full pre-truncation output to the spill
+// directory and returns the file path, or "" when spilling fails. Output is
+// scrubbed with the same configured-key redaction the registry applies at the
+// tool boundary, so a spilled file never holds a secret the transcript would
+// have hidden.
+func spillTruncatedOutput(toolName, output string) string {
+	dir, err := spillDir()
+	if err != nil {
+		return ""
+	}
+	sweepSpillDir(dir)
+	prefix := strings.Map(func(r rune) rune {
+		if r >= 'a' && r <= 'z' || r >= 'A' && r <= 'Z' || r >= '0' && r <= '9' || r == '_' || r == '-' {
+			return r
+		}
+		return '-'
+	}, toolName)
+	file, err := os.CreateTemp(dir, prefix+"-*.txt")
+	if err != nil {
+		return ""
+	}
+	defer file.Close()
+	scrubbed := redaction.RedactString(output, redaction.Options{})
+	if _, err := file.WriteString(scrubbed); err != nil {
+		_ = os.Remove(file.Name())
+		return ""
+	}
+	return file.Name()
+}
+
+// sweepSpillDir removes spill files older than spillRetention. Best-effort:
+// errors are ignored, the directory is small, and a sweep runs only when a new
+// spill is about to happen anyway.
+func sweepSpillDir(dir string) {
+	entries, err := os.ReadDir(dir)
+	if err != nil {
+		return
+	}
+	cutoff := time.Now().Add(-spillRetention)
+	for _, entry := range entries {
+		if entry.IsDir() {
+			continue
+		}
+		info, err := entry.Info()
+		if err != nil {
+			continue
+		}
+		if info.ModTime().Before(cutoff) {
+			_ = os.Remove(filepath.Join(dir, entry.Name()))
+		}
+	}
+}

internal/tools/spill_hardening_test.go

@@ -0,0 +1,91 @@
+package tools
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+	"strings"
+	"testing"
+)
+
+func spillDirName() string {
+	if uid := os.Getuid(); uid >= 0 {
+		return fmt.Sprintf("zero-tool-output-%d", uid)
+	}
+	return "zero-tool-output"
+}
+
+func TestSpillDirRejectsSymlink(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("symlink creation needs privileges on Windows")
+	}
+	tmp := t.TempDir()
+	t.Setenv("TMPDIR", tmp)
+	elsewhere := filepath.Join(tmp, "elsewhere")
+	if err := os.Mkdir(elsewhere, 0o700); err != nil {
+		t.Fatal(err)
+	}
+	// An attacker pre-creates the spill path as a symlink to redirect spills.
+	if err := os.Symlink(elsewhere, filepath.Join(tmp, spillDirName())); err != nil {
+		t.Fatal(err)
+	}
+	if path := spillTruncatedOutput("bash", "sensitive output"); path != "" {
+		t.Fatalf("spill must refuse a symlinked directory, wrote %s", path)
+	}
+	entries, _ := os.ReadDir(elsewhere)
+	if len(entries) != 0 {
+		t.Fatalf("nothing may land behind the symlink: %v", entries)
+	}
+}
+
+func TestSpillDirAcceptsOwnedDirectory(t *testing.T) {
+	t.Setenv("TMPDIR", t.TempDir())
+	if path := spillTruncatedOutput("bash", "ok"); path == "" {
+		t.Fatal("spill must work in a clean per-user temp dir")
+	}
+}
+
+func TestResolveSpillReadPathContainment(t *testing.T) {
+	t.Setenv("TMPDIR", t.TempDir())
+	spillPath := spillTruncatedOutput("exec_command", "spilled body")
+	if spillPath == "" {
+		t.Fatal("spill must succeed in a clean temp dir")
+	}
+	resolved, ok := resolveSpillReadPath(spillPath)
+	if !ok || resolved == "" {
+		t.Fatalf("a real spill file must resolve, got ok=%v", ok)
+	}
+	if _, ok := resolveSpillReadPath(filepath.Join(spillRootPath(), "..", "escape.txt")); ok {
+		t.Fatal("path traversal out of the spill dir must be rejected")
+	}
+	if _, ok := resolveSpillReadPath("/etc/hosts"); ok {
+		t.Fatal("paths outside the spill dir must be rejected")
+	}
+	if _, ok := resolveSpillReadPath(spillRootPath()); ok {
+		t.Fatal("the spill dir itself is not a readable file target")
+	}
+	if runtime.GOOS != "windows" {
+		// A symlink planted inside the spill dir pointing outside must not leak.
+		link := filepath.Join(spillRootPath(), "sneaky-link")
+		if err := os.Symlink("/etc/hosts", link); err == nil {
+			if _, ok := resolveSpillReadPath(link); ok {
+				t.Fatal("symlink escaping the spill dir must be rejected")
+			}
+		}
+	}
+}
+
+func TestReadFileCanReadSpillFile(t *testing.T) {
+	t.Setenv("TMPDIR", t.TempDir())
+	spillPath := spillTruncatedOutput("exec_command", "line one\nline two\n")
+	if spillPath == "" {
+		t.Fatal("spill must succeed")
+	}
+	workspace := t.TempDir()
+	result := NewReadFileTool(workspace).Run(context.Background(), map[string]any{"path": spillPath})
+	if result.Status != StatusOK || !strings.Contains(result.Output, "line two") {
+		t.Fatalf("read_file must be able to follow the truncation notice: %s %q", result.Status, result.Output)
+	}
+}

internal/tools/spill_owner_unix.go

@@ -0,0 +1,24 @@
+//go:build !windows
+
+package tools
+
+import (
+	"fmt"
+	"os"
+	"syscall"
+)
+
+// checkSpillDirOwner rejects a spill directory not owned by the current user:
+// on a shared /tmp another user could have pre-created the path and would then
+// control its lifetime (deletion/renaming) even though the 0600 spill files
+// keep their contents private.
+func checkSpillDirOwner(info os.FileInfo) error {
+	stat, ok := info.Sys().(*syscall.Stat_t)
+	if !ok {
+		return nil
+	}
+	if int(stat.Uid) != os.Geteuid() {
+		return fmt.Errorf("spill directory is owned by uid %d, not the current user", stat.Uid)
+	}
+	return nil
+}

internal/tools/spill_owner_windows.go

@@ -0,0 +1,12 @@
+//go:build windows
+
+package tools
+
+import "os"
+
+// checkSpillDirOwner is a no-op on Windows: %TEMP% is per-user by default and
+// there is no portable uid to compare. The Lstat symlink/dir check in spillDir
+// still applies.
+func checkSpillDirOwner(os.FileInfo) error {
+	return nil
+}

internal/tools/spill_test.go

@@ -0,0 +1,82 @@
+package tools
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestTruncateExecOutputSpillWritesFullOutput(t *testing.T) {
+	t.Setenv("TMPDIR", t.TempDir())
+	long := strings.Repeat("line of build output\n", 5000) // ~100KB > 40KB budget
+
+	truncated, wasTruncated := truncateExecOutputSpill(long, defaultMaxOutputTokens, "bash")
+	if !wasTruncated {
+		t.Fatal("output over budget must truncate")
+	}
+	if !strings.Contains(truncated, "full output saved to ") {
+		t.Fatalf("truncation notice missing spill path: %q", truncated[:200])
+	}
+	// Extract the path and verify the file holds the complete output.
+	start := strings.Index(truncated, "full output saved to ") + len("full output saved to ")
+	end := strings.Index(truncated[start:], " (grep")
+	spillPath := truncated[start : start+end]
+	content, err := os.ReadFile(spillPath)
+	if err != nil {
+		t.Fatalf("spill file unreadable: %v", err)
+	}
+	if string(content) != long {
+		t.Fatalf("spill file must hold the full output: got %d bytes, want %d", len(content), len(long))
+	}
+}
+
+func TestTruncateExecOutputUnderBudgetUnchanged(t *testing.T) {
+	output, wasTruncated := truncateExecOutputSpill("short output", defaultMaxOutputTokens, "bash")
+	if wasTruncated || output != "short output" {
+		t.Fatalf("under-budget output must pass through untouched: %q", output)
+	}
+}
+
+func TestSweepSpillDirRemovesOnlyOldFiles(t *testing.T) {
+	dir := t.TempDir()
+	oldFile := filepath.Join(dir, "bash-old.txt")
+	newFile := filepath.Join(dir, "bash-new.txt")
+	for _, path := range []string{oldFile, newFile} {
+		if err := os.WriteFile(path, []byte("x"), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	stale := time.Now().Add(-spillRetention - time.Hour)
+	if err := os.Chtimes(oldFile, stale, stale); err != nil {
+		t.Fatal(err)
+	}
+
+	sweepSpillDir(dir)
+
+	if _, err := os.Stat(oldFile); !os.IsNotExist(err) {
+		t.Fatal("stale spill file must be removed")
+	}
+	if _, err := os.Stat(newFile); err != nil {
+		t.Fatal("fresh spill file must survive the sweep")
+	}
+}
+
+func TestSpillTruncatedOutputWritesFile(t *testing.T) {
+	t.Setenv("TMPDIR", t.TempDir())
+	path := spillTruncatedOutput("exec_command", "some output body")
+	if path == "" {
+		t.Fatal("spill must return a file path")
+	}
+	if base := filepath.Base(path); !strings.HasPrefix(base, "exec_command-") {
+		t.Fatalf("spill file name must carry the tool prefix: %s", base)
+	}
+	content, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(content) != "some output body" {
+		t.Fatalf("unexpected spill content: %q", content)
+	}
+}

internal/tools/tool_search.go

@@ -261,10 +261,16 @@ func (tool toolSearchTool) rankByKeyword(query string, deferred []Tool) []Tool {
 	for index, candidate := range deferred {
 		name := strings.ToLower(candidate.Name())
 		desc := strings.ToLower(candidate.Description())
+		nameSquashed := squashSeparators(name)
 		score := 0
 		for _, keyword := range keywords {
-			if strings.Contains(name, keyword) {
+			switch {
+			case strings.Contains(name, keyword):
 				score += 2
+			case strings.Contains(nameSquashed, squashSeparators(keyword)):
+				// Separator-insensitive fallback: "webfetch" matches "web_fetch".
+				// Ranked below an exact substring so precise queries still win.
+				score++
 			}
 			if strings.Contains(desc, keyword) {
 				score++
@@ -290,6 +296,19 @@ func (tool toolSearchTool) rankByKeyword(query string, deferred []Tool) []Tool {
 	return matches
 }
 
+// squashSeparators drops separators so "web_fetch", "web-fetch", and "web fetch"
+// all normalize to "webfetch". It lets a query missing the separators still match
+// a tool name, the common way a model mistypes a tool ("webfetch").
+func squashSeparators(s string) string {
+	return strings.Map(func(r rune) rune {
+		switch r {
+		case '_', '-', ' ', '.':
+			return -1
+		}
+		return r
+	}, s)
+}
+
 // noMatchMessage reports that nothing loaded and names the available deferred
 // tools so the model can retry with a valid select: query.
 func (tool toolSearchTool) noMatchMessage(query string, deferred []Tool) string {

internal/tools/tool_search_test.go

@@ -184,6 +184,29 @@ func TestToolSearchKeywordRanksByNameThenDescription(t *testing.T) {
 	}
 }
 
+// A query missing the name's separators still matches: "webfetch" -> web_fetch.
+func TestToolSearchKeywordMatchesSeparatorInsensitive(t *testing.T) {
+	reg := NewRegistry()
+	reg.Register(searchFakeTool{
+		name:        "web_fetch",
+		description: "Fetch a URL.",
+		parameters:  Schema{Type: "object", AdditionalProperties: false},
+	})
+	reg.Register(searchFakeTool{
+		name:        "stock_quote",
+		description: "Get a stock price.",
+		parameters:  Schema{Type: "object", AdditionalProperties: false},
+	})
+	tool := NewToolSearchTool(reg).(optionsAwareTool)
+
+	result := tool.RunWithOptions(context.Background(),
+		map[string]any{"query": "webfetch"}, RunOptions{})
+
+	if got := result.Meta["load_tools"]; got != "web_fetch" {
+		t.Fatalf("load_tools = %q, want web_fetch (separator-insensitive match)", got)
+	}
+}
+
 func TestToolSearchKeywordExcludesNonMatches(t *testing.T) {
 	reg := newDeferredFixtureRegistry()
 	tool := NewToolSearchTool(reg).(optionsAwareTool)

internal/tools/web_fetch.go

@@ -18,9 +18,14 @@ import (
 )
 
 const (
-	defaultWebFetchMaxBytes = 64 * 1024
-	maxWebFetchMaxBytes     = 512 * 1024
-	webFetchTimeout         = 10 * time.Second
+	// Raw-body budgets. HTML responses are converted to markdown before they
+	// reach the model (see web_fetch_markdown.go), so a generous raw budget
+	// does not translate into a generous context cost — conversion typically
+	// shrinks a page by an order of magnitude. 64KiB of raw HTML often held
+	// nothing but a page's <head>, which starved research tasks.
+	defaultWebFetchMaxBytes = 256 * 1024
+	maxWebFetchMaxBytes     = 2 * 1024 * 1024
+	webFetchTimeout         = 30 * time.Second
 	webFetchRedirectLimit   = 5
 	webFetchErrorBodyLimit  = 4 * 1024
 	webFetchPublicOnlyHint  = "web_fetch only supports public remote HTTP/HTTPS URLs. For localhost or private network URLs, use bash with curl so sandbox network permission can apply."
@@ -115,11 +120,17 @@ func newWebFetchToolWithClientAndResolver(client *http.Client, resolver webFetch
 					},
 					"max_bytes": {
 						Type:        "integer",
-						Description: "Maximum response body bytes to return.",
+						Description: "Maximum raw response body bytes to download before conversion.",
 						Default:     defaultWebFetchMaxBytes,
 						Minimum:     intPtr(1),
 						Maximum:     intPtr(maxWebFetchMaxBytes),
 					},
+					"format": {
+						Type:        "string",
+						Description: "auto (default): HTML responses are converted to compact markdown, everything else is returned as-is. raw: never convert. markdown: force conversion.",
+						Enum:        []string{"auto", "raw", "markdown"},
+						Default:     "auto",
+					},
 				},
 				Required:             []string{"url"},
 				AdditionalProperties: false,
@@ -170,6 +181,18 @@ func (tool webFetchTool) run(ctx context.Context, args map[string]any) Result {
 	if err != nil {
 		return errorResult("Error: Invalid arguments for web_fetch: " + err.Error())
 	}
+	format, err := stringArg(args, "format", "auto", false)
+	if err != nil {
+		return errorResult("Error: Invalid arguments for web_fetch: " + err.Error())
+	}
+	format = strings.ToLower(strings.TrimSpace(format))
+	switch format {
+	case "", "auto":
+		format = "auto"
+	case "raw", "markdown":
+	default:
+		return errorResult(`Error: Invalid arguments for web_fetch: format must be "auto", "raw", or "markdown".`)
+	}
 	if err := validateWebFetchURLBeforePermission(rawURL); err != nil {
 		return errorResult("Error: Unsafe URL for web_fetch: " + err.Error())
 	}
@@ -224,14 +247,28 @@ func (tool webFetchTool) run(ctx context.Context, args map[string]any) Result {
 	}
 	body = redactWebFetchText(body)
 
-	output := strings.Join([]string{
+	// HTML responses are converted to compact markdown by default: raw HTML is
+	// mostly markup, so conversion typically shrinks the page by an order of
+	// magnitude and the model reads content instead of boilerplate. format=raw
+	// is the escape hatch for pages the converter mangles.
+	converted := false
+	if format == "markdown" || (format == "auto" && looksLikeHTML(contentType, body)) {
+		if markdown := htmlToMarkdown(body); markdown != "" {
+			body = markdown
+			converted = true
+		}
+	}
+
+	headers := []string{
 		"URL: " + finalURL,
 		"Status: " + webFetchStatusLine(response),
 		"Content-Type: " + firstNonEmptyString(contentType, "unknown"),
 		"Bytes: " + strconv.Itoa(len(body)),
-		"",
-		body,
-	}, "\n")
+	}
+	if converted {
+		headers = append(headers, "Converted: html -> markdown (pass format: \"raw\" for the original HTML)")
+	}
+	output := strings.Join(append(headers, "", body), "\n")
 
 	return Result{
 		Status:    StatusOK,
@@ -243,6 +280,7 @@ func (tool webFetchTool) run(ctx context.Context, args map[string]any) Result {
 			"content_type": contentType,
 			"bytes":        strconv.Itoa(len(body)),
 			"truncated":    strconv.FormatBool(truncated),
+			"converted":    strconv.FormatBool(converted),
 		},
 	}
 }

internal/tools/web_fetch_markdown.go

@@ -0,0 +1,92 @@
+package tools
+
+import (
+	"fmt"
+	"html"
+	"regexp"
+	"strings"
+)
+
+// HTML-to-markdown conversion for web_fetch. Raw HTML is mostly boilerplate —
+// on a typical page well under a tenth of the bytes are readable content — so
+// returning it raw wastes nearly the whole response budget on markup. The
+// converter is deliberately dependency-free (regex passes + stdlib entity
+// unescaping, no HTML parser): the output feeds a language model, not a
+// browser, so best-effort structure (headings, links, list markers) is enough,
+// and the format=raw escape hatch covers pages it mangles.
+
+var (
+	webFetchDropBlockRe = regexp.MustCompile(`(?is)<(script|style|head|noscript|svg|template|iframe)\b[^>]*>.*?</\s*(?:script|style|head|noscript|svg|template|iframe)\s*>`)
+	webFetchCommentRe   = regexp.MustCompile(`(?s)<!--.*?-->`)
+	webFetchAnchorRe    = regexp.MustCompile(`(?is)<a\b[^>]*?href\s*=\s*["']?([^"'\s>]+)["']?[^>]*>(.*?)</\s*a\s*>`)
+	webFetchHeadingRe   = regexp.MustCompile(`(?is)<h([1-6])[^>]*>(.*?)</\s*h[1-6]\s*>`)
+	webFetchListItemRe  = regexp.MustCompile(`(?i)<li\b[^>]*>`)
+	webFetchBreakRe     = regexp.MustCompile(`(?i)<(?:br|hr)\s*/?>`)
+	webFetchBlockTagRe  = regexp.MustCompile(`(?i)</?(?:p|div|section|article|main|header|footer|nav|aside|table|tr|ul|ol|blockquote|pre|figure|form)\b[^>]*>`)
+	webFetchAnyTagRe    = regexp.MustCompile(`(?s)<[^>]*>`)
+	webFetchSpaceRunRe  = regexp.MustCompile(`[ \t]+`)
+	webFetchBlankRunRe  = regexp.MustCompile(`\n{3,}`)
+)
+
+// htmlToMarkdown converts an HTML document to compact markdown-flavored text:
+// headings become #-prefixed lines, anchors become [text](href), list items
+// become "- " bullets, block boundaries become blank lines, all other markup
+// is stripped, and entities are unescaped.
+func htmlToMarkdown(body string) string {
+	text := webFetchCommentRe.ReplaceAllString(body, " ")
+	text = webFetchDropBlockRe.ReplaceAllString(text, " ")
+	text = webFetchHeadingRe.ReplaceAllStringFunc(text, func(match string) string {
+		groups := webFetchHeadingRe.FindStringSubmatch(match)
+		level := int(groups[1][0] - '0')
+		if level < 1 || level > 6 {
+			level = 1
+		}
+		title := strings.TrimSpace(webFetchAnyTagRe.ReplaceAllString(groups[2], " "))
+		return "\n\n" + strings.Repeat("#", level) + " " + title + "\n\n"
+	})
+	text = webFetchAnchorRe.ReplaceAllStringFunc(text, func(match string) string {
+		groups := webFetchAnchorRe.FindStringSubmatch(match)
+		href := groups[1]
+		label := strings.TrimSpace(webFetchAnyTagRe.ReplaceAllString(groups[2], " "))
+		if label == "" {
+			return " "
+		}
+		if strings.HasPrefix(href, "#") || strings.HasPrefix(strings.ToLower(href), "javascript:") {
+			return label
+		}
+		return fmt.Sprintf("[%s](%s)", label, href)
+	})
+	text = webFetchListItemRe.ReplaceAllString(text, "\n- ")
+	text = webFetchBreakRe.ReplaceAllString(text, "\n")
+	text = webFetchBlockTagRe.ReplaceAllString(text, "\n\n")
+	text = webFetchAnyTagRe.ReplaceAllString(text, " ")
+	text = html.UnescapeString(text)
+
+	// Whitespace normalization: collapse space runs, trim line edges, cap blank
+	// runs at one empty line.
+	lines := strings.Split(text, "\n")
+	for i, line := range lines {
+		lines[i] = strings.TrimSpace(webFetchSpaceRunRe.ReplaceAllString(line, " "))
+	}
+	text = strings.Join(lines, "\n")
+	text = webFetchBlankRunRe.ReplaceAllString(text, "\n\n")
+	return strings.TrimSpace(text)
+}
+
+// looksLikeHTML reports whether a response should be treated as an HTML
+// document, by Content-Type first and a body sniff as fallback (servers
+// mislabel HTML as text/plain often enough to matter).
+func looksLikeHTML(contentType, body string) bool {
+	lowered := strings.ToLower(contentType)
+	if strings.Contains(lowered, "text/html") || strings.Contains(lowered, "application/xhtml") {
+		return true
+	}
+	if lowered != "" && !strings.Contains(lowered, "text/plain") {
+		return false
+	}
+	head := strings.ToLower(strings.TrimSpace(body))
+	if len(head) > 512 {
+		head = head[:512]
+	}
+	return strings.Contains(head, "<!doctype html") || strings.Contains(head, "<html")
+}

internal/tools/web_fetch_markdown_test.go

@@ -0,0 +1,127 @@
+package tools
+
+import (
+	"context"
+	"net/http"
+	"strings"
+	"testing"
+)
+
+const sampleHTML = `<!DOCTYPE html>
+<html><head><title>ignored</title><style>body{color:red}</style></head>
+<body>
+<script>var tracking = "noise";</script>
+<h1>Main Title</h1>
+<p>First paragraph with a <a href="https://example.com/docs">docs link</a> and &amp; entity.</p>
+<ul><li>alpha</li><li>beta</li></ul>
+<!-- hidden comment -->
+<h2>Section</h2>
+<p>Body   with   extra   spaces.</p>
+</body></html>`
+
+func TestHTMLToMarkdown(t *testing.T) {
+	markdown := htmlToMarkdown(sampleHTML)
+
+	for _, want := range []string{
+		"# Main Title",
+		"## Section",
+		"[docs link](https://example.com/docs)",
+		"- alpha",
+		"- beta",
+		"with a", // paragraph text survives
+		"& entity",
+	} {
+		if !strings.Contains(markdown, want) {
+			t.Fatalf("markdown missing %q:\n%s", want, markdown)
+		}
+	}
+	for _, banned := range []string{"tracking", "color:red", "hidden comment", "<p>", "</html>", "ignored"} {
+		if strings.Contains(markdown, banned) {
+			t.Fatalf("markdown must not contain %q:\n%s", banned, markdown)
+		}
+	}
+	if strings.Contains(markdown, "extra   spaces") {
+		t.Fatalf("space runs must collapse:\n%s", markdown)
+	}
+	if strings.Contains(markdown, "\n\n\n") {
+		t.Fatalf("blank-line runs must be capped:\n%s", markdown)
+	}
+}
+
+func TestLooksLikeHTML(t *testing.T) {
+	cases := []struct {
+		contentType string
+		body        string
+		want        bool
+	}{
+		{"text/html; charset=utf-8", "anything", true},
+		{"application/xhtml+xml", "anything", true},
+		{"application/json", `{"html":"<html>"}`, false},
+		{"text/plain", "<!doctype html><html>", true}, // mislabeled HTML sniffed
+		{"text/plain", "just text", false},
+		{"", "<html><body>x</body></html>", true},
+	}
+	for _, c := range cases {
+		if got := looksLikeHTML(c.contentType, c.body); got != c.want {
+			t.Errorf("looksLikeHTML(%q, %.20q) = %v, want %v", c.contentType, c.body, got, c.want)
+		}
+	}
+}
+
+func TestWebFetchConvertsHTMLByDefault(t *testing.T) {
+	tool := newWebFetchToolWithClient(webFetchTestClient(func(request *http.Request) (*http.Response, error) {
+		return webFetchTestResponse(request, http.StatusOK, "text/html; charset=utf-8", sampleHTML), nil
+	}))
+
+	result := tool.Run(context.Background(), map[string]any{"url": "https://example.com/page"})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %s: %s", result.Status, result.Output)
+	}
+	if !strings.Contains(result.Output, "# Main Title") || strings.Contains(result.Output, "<h1>") {
+		t.Fatalf("HTML must convert to markdown by default:\n%s", result.Output)
+	}
+	if result.Meta["converted"] != "true" {
+		t.Fatalf("converted meta must be true: %#v", result.Meta)
+	}
+}
+
+func TestWebFetchRawFormatSkipsConversion(t *testing.T) {
+	tool := newWebFetchToolWithClient(webFetchTestClient(func(request *http.Request) (*http.Response, error) {
+		return webFetchTestResponse(request, http.StatusOK, "text/html", sampleHTML), nil
+	}))
+
+	result := tool.Run(context.Background(), map[string]any{"url": "https://example.com/page", "format": "raw"})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %s: %s", result.Status, result.Output)
+	}
+	if !strings.Contains(result.Output, "<h1>Main Title</h1>") {
+		t.Fatalf("format=raw must keep original HTML:\n%s", result.Output)
+	}
+	if result.Meta["converted"] != "false" {
+		t.Fatalf("converted meta must be false: %#v", result.Meta)
+	}
+}
+
+func TestWebFetchLeavesNonHTMLUntouched(t *testing.T) {
+	tool := newWebFetchToolWithClient(webFetchTestClient(func(request *http.Request) (*http.Response, error) {
+		return webFetchTestResponse(request, http.StatusOK, "application/json", `{"key":"<b>value</b>"}`), nil
+	}))
+
+	result := tool.Run(context.Background(), map[string]any{"url": "https://example.com/api"})
+	if result.Status != StatusOK {
+		t.Fatalf("expected ok, got %s: %s", result.Status, result.Output)
+	}
+	if !strings.Contains(result.Output, `{"key":"<b>value</b>"}`) {
+		t.Fatalf("non-HTML must pass through untouched:\n%s", result.Output)
+	}
+}
+
+func TestWebFetchRejectsInvalidFormat(t *testing.T) {
+	tool := newWebFetchToolWithClient(webFetchTestClient(func(request *http.Request) (*http.Response, error) {
+		return webFetchTestResponse(request, http.StatusOK, "text/plain", "x"), nil
+	}))
+	result := tool.Run(context.Background(), map[string]any{"url": "https://example.com", "format": "xml"})
+	if result.Status != StatusError || !strings.Contains(result.Output, "format must be") {
+		t.Fatalf("invalid format must be rejected: %s %s", result.Status, result.Output)
+	}
+}

internal/tools/workspace.go

@@ -254,6 +254,14 @@ func scopedReadRoots(workspaceRoot string, scope PathScope) ([]string, error) {
 }
 
 func resolveScopedReadPath(workspaceRoot string, scope PathScope, requestedPath string) (string, string, error) {
+	// Spill files (truncated tool output saved under the per-uid temp dir) are
+	// readable regardless of scope: the truncation notice tells the model to
+	// read_file/grep them, which must actually work. resolveSpillReadPath
+	// verifies containment after symlink resolution, so this cannot be used to
+	// reach anything outside the spill dir.
+	if spillPath, ok := resolveSpillReadPath(requestedPath); ok {
+		return spillPath, spillPath, nil
+	}
 	if requestedPath == "" || !filepath.IsAbs(requestedPath) || scope == nil {
 		return resolveWorkspacePath(workspaceRoot, requestedPath)
 	}

internal/tools/write_file.go

@@ -44,7 +44,7 @@ func (tool writeFileTool) Run(ctx context.Context, args map[string]any) Result {
 	return tool.RunWithOptions(ctx, args, RunOptions{})
 }
 
-func (tool writeFileTool) RunWithOptions(_ context.Context, args map[string]any, options RunOptions) Result {
+func (tool writeFileTool) 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 write_file: " + err.Error())
@@ -110,6 +110,10 @@ func (tool writeFileTool) RunWithOptions(_ context.Context, args map[string]any,
 	if err := os.WriteFile(absolutePath, []byte(content), 0o644); err != nil {
 		return errorResult("Error writing file " + relativePath + ": " + err.Error())
 	}
+	// Optional format-on-write (ZERO_FORMAT_ON_WRITE). Must run BEFORE the
+	// FileTracker baseline: recording pre-format content would make the very
+	// next edit look like an external modification and trip the conflict guard.
+	content = maybeFormatWrittenFile(ctx, absolutePath, content)
 	// Baseline the freshly written content so a later edit/overwrite in this
 	// session compares against what is now on disk.
 	newInfo, _ := os.Stat(absolutePath)
@@ -126,6 +130,7 @@ func (tool writeFileTool) RunWithOptions(_ context.Context, args map[string]any,
 		lines++
 	}
 	summary := fmt.Sprintf("%s %s (%d lines).", verb, relativePath, lines)
+	summary += inlineDiagnostics(ctx, options, absolutePath, relativePath)
 	result := okResult(summary)
 	result.ChangedFiles = []string{relativePath}
 	// Card-only preview: a real unified diff (all-green for a create, red/green for

internal/tui/coalesce.go

@@ -0,0 +1,109 @@
+package tui
+
+import (
+	"sync"
+	"time"
+
+	tea "charm.land/bubbletea/v2"
+)
+
+// streamCoalesceInterval is roughly one 60fps frame. Assistant-text deltas that
+// arrive within this window are merged into a single agentTextMsg, so the render
+// rate decouples from the token rate: a fast local model (100+ tok/s) no longer
+// forces 100+ full Update→View cycles (each re-parsing the growing markdown) per
+// second. Rendering stays smooth regardless of provider speed.
+const streamCoalesceInterval = 16 * time.Millisecond
+
+// textCoalescer batches agentTextMsg deltas before forwarding them to the Bubble
+// Tea program. Any OTHER message flushes the pending text first, so ordering
+// between streamed prose and tool-call / reasoning / row / usage messages is
+// preserved. The turn's final agentResponseMsg does not pass through here (it is
+// a tea.Cmd return, not a sink message), but the model drops deltas whose runID
+// is no longer active, so a flush that races just past end-of-turn is harmless.
+//
+// Sink messages originate from the single agent goroutine and so arrive
+// serially; the only concurrent caller is the flush timer. The mutex guards the
+// buffer/timer AND is held across the downstream forward, so a timer-fired text
+// flush can never overtake a concurrent non-text message: whoever holds the lock
+// drains and forwards atomically, and the other caller blocks until it is done.
+type textCoalescer struct {
+	forward func(tea.Msg) // downstream sink (external sink + program.Send)
+	// afterFunc schedules fn to run after one frame interval and returns a
+	// stoppable timer. Defaults to a real time.AfterFunc(streamCoalesceInterval, …);
+	// tests swap in a controllable timer so flush timing is deterministic instead of
+	// racing the 16ms wall clock.
+	afterFunc func(fn func()) coalesceTimer
+
+	mu    sync.Mutex
+	buf   []byte
+	runID int
+	timer coalesceTimer
+}
+
+// coalesceTimer is the subset of *time.Timer the coalescer needs. Abstracted
+// behind afterFunc so a test can substitute a timer it controls.
+type coalesceTimer interface {
+	Stop() bool
+}
+
+func newTextCoalescer(forward func(tea.Msg)) *textCoalescer {
+	return &textCoalescer{
+		forward: forward,
+		afterFunc: func(fn func()) coalesceTimer {
+			return time.AfterFunc(streamCoalesceInterval, fn)
+		},
+	}
+}
+
+// send is the coalescing entry point installed as the RuntimeMessageSink.
+func (c *textCoalescer) send(msg tea.Msg) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	text, ok := msg.(agentTextMsg)
+	if !ok {
+		// Non-text message: flush buffered text first (preserving order), then
+		// forward it — both under the lock so nothing can interleave between them.
+		c.drainAndForwardLocked()
+		c.forward(msg)
+		return
+	}
+
+	// A delta for a different run than the one buffered: flush the old run's text
+	// before buffering the new run's. In practice runs are sequential (the prior
+	// run's end already flushed via a non-text message), so this is belt-and-braces.
+	if len(c.buf) > 0 && text.runID != c.runID {
+		c.drainAndForwardLocked()
+	}
+	c.runID = text.runID
+	c.buf = append(c.buf, text.delta...)
+	if c.timer == nil {
+		c.timer = c.afterFunc(c.flush)
+	}
+}
+
+// flush forwards any buffered text as one agentTextMsg. Runs on the timer
+// goroutine; the lock it takes serializes it against send so its output can't be
+// reordered around a concurrent non-text message.
+func (c *textCoalescer) flush() {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	c.drainAndForwardLocked()
+}
+
+// drainAndForwardLocked forwards any buffered text as one agentTextMsg and stops
+// the timer, all while the caller holds c.mu — so a text flush and any non-text
+// forward are strictly ordered and never interleave. A no-op when nothing is
+// buffered. string(c.buf) copies, so reusing the backing array via [:0] is safe.
+func (c *textCoalescer) drainAndForwardLocked() {
+	if len(c.buf) == 0 {
+		return
+	}
+	if c.timer != nil {
+		c.timer.Stop()
+		c.timer = nil
+	}
+	msg := agentTextMsg{runID: c.runID, delta: string(c.buf)}
+	c.buf = c.buf[:0]
+	c.forward(msg)
+}

internal/tui/coalesce_test.go

@@ -0,0 +1,182 @@
+package tui
+
+import (
+	"sync"
+	"testing"
+	"time"
+
+	tea "charm.land/bubbletea/v2"
+)
+
+// recorder is a thread-safe sink that captures forwarded messages in order.
+type recorder struct {
+	mu   sync.Mutex
+	msgs []tea.Msg
+}
+
+func (r *recorder) forward(msg tea.Msg) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	r.msgs = append(r.msgs, msg)
+}
+
+func (r *recorder) snapshot() []tea.Msg {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	out := make([]tea.Msg, len(r.msgs))
+	copy(out, r.msgs)
+	return out
+}
+
+// manualTimer is a coalesceTimer that never fires on its own, so a test can prove
+// buffering deterministically without racing the real 16ms frame timer: the flush
+// happens only when the test calls c.flush() explicitly.
+type manualTimer struct{ stopped bool }
+
+func (m *manualTimer) Stop() bool {
+	prev := m.stopped
+	m.stopped = true
+	return !prev
+}
+
+// Rapid deltas within one frame collapse into a single agentTextMsg carrying the
+// concatenated text.
+func TestCoalescerBatchesDeltas(t *testing.T) {
+	rec := &recorder{}
+	c := newTextCoalescer(rec.forward)
+	// Swap in a timer that never auto-fires so the "still buffered" assertion below
+	// cannot flake if the scheduler pauses this goroutine past the frame interval.
+	c.afterFunc = func(func()) coalesceTimer { return &manualTimer{} }
+
+	c.send(agentTextMsg{runID: 1, delta: "Hel"})
+	c.send(agentTextMsg{runID: 1, delta: "lo, "})
+	c.send(agentTextMsg{runID: 1, delta: "world"})
+
+	// Nothing forwarded yet — still buffered within the frame (the timer we injected
+	// never fires; only the explicit flush below delivers the text).
+	if got := rec.snapshot(); len(got) != 0 {
+		t.Fatalf("deltas should buffer, forwarded %d early: %#v", len(got), got)
+	}
+
+	c.flush()
+
+	got := rec.snapshot()
+	if len(got) != 1 {
+		t.Fatalf("expected 1 coalesced message, got %d: %#v", len(got), got)
+	}
+	text, ok := got[0].(agentTextMsg)
+	if !ok || text.delta != "Hello, world" || text.runID != 1 {
+		t.Fatalf("coalesced message = %#v, want agentTextMsg{1, \"Hello, world\"}", got[0])
+	}
+}
+
+// A non-text message flushes buffered text first, so ordering (text before the
+// tool call it precedes) is preserved.
+func TestCoalescerFlushesTextBeforeOtherMessages(t *testing.T) {
+	rec := &recorder{}
+	c := newTextCoalescer(rec.forward)
+
+	c.send(agentTextMsg{runID: 1, delta: "about to run"})
+	c.send(toolCallStreamStartMsg{runID: 1, id: "t1", name: "bash"})
+
+	got := rec.snapshot()
+	if len(got) != 2 {
+		t.Fatalf("expected text then tool-call, got %d: %#v", len(got), got)
+	}
+	if text, ok := got[0].(agentTextMsg); !ok || text.delta != "about to run" {
+		t.Fatalf("first forwarded message must be the flushed text, got %#v", got[0])
+	}
+	if _, ok := got[1].(toolCallStreamStartMsg); !ok {
+		t.Fatalf("second forwarded message must be the tool-call start, got %#v", got[1])
+	}
+}
+
+// A delta for a new run flushes the previous run's buffered text before buffering
+// the new run's, so text is never mis-attributed across runs.
+func TestCoalescerFlushesOnRunSwitch(t *testing.T) {
+	rec := &recorder{}
+	c := newTextCoalescer(rec.forward)
+
+	c.send(agentTextMsg{runID: 1, delta: "run one text"})
+	c.send(agentTextMsg{runID: 2, delta: "run two text"})
+	c.flush()
+
+	got := rec.snapshot()
+	if len(got) != 2 {
+		t.Fatalf("expected one message per run, got %d: %#v", len(got), got)
+	}
+	if first := got[0].(agentTextMsg); first.runID != 1 || first.delta != "run one text" {
+		t.Fatalf("first message = %#v, want run 1 text flushed on switch", got[0])
+	}
+	if second := got[1].(agentTextMsg); second.runID != 2 || second.delta != "run two text" {
+		t.Fatalf("second message = %#v, want run 2 text", got[1])
+	}
+}
+
+// Under concurrency (the frame timer firing while a non-text message is sent),
+// forwarding must stay serialized and ordered: buffered text is always delivered
+// before a following non-text message, never after it. Run with -race to catch
+// the interleaving CodeRabbit flagged.
+func TestCoalescerOrdersTextBeforeNonTextUnderConcurrency(t *testing.T) {
+	for iter := 0; iter < 200; iter++ {
+		rec := &recorder{}
+		c := newTextCoalescer(rec.forward)
+
+		// Buffer text, then let the timer race against an inline non-text send.
+		c.send(agentTextMsg{runID: 1, delta: "before-tool"})
+		done := make(chan struct{})
+		go func() {
+			c.send(toolCallStreamStartMsg{runID: 1, id: "t1", name: "bash"})
+			close(done)
+		}()
+		<-done
+
+		// Give the timer a chance to also fire, then settle.
+		time.Sleep(2 * streamCoalesceInterval)
+		c.flush()
+
+		got := rec.snapshot()
+		textIdx, toolIdx := -1, -1
+		for i, m := range got {
+			switch m.(type) {
+			case agentTextMsg:
+				if textIdx == -1 {
+					textIdx = i
+				}
+			case toolCallStreamStartMsg:
+				toolIdx = i
+			}
+		}
+		if textIdx == -1 || toolIdx == -1 {
+			t.Fatalf("iter %d: missing messages: %#v", iter, got)
+		}
+		if textIdx > toolIdx {
+			t.Fatalf("iter %d: text (%d) forwarded after tool-call (%d): %#v", iter, textIdx, toolIdx, got)
+		}
+	}
+}
+
+// The frame timer flushes buffered text on its own without an explicit flush or a
+// following message.
+func TestCoalescerTimerFlushes(t *testing.T) {
+	rec := &recorder{}
+	c := newTextCoalescer(rec.forward)
+
+	c.send(agentTextMsg{runID: 1, delta: "timer-driven"})
+
+	deadline := time.Now().Add(time.Second)
+	for time.Now().Before(deadline) {
+		if len(rec.snapshot()) > 0 {
+			break
+		}
+		time.Sleep(streamCoalesceInterval)
+	}
+
+	got := rec.snapshot()
+	if len(got) != 1 {
+		t.Fatalf("timer should have flushed exactly one message, got %d: %#v", len(got), got)
+	}
+	if text := got[0].(agentTextMsg); text.delta != "timer-driven" {
+		t.Fatalf("timer-flushed message = %#v", got[0])
+	}
+}

internal/tui/command_center.go

@@ -482,6 +482,9 @@ func (m model) handleModelCommand(args string) (model, string) {
 		lines = append(lines, target.notice)
 	}
 	lines = append(lines, summary)
+	if warn := m.visionDropWarning(); warn != "" {
+		lines = append(lines, warn)
+	}
 	return m, strings.Join(lines, "\n")
 }
 
@@ -536,7 +539,11 @@ func (m model) switchProviderModel(providerName, modelID string) (model, string,
 			cmds = append(cmds, cmd)
 		}
 	}
-	return m, fmt.Sprintf("Model\nSwitched to %s · %s", target.Name, target.Model), tea.Batch(cmds...)
+	status := fmt.Sprintf("Model\nSwitched to %s · %s", target.Name, target.Model)
+	if warn := m.visionDropWarning(); warn != "" {
+		status += "\n" + warn
+	}
+	return m, status, tea.Batch(cmds...)
 }
 
 // profileWithCredential fills a profile's APIKey for provider construction the same

internal/tui/commands.go

@@ -41,6 +41,10 @@ const (
 	commandAddDir
 	commandSelfCorrect
 	commandTurns
+	commandRetry
+	commandEdit
+	commandCopy
+	commandExport
 	commandNew
 	commandUnknown
 )
@@ -209,9 +213,9 @@ var commandDefinitions = []commandDefinition{
 	},
 	{
 		name:        "/compact",
-		usage:       "/compact [status]",
+		usage:       "/compact [status|now]",
 		group:       commandGroupSession,
-		description: "Show or request transcript compaction state.",
+		description: "Compact the transcript now, or show compaction state (/compact status).",
 		kind:        commandCompact,
 	},
 	{
@@ -257,6 +261,34 @@ var commandDefinitions = []commandDefinition{
 		description: "Show or set the per-run tool-turn budget for this session (raise it for long multi-step tasks).",
 		kind:        commandTurns,
 	},
+	{
+		name:        "/retry",
+		usage:       "/retry",
+		group:       commandGroupSession,
+		description: "Resend your last prompt.",
+		kind:        commandRetry,
+	},
+	{
+		name:        "/edit",
+		usage:       "/edit",
+		group:       commandGroupSession,
+		description: "Recall your last prompt into the composer to edit and resend.",
+		kind:        commandEdit,
+	},
+	{
+		name:        "/copy",
+		usage:       "/copy",
+		group:       commandGroupSession,
+		description: "Copy the last answer to the clipboard.",
+		kind:        commandCopy,
+	},
+	{
+		name:        "/export",
+		usage:       "/export [path]",
+		group:       commandGroupSession,
+		description: "Write the conversation transcript to a file.",
+		kind:        commandExport,
+	},
 	{
 		name:        "/help",
 		usage:       "/help",

internal/tui/compact_now_test.go

@@ -0,0 +1,27 @@
+package tui
+
+import (
+	"strings"
+	"testing"
+)
+
+// "/compact now" is accepted as a manual trigger (not a usage error), while a
+// bogus argument still reports usage.
+func TestCompactNowAccepted(t *testing.T) {
+	m := model{}
+
+	_, nowText, _ := m.handleCompactCommand("now")
+	if strings.Contains(nowText, "usage:") {
+		t.Fatalf("/compact now must be accepted as a trigger, got usage error: %q", nowText)
+	}
+
+	_, statusText, _ := m.handleCompactCommand("status")
+	if strings.Contains(statusText, "usage:") {
+		t.Fatalf("/compact status must not error, got: %q", statusText)
+	}
+
+	_, bogusText, _ := m.handleCompactCommand("frobnicate")
+	if !strings.Contains(bogusText, "usage:") {
+		t.Fatalf("/compact frobnicate must report usage, got: %q", bogusText)
+	}
+}

internal/tui/error_hint_test.go

@@ -0,0 +1,27 @@
+package tui
+
+import (
+	"strings"
+	"testing"
+)
+
+// A hinted error row renders the raw error and the faint one-line next step
+// below it; an unhinted row renders just the error.
+func TestRenderErrorRowShowsHint(t *testing.T) {
+	withHint := renderErrorRow(transcriptRow{
+		kind: rowError,
+		text: "auth error: your API key is missing or invalid",
+		hint: "API key rejected — run /provider to re-check your credentials",
+	}, 80)
+	if !strings.Contains(withHint, "auth error:") {
+		t.Fatalf("expected raw error text in output, got:\n%s", withHint)
+	}
+	if !strings.Contains(withHint, "/provider") {
+		t.Fatalf("expected hint referencing /provider, got:\n%s", withHint)
+	}
+
+	noHint := renderErrorRow(transcriptRow{kind: rowError, text: "provider error: mystery"}, 80)
+	if strings.Contains(noHint, "→") {
+		t.Fatalf("unhinted error must not render a hint arrow, got:\n%s", noHint)
+	}
+}

internal/tui/flush_test.go

@@ -121,7 +121,7 @@ func TestEscCancellationLeavesVisibleMarker(t *testing.T) {
 	m.pending = true
 	m.activeRunID = 5
 	m.runCancel = func() {}
-	m.streamingText = "half an answer"
+	m.streamingText = []byte("half an answer")
 
 	updated, _ := m.Update(testKey(tea.KeyEsc))
 	next := updated.(model)
@@ -133,7 +133,7 @@ func TestEscCancellationLeavesVisibleMarker(t *testing.T) {
 	if !transcriptContains(next.transcript, "half an answer") {
 		t.Fatalf("expected the partial streamed answer to be preserved, got %#v", next.transcript)
 	}
-	if next.streamingText != "" {
+	if len(next.streamingText) != 0 {
 		t.Fatal("streaming text should be cleared after cancel")
 	}
 }

internal/tui/image_attach.go

@@ -292,6 +292,18 @@ func renderImageChips(labels []string) string {
 // and staged documents, e.g. "[Image #1] [Image #2] [Doc #1]". Returns "" when
 // nothing is staged. Numbered (not named) so a long screenshot path never shows
 // in the composer.
+// visionDropWarning returns a one-line notice when images are staged but the
+// (now active) model can't accept them, so switching to a non-vision model warns
+// the user immediately at switch time instead of silently dropping the images at
+// submit. Empty when there is nothing staged or the model supports vision.
+func (m model) visionDropWarning() string {
+	if len(m.pendingImages) == 0 || m.modelSupportsVisionTUI() {
+		return ""
+	}
+	return fmt.Sprintf("⚠ %d staged image(s) will be dropped — %s has no vision support.",
+		len(m.pendingImages), displayValue(m.modelName, "the active model"))
+}
+
 func renderAttachmentChips(imageLabels []string, docs []pendingDocument) string {
 	chips := make([]string, 0, len(imageLabels)+len(docs))
 	for i := range imageLabels {

internal/tui/image_attach_test.go

@@ -483,3 +483,76 @@ func TestRenderAttachmentChips(t *testing.T) {
 		t.Fatalf("chip row %q should not show file names", got)
 	}
 }
+
+// TestRetryResendsAttachments guards the /retry contract: it must resend the exact
+// request the last prompt carried — including its staged image and PDF document —
+// not a degraded text-only prompt. launchPrompt clears the pending queues once a
+// turn is sent, so without re-staging the remembered snapshot a vision/document-
+// backed turn that failed would silently retry as a different task.
+func TestRetryResendsAttachments(t *testing.T) {
+	root := t.TempDir()
+
+	captured := make(chan []zeroruntime.ImageBlock, 1)
+	provider := &fakeProvider{events: []zeroruntime.StreamEvent{
+		{Type: zeroruntime.StreamEventText, Content: "ok"},
+		{Type: zeroruntime.StreamEventDone},
+	}}
+	m := newModel(context.Background(), Options{
+		Cwd:          root,
+		ProviderName: "openai",
+		ModelName:    "gpt-4.1",
+		Provider:     provider,
+		Registry:     tools.NewRegistry(),
+		SessionStore: testSessionStore(t),
+	})
+	m.agentOptions.OnText = func(string) {}
+	m.captureRunImages = func(imgs []zeroruntime.ImageBlock) { captured <- imgs }
+
+	// State left after a prior vision+PDF prompt was submitted: the pending queues
+	// are cleared, but the remembered snapshot survives so /retry can reproduce it.
+	m.lastPrompt = "describe both"
+	m.lastImages = []zeroruntime.ImageBlock{{MediaType: "image/png", Data: []byte{0x89, 'P', 'N', 'G'}}}
+	m.lastImageLabels = []string{"photo.png"}
+	m.lastDocuments = []pendingDocument{{label: "spec.pdf", text: "Top secret design notes"}}
+
+	m.input.SetValue("/retry")
+	updated, cmd := m.handleSubmit()
+	next := updated.(model)
+	if cmd == nil {
+		t.Fatal("/retry with a remembered prompt should launch a run")
+	}
+	// The retried turn re-consumes and then clears the queues, exactly like a fresh
+	// submit; the snapshot must survive so a second /retry stays reproducible.
+	if len(next.pendingImages) != 0 || len(next.pendingImageLabels) != 0 || len(next.pendingDocuments) != 0 {
+		t.Fatalf("retry must clear pending queues after resend, got imgs=%d labels=%d docs=%d",
+			len(next.pendingImages), len(next.pendingImageLabels), len(next.pendingDocuments))
+	}
+	if len(next.lastImages) != 1 || len(next.lastDocuments) != 1 {
+		t.Fatalf("retry must keep the snapshot for a subsequent retry, got imgs=%d docs=%d",
+			len(next.lastImages), len(next.lastDocuments))
+	}
+
+	updated, _ = next.Update(execCmd(cmd))
+	_ = updated.(model)
+
+	select {
+	case imgs := <-captured:
+		if len(imgs) != 1 || imgs[0].MediaType != "image/png" {
+			t.Fatalf("retried run must resend the remembered image, got %#v", imgs)
+		}
+	default:
+		t.Fatal("expected the retried run to receive the remembered image")
+	}
+
+	if len(provider.requests) != 1 {
+		t.Fatalf("expected one provider request, got %d", len(provider.requests))
+	}
+	msgs := provider.requests[0].Messages
+	last := msgs[len(msgs)-1]
+	if !strings.Contains(last.Content, "Top secret design notes") {
+		t.Fatalf("retried prompt should re-prepend the document text, got:\n%s", last.Content)
+	}
+	if !strings.Contains(last.Content, "describe both") {
+		t.Fatalf("retried prompt should include the remembered user text, got:\n%s", last.Content)
+	}
+}

internal/tui/lsp_manager_test.go

@@ -0,0 +1,31 @@
+package tui
+
+import (
+	"context"
+	"testing"
+)
+
+// The LSP manager is built once at construction (when cwd is known) and reused —
+// a fresh manager per run would cold-start gopls on the first edit of every turn.
+func TestNewModelBuildsSharedLSPManager(t *testing.T) {
+	m := newModel(context.Background(), Options{Cwd: t.TempDir()})
+	if m.lspManager == nil {
+		t.Fatal("expected a session-long lspManager when cwd is set")
+	}
+	// The pointer is stable across model copies (Bubble Tea passes the model by
+	// value every Update), so every run shares the same warm servers.
+	copied := m
+	if copied.lspManager != m.lspManager {
+		t.Fatal("lspManager pointer must survive a model copy")
+	}
+}
+
+// shutdownLSPManager must be nil-safe: a model built without a cwd has no manager,
+// and quitting it must not panic.
+func TestShutdownLSPManagerNilSafe(t *testing.T) {
+	m := model{}
+	if m.lspManager != nil {
+		t.Fatal("expected no manager on a zero-value model")
+	}
+	m.shutdownLSPManager() // must not panic
+}

internal/tui/model.go

@@ -19,6 +19,7 @@ import (
 	"github.com/Gitlawb/zero/internal/agent"
 	"github.com/Gitlawb/zero/internal/config"
 	"github.com/Gitlawb/zero/internal/doctor"
+	"github.com/Gitlawb/zero/internal/errhint"
 	"github.com/Gitlawb/zero/internal/lsp"
 	internalmcp "github.com/Gitlawb/zero/internal/mcp"
 	"github.com/Gitlawb/zero/internal/modelregistry"
@@ -76,24 +77,29 @@ type model struct {
 	discoverProviderModels      func(context.Context, config.ProviderProfile) ([]providermodeldiscovery.Model, error)
 	discoverOllamaContextWindow func(ctx context.Context, baseURL string, model string) (int, error)
 	registry                    *tools.Registry
-	sessionStore                *sessions.Store
-	sandboxStore                *sandbox.GrantStore
-	mcpConfig                   config.MCPConfig
-	mcpPermissionStore          *internalmcp.PermissionStore
-	mcpTokenStore               *internalmcp.TokenStore
-	mcpCommand                  func(context.Context, []string) MCPCommandResult
-	sandboxSetupCommand         func(context.Context) SandboxSetupCommandResult
-	mcpViewStateCache           MCPViewState
-	mcpViewStateReady           bool
-	mcpCommandSeq               int
-	mcpCommandCancel            context.CancelFunc
-	sandboxSetupSeq             int
-	sandboxSetupInFlight        bool
-	doctorCommandSeq            int
-	doctorInFlight              bool
-	doctorFrame                 int
-	activeSession               sessions.Metadata
-	sessionEvents               []sessions.Event
+	// lspManager is created once per session and reused across prompts so gopls (and
+	// other language servers) stay warm — a fresh manager per run would cold-start
+	// the server on the first edit of every turn. Nil when cwd is unknown; runs then
+	// fall back to a per-run manager. Torn down in quit().
+	lspManager           *lsp.Manager
+	sessionStore         *sessions.Store
+	sandboxStore         *sandbox.GrantStore
+	mcpConfig            config.MCPConfig
+	mcpPermissionStore   *internalmcp.PermissionStore
+	mcpTokenStore        *internalmcp.TokenStore
+	mcpCommand           func(context.Context, []string) MCPCommandResult
+	sandboxSetupCommand  func(context.Context) SandboxSetupCommandResult
+	mcpViewStateCache    MCPViewState
+	mcpViewStateReady    bool
+	mcpCommandSeq        int
+	mcpCommandCancel     context.CancelFunc
+	sandboxSetupSeq      int
+	sandboxSetupInFlight bool
+	doctorCommandSeq     int
+	doctorInFlight       bool
+	doctorFrame          int
+	activeSession        sessions.Metadata
+	sessionEvents        []sessions.Event
 	// titledSessions records session ids for which a model-generated title has
 	// already been attempted this process, so a finished turn re-fires the title
 	// generator at most once per session (even before its async result lands).
@@ -261,13 +267,30 @@ type model struct {
 	headerPrinted bool
 
 	// Composer input history (shell-style ↑/↓ recall of submitted inputs).
+	// lastPrompt is the verbatim text of the most recent submitted prompt, so
+	// /retry can resend it and /edit can recall it into the composer.
+	lastPrompt string
+	// lastImages/lastImageLabels/lastDocuments remember the attachments consumed
+	// by the most recent submitted prompt. launchPrompt clears the pending queues
+	// once a turn is sent, so /retry re-stages these to reproduce the exact same
+	// request — otherwise a vision/PDF-backed prompt would silently retry as
+	// text-only and answer a different task. They share the underlying image bytes
+	// with the sent turn (never mutated in place), so no deep copy is needed.
+	lastImages      []zeroruntime.ImageBlock
+	lastImageLabels []string
+	lastDocuments   []pendingDocument
 	// historyIdx == len(inputHistory) means "not navigating"; historyDraft
 	// preserves whatever was typed before recall started.
 	inputHistory []string
 	historyIdx   int
 	historyDraft string
 
-	streamingText              string // live assistant text for the current segment
+	// streamingText is the live assistant text for the current segment, accumulated
+	// as []byte so each delta is an O(1) amortized append instead of the O(n²) that
+	// string += delta incurs across a long generation. Read via streamingTextString().
+	// A []byte (not strings.Builder) because the model is copied by value on every
+	// Update, which would trip strings.Builder's copy check.
+	streamingText              []byte
 	streamingReasoning         string // live provider reasoning for the current segment
 	streamingReasoningExpanded bool
 	// turnStreamedRunes accumulates every reasoning+answer rune streamed in the
@@ -773,6 +796,11 @@ func newModel(ctx context.Context, options Options) model {
 	// Streaming text always renders statically at base ink (the disabled path in
 	// styleStreamingLine), so no accent glow and no per-line fade ticks.
 	m.fadeDisabled = true
+	// One session-long LSP manager (cheap to build — servers start lazily on the
+	// first Check), reused across prompts so gopls stays warm between turns.
+	if cwd != "" {
+		m.lspManager = lsp.NewManager(cwd)
+	}
 	m.refreshMCPViewState()
 	return m
 }
@@ -902,9 +930,22 @@ func (m model) noBlockingModal() bool {
 func (m model) quit() (tea.Model, tea.Cmd) {
 	m.stopPRWatcher()
 	m.stopAllBackgroundTerminalSessions()
+	m.shutdownLSPManager()
 	return m, tea.Quit
 }
 
+// shutdownLSPManager gracefully stops the session-long language servers on exit.
+// Best-effort with a short deadline so a slow server can't hang the quit; the
+// servers are our child processes and would be reaped on exit regardless.
+func (m model) shutdownLSPManager() {
+	if m.lspManager == nil {
+		return
+	}
+	shutdownCtx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+	_ = m.lspManager.Shutdown(shutdownCtx)
+}
+
 func (m model) handleCtrlC() (tea.Model, tea.Cmd) {
 	if !m.pending && m.composerValue() != "" && m.noBlockingModal() && !m.transcriptDetailed && !m.subchat.active {
 		m.clearComposer()
@@ -1576,7 +1617,7 @@ func (m model) updateModel(msg tea.Msg) (tea.Model, tea.Cmd) {
 		// Streaming text means any in-progress tool call has finished — clear the
 		// live "writing" block so it doesn't linger over new prose.
 		m.clearStreamingToolCall()
-		m.streamingText += msg.delta
+		m.streamingText = append(m.streamingText, msg.delta...)
 		m.turnStreamedRunes += utf8.RuneCountInString(msg.delta)
 		// recordStreamingDelta appends a time.Time to lineAges for every
 		// newline in the delta and bumps lastStreamActivity. It also
@@ -1859,20 +1900,23 @@ func (m model) updateModel(msg tea.Msg) (tea.Model, tea.Cmd) {
 			if row, ok := reasoningTranscriptRow("", msg.runID, m.streamingReasoning); ok {
 				m.transcript = appendTranscriptRow(m.transcript, row)
 			}
-			if text := strings.TrimRight(m.streamingText, "\n"); strings.TrimSpace(text) != "" {
+			if text := strings.TrimRight(m.streamingTextString(), "\n"); strings.TrimSpace(text) != "" {
 				m.transcript = appendTranscriptRow(m.transcript, transcriptRow{kind: rowAssistant, text: text})
 			}
 			// The error row terminates the turn, so it carries the done-line
-			// metadata a final assistant row would have carried.
+			// metadata a final assistant row would have carried. A recognized
+			// provider failure (auth/rate-limit/connectivity/…) also carries a
+			// one-line next step so the user isn't left staring at a raw blob.
 			m.transcript = appendTranscriptRow(m.transcript, transcriptRow{
 				kind:        rowError,
 				text:        msg.err.Error(),
+				hint:        errhint.TUIHint(msg.err),
 				final:       true,
 				turnTools:   msg.turnTools,
 				turnElapsed: msg.turnElapsed,
 			})
 		}
-		m.streamingText = ""
+		m.streamingText = nil
 		m.streamingReasoning = ""
 		m.streamingReasoningExpanded = false
 		// Roll the completed run's wall-time into the session's rolling average so
@@ -2047,14 +2091,14 @@ func (m model) updateModel(msg tea.Msg) (tea.Model, tea.Cmd) {
 				m.streamingReasoning = ""
 				m.streamingReasoningExpanded = false
 			}
-			if text := strings.TrimRight(m.streamingText, "\n"); strings.TrimSpace(text) != "" {
+			if text := strings.TrimRight(m.streamingTextString(), "\n"); strings.TrimSpace(text) != "" {
 				m.transcript = appendTranscriptRow(m.transcript, transcriptRow{kind: rowAssistant, text: text})
 				// This interim narration is the agent explaining what it's about to
 				// do — attribute it to the active plan step so the step-detail card
 				// can replay the agent's own account of the work.
 				m = m.captureStepNarration(text)
 			}
-			m.streamingText = ""
+			m.streamingText = nil
 			// The tool call has finalized into its card — drop the live "writing"
 			// preview so it doesn't linger or duplicate beneath the card.
 			m.clearStreamingToolCall()
@@ -2785,7 +2829,7 @@ func (m model) liveReasoningBodyCap() int {
 }
 
 func (m model) interimBlock(width int) string {
-	text := strings.TrimRight(m.streamingText, "\n")
+	text := strings.TrimRight(m.streamingTextString(), "\n")
 	reasoning := strings.TrimRight(m.streamingReasoning, "\n")
 	blocks := []string{}
 	if strings.TrimSpace(reasoning) != "" {
@@ -2849,7 +2893,7 @@ func (m model) spinnerGlyph() string {
 // (reasoning, waiting on the model, or a tool in flight). Cheap and robust — no
 // transcript scan — so it can't misreport on a long, output-less step.
 func (m model) workingActivity() string {
-	if strings.TrimSpace(m.streamingText) != "" {
+	if strings.TrimSpace(m.streamingTextString()) != "" {
 		return "writing"
 	}
 	return "thinking"
@@ -3897,6 +3941,62 @@ func (m model) handleSubmit() (tea.Model, tea.Cmd) {
 		}
 		m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: "$ " + cmdText})
 		return m, runBashEscape(m.cwd, cmdText)
+	case commandRetry:
+		// /retry launches a run, so it needs the same guards a normal prompt gets:
+		// never start one while exiting (would strand the shutdown flush) or during
+		// compaction (would race compactResultMsg's wholesale rewrite of
+		// transcript/sessionEvents and silently drop events).
+		if m.exiting {
+			return m, nil
+		}
+		if m.pending {
+			m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: "Retry\ncannot retry while a run is in progress."})
+			return m, nil
+		}
+		if m.compactInFlight {
+			m.transcript = reduceTranscript(m.transcript, transcriptAction{
+				kind: actionAppendSystem,
+				text: "Retry\nstatus: warning\nCompaction is running. Retry once it finishes.",
+			})
+			return m, nil
+		}
+		if strings.TrimSpace(m.lastPrompt) == "" {
+			m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: "Retry\nno previous prompt to resend."})
+			return m, nil
+		}
+		// Re-stage the attachments the last prompt carried so launchPrompt rebuilds
+		// an identical request (document preamble + images + vision re-check). Without
+		// this the queues are empty and /retry would resend a text-only prompt,
+		// silently dropping the image/PDF context and answering a different task.
+		m.pendingImages = m.lastImages
+		m.pendingImageLabels = m.lastImageLabels
+		m.pendingDocuments = m.lastDocuments
+		return m.launchPrompt(m.lastPrompt)
+	case commandEdit:
+		if strings.TrimSpace(m.lastPrompt) == "" {
+			m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: "Edit\nno previous prompt to recall."})
+			return m, nil
+		}
+		// Re-stage the remembered attachments alongside the recalled text so an
+		// edited resend carries the same image/PDF context — the reappearing chip
+		// row is the visible confirmation. Without this, editing a vision- or
+		// document-backed prompt would silently submit a text-only version and
+		// answer a different task (the same gap /retry guards against).
+		m.pendingImages = m.lastImages
+		m.pendingImageLabels = m.lastImageLabels
+		m.pendingDocuments = m.lastDocuments
+		m.input.SetValue(m.lastPrompt)
+		return m, nil
+	case commandCopy:
+		text := m.lastAssistantAnswer()
+		if strings.TrimSpace(text) == "" {
+			m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: "Copy\nno answer to copy yet."})
+			return m, nil
+		}
+		return m, copyTranscriptSelectionCmd(text)
+	case commandExport:
+		m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendSystem, text: m.handleExportCommand(command.text)})
+		return m, nil
 	case commandPrompt:
 		if intent, ok := detectMCPSetupIntent(command.text); ok {
 			return m.openMCPAddWizardFromIntent(intent), nil
@@ -3911,6 +4011,15 @@ func (m model) handleSubmit() (tea.Model, tea.Cmd) {
 // composer. Queued prompts use this path too, so session and image behavior
 // stays identical to immediate submissions.
 func (m model) launchPrompt(prompt string) (model, tea.Cmd) {
+	// Remember the verbatim prompt (before specialist/document expansion) so /retry
+	// and /edit can act on exactly what the user submitted. Snapshot the staged
+	// attachments too: launchPrompt clears the pending queues below, so /retry
+	// re-stages these to resend an identical vision/PDF-backed request rather than
+	// a degraded text-only one.
+	m.lastPrompt = prompt
+	m.lastImages = m.pendingImages
+	m.lastImageLabels = m.pendingImageLabels
+	m.lastDocuments = m.pendingDocuments
 	m.transcript = reduceTranscript(m.transcript, transcriptAction{kind: actionAppendUser, text: prompt})
 	if m.provider == nil {
 		m.transcript = reduceTranscript(m.transcript, transcriptAction{
@@ -4098,7 +4207,7 @@ func (m *model) cancelRun() {
 		if row, ok := reasoningTranscriptRow("", m.activeRunID, m.streamingReasoning); ok {
 			m.transcript = appendTranscriptRow(m.transcript, row)
 		}
-		if text := strings.TrimRight(m.streamingText, "\n"); strings.TrimSpace(text) != "" {
+		if text := strings.TrimRight(m.streamingTextString(), "\n"); strings.TrimSpace(text) != "" {
 			m.transcript = appendTranscriptRow(m.transcript, transcriptRow{kind: rowAssistant, text: text})
 		}
 		m.transcript = appendTranscriptRow(m.transcript, transcriptRow{kind: rowSystem, text: "Run cancelled."})
@@ -4119,7 +4228,7 @@ func (m *model) cancelRun() {
 	m.pendingAskUser = nil
 	// The interim block renders streamingText live; a cancelled run's partial
 	// answer must not leak into (and concatenate with) the next turn's stream.
-	m.streamingText = ""
+	m.streamingText = nil
 	m.streamingReasoning = ""
 	m.streamingReasoningExpanded = false
 	// Hard-stop the fade and drop the per-line age map. The next turn's
@@ -4196,18 +4305,28 @@ func (m model) runAgentWithOptions(runID int, runCtx context.Context, prompt str
 		// matching exec; the per-turn lsp.Manager is torn down when this run
 		// returns; auto-fix vs report-only follows the active permission mode.
 		if !runOptions.specDraft && options.Cwd != "" {
-			lspManager := lsp.NewManager(options.Cwd)
-			defer func() {
-				shutdownCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
-				defer cancel()
-				_ = lspManager.Shutdown(shutdownCtx)
-			}()
+			// Prefer the session-long manager (kept warm across prompts). Only when it
+			// is absent — e.g. cwd was unknown at construction, or a test built the
+			// model directly — fall back to a per-run manager that is shut down here.
+			lspManager := m.lspManager
+			if lspManager == nil {
+				lspManager = lsp.NewManager(options.Cwd)
+				defer func() {
+					shutdownCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+					defer cancel()
+					_ = lspManager.Shutdown(shutdownCtx)
+				}()
+			}
 			options.SelfCorrect = agent.NewSelfCorrector(options.Cwd, agent.NewLSPDiagnosticsChecker(lspManager), agent.NewProjectVerifier(options.Cwd), agent.SelfCorrectConfig{
 				Enabled:      true,
 				IncludeTests: m.selfCorrectTests,
 				IncludeLSP:   true,
 				Autonomy:     selfCorrectAutonomyForMode(options.PermissionMode),
 			})
+			// Inline post-edit diagnostics: edit_file/write_file append error
+			// diagnostics for the file they just wrote to their own output, so the
+			// model sees a break in the same turn. Shares the run's lazy manager.
+			options.FileDiagnostics = agent.NewFileDiagnostics(lspManager, options.Cwd)
 		}
 
 		// Some providers synthesize tool-call ids that repeat within a run (e.g.
@@ -4659,6 +4778,13 @@ func (m model) sendAgentText(runID int, delta string) {
 	m.runtimeMessageSink(agentTextMsg{runID: runID, delta: delta})
 }
 
+// streamingTextString returns the accumulated live assistant text. streamingText
+// is stored as []byte for O(1) amortized appends; the conversion here is bounded
+// by the segment length, the same cost the renderer already pays.
+func (m model) streamingTextString() string {
+	return string(m.streamingText)
+}
+
 func (m model) sendToolCallStreamStart(runID int, id, name string) {
 	if m.runtimeMessageSink == nil {
 		return

internal/tui/model_test.go

@@ -2301,7 +2301,7 @@ func TestStreamingAssistantAfterToolCardGetsRuleSeparator(t *testing.T) {
 	m.width, m.height = 120, 40
 	m.altScreen = true
 	m.pending = true
-	m.streamingText = "Done."
+	m.streamingText = []byte("Done.")
 	m.transcript = append(m.transcript,
 		transcriptRow{kind: rowUser, text: "run it"},
 		transcriptRow{kind: rowAssistant, text: "I'll run it first."},

internal/tui/new_session_test.go

@@ -20,6 +20,10 @@ func TestStartNewSessionResetsState(t *testing.T) {
 	m.pendingImageLabels = []string{"pic.png"}
 	m.pendingDocuments = []pendingDocument{{label: "doc.pdf"}}
 	m.queuedMessage = "queued"
+	// The /retry attachment snapshot is prior-session state too and must not survive.
+	m.lastImages = make([]zeroruntime.ImageBlock, 1)
+	m.lastImageLabels = []string{"pic.png"}
+	m.lastDocuments = []pendingDocument{{label: "doc.pdf"}}
 
 	next := m.startNewSession()
 
@@ -41,6 +45,11 @@ func TestStartNewSessionResetsState(t *testing.T) {
 		t.Fatalf("startNewSession must clear staged input, got images=%d labels=%d docs=%d queued=%q",
 			len(next.pendingImages), len(next.pendingImageLabels), len(next.pendingDocuments), next.queuedMessage)
 	}
+	// The /retry snapshot must not leak the previous session's attachments.
+	if len(next.lastImages) != 0 || len(next.lastImageLabels) != 0 || len(next.lastDocuments) != 0 {
+		t.Fatalf("startNewSession must clear the retry snapshot, got images=%d labels=%d docs=%d",
+			len(next.lastImages), len(next.lastImageLabels), len(next.lastDocuments))
+	}
 }
 
 func TestNewCommandStartsFreshSession(t *testing.T) {

internal/tui/onboarding.go

@@ -1151,6 +1151,29 @@ func (m model) setupProvider() SetupProviderOption {
 	return m.setup.providers[index]
 }
 
+// setupErrorAffordance returns a faint one-line recovery hint tailored to the
+// current setup stage, so a first-run error points the user at the inline fix
+// instead of leaving them staring at a red line. Empty for stages whose footer
+// already makes recovery obvious. Each hint names only keys the stage's footer
+// confirms, so the guidance is always accurate.
+func (m model) setupErrorAffordance() string {
+	switch m.setup.stage {
+	case setupStageEndpoint:
+		return "→ edit the endpoint above, then Enter to retry (← to go back)"
+	case setupStageCredentials:
+		if m.setupCredentialInputActive() {
+			return "→ re-enter the key above, then Enter to retry (← to go back)"
+		}
+	case setupStageName:
+		return "→ edit the name above, then Enter to continue (← to go back)"
+	case setupStageModel:
+		if !m.setup.modelLoad {
+			return "→ pick another model with ↑/↓, or type to search"
+		}
+	}
+	return ""
+}
+
 func (m model) setupView(width int) string {
 	if width <= 0 {
 		width = defaultStartupWidth
@@ -1159,6 +1182,9 @@ func (m model) setupView(width int) string {
 	content := m.setupStageLines(width, height)
 	if m.setup.err != "" {
 		content = append(content, "", zeroTheme.red.Render("error: "+m.setup.err))
+		if hint := m.setupErrorAffordance(); hint != "" {
+			content = append(content, zeroTheme.faint.Render(hint))
+		}
 	}
 	progress := m.setupProgressText()
 	footer := m.setupFooter()

internal/tui/onboarding_affordance_test.go

@@ -0,0 +1,29 @@
+package tui
+
+import (
+	"strings"
+	"testing"
+)
+
+// Each input/choice stage's error affordance names an accurate recovery step;
+// stages without a tailored hint return empty.
+func TestSetupErrorAffordance(t *testing.T) {
+	endpoint := model{}
+	endpoint.setup.stage = setupStageEndpoint
+	if got := endpoint.setupErrorAffordance(); !strings.Contains(got, "endpoint") || !strings.Contains(got, "Enter") {
+		t.Fatalf("endpoint affordance = %q", got)
+	}
+
+	name := model{}
+	name.setup.stage = setupStageName
+	if got := name.setupErrorAffordance(); !strings.Contains(got, "name") {
+		t.Fatalf("name affordance = %q", got)
+	}
+
+	// A stage with no tailored hint (method chooser) returns empty.
+	method := model{}
+	method.setup.stage = setupStageMethod
+	if got := method.setupErrorAffordance(); got != "" {
+		t.Fatalf("method stage should have no affordance, got %q", got)
+	}
+}

internal/tui/picker.go

@@ -102,16 +102,83 @@ func (p *commandPicker) applyQuery() {
 		p.selected = clampInt(p.selected, 0, maxInt(0, len(p.items)-1))
 		return
 	}
-	filtered := make([]pickerItem, 0, len(source))
-	for _, item := range source {
-		if strings.Contains(strings.ToLower(strings.Join([]string{item.Group, item.Label, item.Value, item.Meta}, " ")), query) {
-			filtered = append(filtered, item)
+
+	// Rank matches (exact < prefix < contains < subsequence) instead of a flat
+	// substring filter, so the closest match to "sonnet 4.5" lands at the top
+	// rather than buried behind scrolling. Groups stay contiguous — a group is
+	// ordered by its best-matching item and never split into two header blocks —
+	// so the grouped model picker still renders one header per provider.
+	type entry struct {
+		item  pickerItem
+		score int
+		order int
+	}
+	groupFirst := map[string]int{}
+	groupBest := map[string]int{}
+	entries := make([]entry, 0, len(source))
+	for index, item := range source {
+		score, ok := scorePickerItem(item, query)
+		if !ok {
+			continue
 		}
+		if _, seen := groupFirst[item.Group]; !seen {
+			groupFirst[item.Group] = index
+		}
+		if best, seen := groupBest[item.Group]; !seen || score < best {
+			groupBest[item.Group] = score
+		}
+		entries = append(entries, entry{item: item, score: score, order: index})
+	}
+	sort.SliceStable(entries, func(a, b int) bool {
+		ga, gb := entries[a].item.Group, entries[b].item.Group
+		if ga != gb {
+			// Most-relevant group first; ties keep original group appearance order so
+			// a group is never scattered across the list.
+			if groupBest[ga] != groupBest[gb] {
+				return groupBest[ga] < groupBest[gb]
+			}
+			return groupFirst[ga] < groupFirst[gb]
+		}
+		// Within a group: best match first, then original order.
+		if entries[a].score != entries[b].score {
+			return entries[a].score < entries[b].score
+		}
+		return entries[a].order < entries[b].order
+	})
+	filtered := make([]pickerItem, 0, len(entries))
+	for _, e := range entries {
+		filtered = append(filtered, e.item)
 	}
 	p.items = filtered
 	p.selected = 0
 }
 
+// scorePickerItem ranks an item against a lowercased query; lower is better, and
+// ok is false when it doesn't match at all. Tiers mirror scoreFileSuggestion:
+// exact/prefix/contains on the label (what the user reads) beat matches deeper in
+// the joined haystack, and a fuzzy subsequence is the last-resort match.
+func scorePickerItem(item pickerItem, query string) (int, bool) {
+	label := strings.ToLower(item.Label)
+	hay := strings.ToLower(strings.Join([]string{item.Group, item.Label, item.Value, item.Meta}, " "))
+	switch {
+	case label == query:
+		return 0, true
+	case strings.HasPrefix(label, query):
+		return 20, true
+	case strings.Contains(label, query):
+		return 40, true
+	case strings.HasPrefix(hay, query):
+		return 60, true
+	case strings.Contains(hay, query):
+		return 80, true
+	default:
+		if gap, ok := fuzzySubsequenceGap(hay, query); ok {
+			return 120 + gap, true
+		}
+		return 0, false
+	}
+}
+
 // newModelPicker lists active (non-deprecated) models, preselecting the active
 // one. Returns nil when the catalog is unavailable so the caller falls back to
 // the plain status text.

internal/tui/picker_fuzzy_test.go

@@ -0,0 +1,72 @@
+package tui
+
+import (
+	"testing"
+)
+
+func newFuzzyTestPicker(items []pickerItem) *commandPicker {
+	return &commandPicker{
+		kind:     pickerModel,
+		items:    append([]pickerItem{}, items...),
+		allItems: append([]pickerItem{}, items...),
+	}
+}
+
+// A closer match ranks above a merely-containing one, so the best result lands at
+// the top instead of in original order.
+func TestPickerRanksBestMatchFirst(t *testing.T) {
+	p := newFuzzyTestPicker([]pickerItem{
+		{Label: "Claude Opus 4.5", Value: "opus-4.5"},
+		{Label: "Claude Sonnet 4.5", Value: "sonnet-4.5"},
+		{Label: "Sonnet", Value: "sonnet"}, // exact-ish prefix match should win
+	})
+	p.query = "sonnet"
+	p.applyQuery()
+
+	if len(p.items) != 2 {
+		t.Fatalf("expected 2 matches for 'sonnet', got %d: %#v", len(p.items), p.items)
+	}
+	if p.items[0].Value != "sonnet" {
+		t.Fatalf("exact/prefix match should rank first, got %q", p.items[0].Value)
+	}
+}
+
+// A subsequence query (non-contiguous characters) still matches when no substring does.
+func TestPickerFuzzySubsequenceMatches(t *testing.T) {
+	p := newFuzzyTestPicker([]pickerItem{
+		{Label: "Claude Sonnet 4.5", Value: "sonnet-4.5"},
+		{Label: "GPT-5", Value: "gpt-5"},
+	})
+	p.query = "snt45" // subsequence of "sonnet 4.5", not a substring
+	p.applyQuery()
+
+	if len(p.items) != 1 || p.items[0].Value != "sonnet-4.5" {
+		t.Fatalf("subsequence query should match sonnet-4.5, got %#v", p.items)
+	}
+}
+
+// Ranking keeps each group contiguous: a filtered result never splits one group
+// into two separate blocks, so group headers still render once per group.
+func TestPickerRankingKeepsGroupsContiguous(t *testing.T) {
+	p := newFuzzyTestPicker([]pickerItem{
+		{Group: "openai", Label: "gpt-5 mini", Value: "a"},
+		{Group: "anthropic", Label: "sonnet 4.5", Value: "b"},
+		{Group: "openai", Label: "gpt-5 codex", Value: "c"},
+		{Group: "anthropic", Label: "opus 4.5", Value: "d"},
+	})
+	p.query = "5" // matches both openai models and the "4.5" anthropic ones
+	p.applyQuery()
+
+	// Every item's group must appear as one contiguous run.
+	seen := map[string]bool{}
+	last := ""
+	for _, item := range p.items {
+		if item.Group != last {
+			if seen[item.Group] {
+				t.Fatalf("group %q split into non-contiguous blocks: %#v", item.Group, p.items)
+			}
+			seen[item.Group] = true
+			last = item.Group
+		}
+	}
+}

internal/tui/rendering.go

@@ -987,6 +987,12 @@ func isDoctorResultHeading(value string) bool {
 
 func renderErrorRow(row transcriptRow, width int) string {
 	note := noteBox(row.text, width, zeroTheme.cardErr, zeroTheme.red)
+	// A recognized failure carries a one-line next step. Render it just below the
+	// red box in the faint metadata style so it reads as guidance, not more error
+	// text (and to avoid nesting ANSI styles inside noteBox's per-line red wrap).
+	if hint := strings.TrimSpace(row.hint); hint != "" {
+		note += "\n" + fitStyledLine(zeroTheme.faint.Render("→ "+hint), width)
+	}
 	return note
 }
 

internal/tui/rendering_lime_test.go

@@ -224,14 +224,14 @@ func TestCommandCardRowTrimsIndentedActionsLabel(t *testing.T) {
 func TestInterimBlockShowsStreamingTextWithCursor(t *testing.T) {
 	m := limeTestModel()
 	m.pending = true
-	m.streamingText = "I'll add a --version flag"
+	m.streamingText = []byte("I'll add a --version flag")
 	got := plainRender(t, m.interimBlock(96))
 	if !strings.Contains(got, "I'll add a --version flag") || !strings.Contains(got, "▌") {
 		t.Fatalf("interim block = %q, want streamed text with trailing cursor", got)
 	}
 
 	// Before the first delta the block falls back to the liveness spinner.
-	m.streamingText = ""
+	m.streamingText = nil
 	if got := plainRender(t, m.interimBlock(96)); !strings.Contains(got, "Working") {
 		t.Fatalf("empty interim block = %q, want the liveness label", got)
 	}
@@ -240,13 +240,13 @@ func TestInterimBlockShowsStreamingTextWithCursor(t *testing.T) {
 func TestInterimBlockRendersStreamingMarkdownTable(t *testing.T) {
 	m := limeTestModel()
 	m.pending = true
-	m.streamingText = strings.Join([]string{
+	m.streamingText = []byte(strings.Join([]string{
 		"Here's the comparison:",
 		"",
 		"| Category | System A | System B |",
 		"|---|---|---|",
 		"| **Label** | Alpha | Beta |",
-	}, "\n")
+	}, "\n"))
 
 	rendered := m.interimBlock(72)
 	got := plainRender(t, rendered)
@@ -1528,10 +1528,10 @@ func TestCancelRunClearsStreamingText(t *testing.T) {
 	m := limeTestModel()
 	m.pending = true
 	m.activeRunID = 3
-	m.streamingText = "partial answer from a doomed run"
+	m.streamingText = []byte("partial answer from a doomed run")
 	m.cancelRun()
-	if m.streamingText != "" {
-		t.Fatalf("cancelRun must clear streamingText, got %q", m.streamingText)
+	if len(m.streamingText) != 0 {
+		t.Fatalf("cancelRun must clear streamingText, got %q", string(m.streamingText))
 	}
 }
 

internal/tui/resume_scope_test.go

@@ -0,0 +1,31 @@
+package tui
+
+import (
+	"runtime"
+	"testing"
+)
+
+func TestSessionMatchesWorkspace(t *testing.T) {
+	cases := []struct {
+		name        string
+		sessionCwd  string
+		workspace   string
+		wantVisible bool
+	}{
+		{"same workspace", "/home/u/proj", "/home/u/proj", true},
+		{"trailing slash normalizes", "/home/u/proj/", "/home/u/proj", true},
+		{"different workspace hidden", "/home/u/other", "/home/u/proj", false},
+		{"session with no cwd stays visible", "", "/home/u/proj", true},
+		{"unknown current workspace keeps all", "/home/u/other", "", true},
+		// Casing: matched case-insensitively on Windows (its FS is), case-sensitively
+		// elsewhere. This exercises the platform-specific branch on each OS's CI.
+		{"case differs only", "/home/U/Proj", "/home/u/proj", runtime.GOOS == "windows"},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := sessionMatchesWorkspace(tc.sessionCwd, tc.workspace); got != tc.wantVisible {
+				t.Fatalf("sessionMatchesWorkspace(%q, %q) = %v, want %v", tc.sessionCwd, tc.workspace, got, tc.wantVisible)
+			}
+		})
+	}
+}

internal/tui/retry_edit_test.go

@@ -0,0 +1,96 @@
+package tui
+
+import (
+	"context"
+	"testing"
+
+	tea "charm.land/bubbletea/v2"
+
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+func TestCommandsRegistered(t *testing.T) {
+	for _, name := range []string{"/retry", "/edit", "/copy", "/export"} {
+		if cmd, ok := resolveCommand(name); !ok {
+			t.Fatalf("%s should be a registered command", name)
+		} else if cmd.name != name {
+			t.Fatalf("resolveCommand(%q) = %q", name, cmd.name)
+		}
+	}
+}
+
+// /edit recalls the last prompt into the composer for editing.
+func TestEditRecallsLastPrompt(t *testing.T) {
+	m := newModel(context.Background(), Options{})
+	m.lastPrompt = "refactor the parser"
+	m.input.SetValue("/edit")
+
+	updated, _ := m.Update(testKey(tea.KeyEnter))
+	next := updated.(model)
+
+	if got := next.composerValue(); got != "refactor the parser" {
+		t.Fatalf("/edit should recall last prompt into composer, got %q", got)
+	}
+}
+
+// /edit must re-stage the remembered attachments alongside the recalled text, so an
+// edited resend of a vision/PDF-backed prompt carries the same image/document
+// context instead of silently sending a text-only version.
+func TestEditRestagesAttachments(t *testing.T) {
+	m := newModel(context.Background(), Options{ModelName: "gpt-4.1"})
+	m.lastPrompt = "describe the diagram"
+	m.lastImages = []zeroruntime.ImageBlock{{MediaType: "image/png"}}
+	m.lastImageLabels = []string{"diagram.png"}
+	m.lastDocuments = []pendingDocument{{label: "spec.pdf", text: "notes"}}
+	m.input.SetValue("/edit")
+
+	updated, _ := m.Update(testKey(tea.KeyEnter))
+	next := updated.(model)
+
+	if got := next.composerValue(); got != "describe the diagram" {
+		t.Fatalf("/edit should recall the prompt text, got %q", got)
+	}
+	if len(next.pendingImages) != 1 || len(next.pendingImageLabels) != 1 || next.pendingImageLabels[0] != "diagram.png" {
+		t.Fatalf("/edit should re-stage the remembered image, got imgs=%d labels=%v",
+			len(next.pendingImages), next.pendingImageLabels)
+	}
+	if len(next.pendingDocuments) != 1 || next.pendingDocuments[0].label != "spec.pdf" {
+		t.Fatalf("/edit should re-stage the remembered document, got %#v", next.pendingDocuments)
+	}
+}
+
+// /retry with no prior prompt reports that there's nothing to resend rather than
+// launching an empty run.
+func TestRetryWithoutPriorPromptIsNoOp(t *testing.T) {
+	m := newModel(context.Background(), Options{})
+	m.input.SetValue("/retry")
+
+	updated, _ := m.Update(testKey(tea.KeyEnter))
+	next := updated.(model)
+
+	if !transcriptContains(next.transcript, "no previous prompt") {
+		t.Fatalf("/retry with no history should note there's nothing to resend, got %#v", next.transcript)
+	}
+}
+
+// /retry must not launch a run while compaction is rewriting session state — the
+// same guard a normal prompt has.
+func TestRetryBlockedDuringCompaction(t *testing.T) {
+	m := newModel(context.Background(), Options{})
+	m.lastPrompt = "do the thing"
+	m.compactInFlight = true
+	m.input.SetValue("/retry")
+
+	updated, cmd := m.Update(testKey(tea.KeyEnter))
+	next := updated.(model)
+
+	if next.pending {
+		t.Fatal("/retry must not start a run during compaction")
+	}
+	if cmd != nil {
+		t.Fatal("/retry during compaction must not return a run command")
+	}
+	if !transcriptContains(next.transcript, "Compaction is running") {
+		t.Fatalf("/retry during compaction should warn, got %#v", next.transcript)
+	}
+}

internal/tui/run.go

@@ -25,7 +25,7 @@ func Run(ctx context.Context, options Options) int {
 
 	externalSink := options.RuntimeMessageSink
 	var program *tea.Program
-	options.RuntimeMessageSink = func(msg tea.Msg) {
+	forward := func(msg tea.Msg) {
 		if externalSink != nil {
 			externalSink(msg)
 		}
@@ -33,6 +33,10 @@ func Run(ctx context.Context, options Options) int {
 			program.Send(msg)
 		}
 	}
+	// Coalesce streamed assistant-text deltas to ~one frame each so a fast provider
+	// can't drive a full Update→View per token; every other message flushes pending
+	// text first, keeping order intact.
+	options.RuntimeMessageSink = newTextCoalescer(forward).send
 	options.AltScreen = useAltScreen(options)
 
 	programOpts := []tea.ProgramOption{

internal/tui/session.go

@@ -4,6 +4,8 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"path/filepath"
+	"runtime"
 	"strings"
 	"time"
 
@@ -80,6 +82,12 @@ func (m model) startNewSession() model {
 	m.pendingImageLabels = nil
 	m.pendingDocuments = nil
 	m.queuedMessage = ""
+	// The remembered /retry attachment snapshot belongs to the previous session
+	// too — dropping it keeps a post-/new /retry from re-staging old images or
+	// documents. lastPrompt is composer history (like inputHistory) and stays.
+	m.lastImages = nil
+	m.lastImageLabels = nil
+	m.lastDocuments = nil
 
 	note := "Started a new session."
 	if previousID != "" {
@@ -220,9 +228,11 @@ func (m model) sessionPrompt(prompt string) string {
 
 func (m model) resolveResumeSession(args string) (*sessions.Metadata, error) {
 	if strings.EqualFold(args, "latest") {
-		// Latest *resumable* conversation, so "latest" never lands on a child or
-		// spec sub-run. An explicit `/resume <id>` below still resolves any session.
-		latest, err := m.sessionStore.LatestResumable()
+		// Latest *resumable* conversation IN THIS WORKSPACE, so "latest" never lands
+		// on a child, a spec sub-run, or a session from another project (matching the
+		// workspace-scoped picker). An explicit `/resume <id>` below still resolves
+		// any session regardless of workspace.
+		latest, err := m.latestResumableInWorkspace()
 		if err != nil {
 			return nil, err
 		}
@@ -321,6 +331,18 @@ func (m model) newSessionPicker() *commandPicker {
 	now := m.now()
 	items := make([]pickerItem, 0, len(metas))
 	for _, meta := range metas {
+		// Workspace-scoped: hide sessions from other project directories so /resume
+		// lists this workspace's history, not every project's. Checked BEFORE the
+		// per-session event read below, so a large global history doesn't pay 50
+		// full file reads to build one workspace's list. Sessions with no recorded
+		// Cwd (older runs) stay visible rather than vanishing.
+		if !sessionMatchesWorkspace(meta.Cwd, m.cwd) {
+			continue
+		}
+		// A zero-event session has nothing to resume — skip it without a file read.
+		if meta.EventCount == 0 {
+			continue
+		}
 		// Skip empty/failed runs (no assistant output, no tool calls) — e.g. the
 		// same prompt retried while the model wasn't responding. They have nothing
 		// to resume and otherwise flood the list with identical rows. Still on disk.
@@ -356,6 +378,52 @@ func (m model) newSessionPicker() *commandPicker {
 // resuming: a tool call/result, or a non-user message with real content (not the
 // no-output guardrail stop). Empty/failed runs return false and are hidden from
 // the picker (they stay on disk). Errors fail open (the session is kept).
+// latestResumableInWorkspace returns the most-recently-updated resumable session
+// belonging to the current workspace, or nil when none exist. ListResumable is
+// ordered latest-first, so the first qualifying match is the latest. It applies
+// the SAME filters as newSessionPicker — workspace membership, non-empty
+// metadata, and real resumable content — so `/resume latest` never lands on a
+// zero-event or empty/failed run the picker would have hidden.
+func (m model) latestResumableInWorkspace() (*sessions.Metadata, error) {
+	metas, err := m.sessionStore.ListResumable()
+	if err != nil {
+		return nil, err
+	}
+	for i := range metas {
+		if !sessionMatchesWorkspace(metas[i].Cwd, m.cwd) {
+			continue
+		}
+		if metas[i].EventCount == 0 {
+			continue
+		}
+		if !m.sessionHasResumableContent(metas[i].SessionID) {
+			continue
+		}
+		return &metas[i], nil
+	}
+	return nil, nil
+}
+
+// sessionMatchesWorkspace reports whether a session recorded in sessionCwd
+// belongs to the current workspaceCwd. A session with no recorded Cwd (older
+// runs) or an unknown current workspace is kept visible rather than filtered out,
+// so the scoping never hides history it can't confidently place elsewhere. On
+// Windows the comparison is case-insensitive, since the filesystem is and the
+// same workspace can be spelled with different casing (C:\Proj vs c:\proj).
+func sessionMatchesWorkspace(sessionCwd, workspaceCwd string) bool {
+	sessionCwd = strings.TrimSpace(sessionCwd)
+	workspaceCwd = strings.TrimSpace(workspaceCwd)
+	if sessionCwd == "" || workspaceCwd == "" {
+		return true
+	}
+	a := filepath.Clean(sessionCwd)
+	b := filepath.Clean(workspaceCwd)
+	if runtime.GOOS == "windows" {
+		return strings.EqualFold(a, b)
+	}
+	return a == b
+}
+
 func (m model) sessionHasResumableContent(sessionID string) bool {
 	events, err := m.sessionStore.ReadEvents(sessionID)
 	if err != nil {

internal/tui/session_controls.go

@@ -326,8 +326,10 @@ func (m model) handleCompactCommand(args string) (model, string, tea.Cmd) {
 	if args == "status" {
 		return m, m.compactText(false), nil
 	}
-	if args != "" {
-		return m, "Compact\nusage: /compact [status]", nil
+	// Bare "/compact" already triggers compaction; accept "now" too since that is
+	// what users reach for when they see the context gauge climbing.
+	if args != "" && args != "now" {
+		return m, "Compact\nusage: /compact [status|now]", nil
 	}
 	if m.compactInFlight {
 		return m, m.compactText(true), nil

internal/tui/syntax_highlight_test.go

@@ -7,7 +7,7 @@ import (
 
 func TestStreamingCodeRendersHighlighted(t *testing.T) {
 	m := model{
-		streamingText: "```go\nfunc main() {}\n```",
+		streamingText: []byte("```go\nfunc main() {}\n```"),
 		pending:       true,
 	}
 	out := m.interimBlock(80)
@@ -86,7 +86,7 @@ func TestStreamingMarkdownStablePrefixUsesRenderCache(t *testing.T) {
 
 func TestStreamingBuffersOpenFencedCodeBlock(t *testing.T) {
 	open := model{
-		streamingText: "Here is the script:\n```python\nfrom datetime import datetime\nprint(datetime.now())",
+		streamingText: []byte("Here is the script:\n```python\nfrom datetime import datetime\nprint(datetime.now())"),
 		pending:       true,
 	}
 	openOut := plainRender(t, open.interimBlock(90))
@@ -98,7 +98,7 @@ func TestStreamingBuffersOpenFencedCodeBlock(t *testing.T) {
 	}
 
 	closed := model{
-		streamingText: open.streamingText + "\n```",
+		streamingText: []byte(string(open.streamingText) + "\n```"),
 		pending:       true,
 	}
 	closedOut := closed.interimBlock(90)

internal/tui/transcript.go

@@ -35,6 +35,7 @@ type transcriptRow struct {
 	tool       string       // tool name, for tool call/result rows
 	status     tools.Status // result status, for tool result rows
 	detail     string       // raw multi-line output (e.g. a diff to render as a card)
+	hint       string       // one-line actionable hint, rendered faintly below error rows
 	arg        string       // secondary argument hint (pattern/command), for tool call rows
 	runID      int          // owning run, for tool call rows (0 = rehydrated/unknown)
 	permission *agent.PermissionEvent

internal/tui/transcript_export.go

@@ -0,0 +1,83 @@
+package tui
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+)
+
+// lastAssistantAnswer returns the text of the most recent assistant row — the
+// last final answer if one exists, else the last assistant row of any kind. Empty
+// when the conversation has no assistant text yet.
+func (m model) lastAssistantAnswer() string {
+	lastAny := ""
+	for i := len(m.transcript) - 1; i >= 0; i-- {
+		row := m.transcript[i]
+		if row.kind != rowAssistant {
+			continue
+		}
+		if row.final {
+			return row.text
+		}
+		if lastAny == "" {
+			lastAny = row.text
+		}
+	}
+	return lastAny
+}
+
+// plainTranscriptText renders the conversation as a plain, role-prefixed text
+// document for /export — the readable content (user prompts, assistant answers,
+// system notes, errors), skipping tool-call/permission UI noise.
+func (m model) plainTranscriptText() string {
+	var b strings.Builder
+	for _, row := range m.transcript {
+		var prefix string
+		switch row.kind {
+		case rowUser:
+			prefix = "you: "
+		case rowAssistant:
+			prefix = "zero: "
+		case rowSystem:
+			prefix = "· "
+		case rowError:
+			prefix = "error: "
+		default:
+			continue
+		}
+		text := strings.TrimRight(row.text, "\n")
+		if strings.TrimSpace(text) == "" {
+			continue
+		}
+		b.WriteString(prefix)
+		b.WriteString(text)
+		b.WriteString("\n\n")
+	}
+	return b.String()
+}
+
+// handleExportCommand writes the transcript to a file and returns a status
+// message. With no argument it derives a timestamped filename in the workspace;
+// a relative path is resolved against the workspace root.
+func (m model) handleExportCommand(args string) string {
+	body := m.plainTranscriptText()
+	if strings.TrimSpace(body) == "" {
+		return "Export\nnothing to export yet."
+	}
+
+	path := strings.TrimSpace(args)
+	if path == "" {
+		stamp := m.now().Format("20060102-150405")
+		path = fmt.Sprintf("zero-transcript-%s.txt", stamp)
+	}
+	if !filepath.IsAbs(path) && m.cwd != "" {
+		path = filepath.Join(m.cwd, path)
+	}
+	// 0o600: a transcript can include tool/bash output that echoed secrets, so it
+	// must not be world/group-readable on a shared machine.
+	if err := os.WriteFile(path, []byte(body), 0o600); err != nil {
+		return "Export\nfailed to write " + path + ": " + err.Error()
+	}
+	return "Export\nwrote transcript to " + path
+}

internal/tui/transcript_export_test.go

@@ -0,0 +1,103 @@
+package tui
+
+import (
+	"os"
+	"path/filepath"
+	"runtime"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestLastAssistantAnswerPrefersFinal(t *testing.T) {
+	m := model{transcript: []transcriptRow{
+		{kind: rowUser, text: "hi"},
+		{kind: rowAssistant, text: "interim narration"},
+		{kind: rowAssistant, text: "the final answer", final: true},
+		{kind: rowSystem, text: "worked for 3s"},
+	}}
+	if got := m.lastAssistantAnswer(); got != "the final answer" {
+		t.Fatalf("lastAssistantAnswer = %q, want the final answer", got)
+	}
+
+	// With no final row, falls back to the most recent assistant row.
+	m2 := model{transcript: []transcriptRow{
+		{kind: rowAssistant, text: "first"},
+		{kind: rowAssistant, text: "second"},
+	}}
+	if got := m2.lastAssistantAnswer(); got != "second" {
+		t.Fatalf("fallback lastAssistantAnswer = %q, want second", got)
+	}
+
+	// Empty when there's no assistant text.
+	if got := (model{transcript: []transcriptRow{{kind: rowUser, text: "hi"}}}).lastAssistantAnswer(); got != "" {
+		t.Fatalf("expected empty answer, got %q", got)
+	}
+}
+
+func TestPlainTranscriptTextSkipsNoise(t *testing.T) {
+	m := model{transcript: []transcriptRow{
+		{kind: rowUser, text: "add a flag"},
+		{kind: rowToolCall, tool: "bash", text: "go build"},
+		{kind: rowToolResult, tool: "bash", text: "ok"},
+		{kind: rowAssistant, text: "Done — added --version.", final: true},
+		{kind: rowError, text: "provider error: boom"},
+	}}
+	out := m.plainTranscriptText()
+	if !strings.Contains(out, "you: add a flag") || !strings.Contains(out, "zero: Done — added --version.") {
+		t.Fatalf("export missing conversation text:\n%s", out)
+	}
+	if !strings.Contains(out, "error: provider error: boom") {
+		t.Fatalf("export should include error rows:\n%s", out)
+	}
+	if strings.Contains(out, "go build") || strings.Contains(out, "bash") {
+		t.Fatalf("export should skip tool-call noise:\n%s", out)
+	}
+}
+
+func TestHandleExportCommandWritesFile(t *testing.T) {
+	dir := t.TempDir()
+	m := model{
+		cwd: dir,
+		now: func() time.Time { return time.Date(2026, 7, 4, 9, 30, 0, 0, time.UTC) },
+		transcript: []transcriptRow{
+			{kind: rowUser, text: "hello"},
+			{kind: rowAssistant, text: "hi there", final: true},
+		},
+	}
+
+	// Explicit relative path resolves against cwd.
+	msg := m.handleExportCommand("out.txt")
+	if !strings.Contains(msg, "wrote transcript to") {
+		t.Fatalf("export status = %q", msg)
+	}
+	outPath := filepath.Join(dir, "out.txt")
+	data, err := os.ReadFile(outPath)
+	if err != nil {
+		t.Fatalf("reading exported file: %v", err)
+	}
+	if !strings.Contains(string(data), "you: hello") || !strings.Contains(string(data), "zero: hi there") {
+		t.Fatalf("exported content = %q", string(data))
+	}
+	// The transcript may contain secrets echoed in tool output; it must not be
+	// world/group-readable. Unix perm bits are meaningless on Windows (os.WriteFile
+	// ignores them and Stat reports 0666), so only assert the mode where it applies.
+	if runtime.GOOS != "windows" {
+		if info, err := os.Stat(outPath); err != nil {
+			t.Fatalf("stat exported file: %v", err)
+		} else if perm := info.Mode().Perm(); perm != 0o600 {
+			t.Fatalf("exported file mode = %o, want 0600", perm)
+		}
+	}
+
+	// No-arg export derives a timestamped filename in cwd.
+	if msg := m.handleExportCommand(""); !strings.Contains(msg, "zero-transcript-20260704-093000.txt") {
+		t.Fatalf("default export status = %q", msg)
+	}
+
+	// Nothing to export on an empty conversation.
+	empty := model{cwd: dir, now: m.now}
+	if msg := empty.handleExportCommand(""); !strings.Contains(msg, "nothing to export") {
+		t.Fatalf("empty export status = %q", msg)
+	}
+}

internal/tui/transcript_selection.go

@@ -366,7 +366,7 @@ func (m model) transcriptBodyItems(width int, emptyOverlay string) []transcriptB
 	}
 
 	if m.pending {
-		pendingShowsAssistantText := m.pendingPermission == nil && m.pendingAskUser == nil && strings.TrimSpace(m.streamingText) != ""
+		pendingShowsAssistantText := m.pendingPermission == nil && m.pendingAskUser == nil && strings.TrimSpace(m.streamingTextString()) != ""
 		if pendingShowsAssistantText && havePreviousKind && shouldRuleBeforeTurn(previousKind, rowAssistant) {
 			items = append(items, transcriptRuleBodyItem(contentWidth, gutter))
 		} else {

internal/tui/vision_drop_test.go

@@ -0,0 +1,23 @@
+package tui
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/Gitlawb/zero/internal/zeroruntime"
+)
+
+func TestVisionDropWarning(t *testing.T) {
+	// No staged images: never warns, regardless of model.
+	if got := (model{}).visionDropWarning(); got != "" {
+		t.Fatalf("no images should give no warning, got %q", got)
+	}
+
+	// Staged images + a model with no vision support (empty model name qualifies):
+	// warn immediately, naming the count.
+	withImg := model{pendingImages: make([]zeroruntime.ImageBlock, 2)}
+	warn := withImg.visionDropWarning()
+	if !strings.Contains(warn, "will be dropped") || !strings.Contains(warn, "2 staged") {
+		t.Fatalf("expected a drop warning naming the count, got %q", warn)
+	}
+}

internal/tui/working_status_test.go

@@ -191,7 +191,7 @@ func TestInterimBlockShowsWorkingLineWithStreamedText(t *testing.T) {
 	base := time.Date(2026, 6, 14, 10, 0, 0, 0, time.UTC)
 	m.now = func() time.Time { return base.Add(12 * time.Second) }
 	m.turnStartedAt = base
-	m.streamingText = "partial answer so far"
+	m.streamingText = []byte("partial answer so far")
 
 	got := plainRender(t, m.interimBlock(96))
 	if !strings.Contains(got, "partial answer so far") {
@@ -241,7 +241,7 @@ func TestWorkingTokenIndicatorAccumulatesAcrossSegmentClears(t *testing.T) {
 	// Simulate the segment boundary that clears the live buffers, then stream
 	// answer text in the next segment.
 	m.streamingReasoning = ""
-	m.streamingText = ""
+	m.streamingText = nil
 	updated, _ = m.Update(agentTextMsg{runID: rid, delta: strings.Repeat("b", 40)})
 	m = updated.(model)
 
@@ -288,7 +288,7 @@ func TestInterimBlockShowsReasoningPreviewWhileThinking(t *testing.T) {
 	m.now = func() time.Time { return base.Add(90 * time.Second) }
 	m.turnStartedAt = base
 	m.streamingReasoning = "analyzing the layout\nthe patch was corrupt so re-planning the css edits"
-	m.streamingText = "" // thinking phase: no answer yet
+	m.streamingText = nil // thinking phase: no answer yet
 
 	got := plainRender(t, m.interimBlock(96))
 	if !strings.Contains(got, "re-planning the css edits") {
@@ -307,7 +307,7 @@ func TestInterimBlockNoPreviewWhenReasoningExpanded(t *testing.T) {
 	m.now = func() time.Time { return time.Date(2026, 6, 18, 23, 0, 0, 0, time.UTC) }
 	m.streamingReasoningExpanded = true
 	m.streamingReasoning = "only line of reasoning here"
-	m.streamingText = ""
+	m.streamingText = nil
 	got := plainRender(t, m.interimBlock(96))
 	if strings.Count(got, "only line of reasoning here") != 1 {
 		t.Fatalf("reasoning should appear exactly once when expanded (no preview dup):\n%s", got)