merge: feat/task-2-subagent-orchestration

TASKS-2 Subagent Orchestration (Container Model):
- ContainerMessage + channel-based escalation tools
- Deliberate (plan-first) and Exploratory preset workflows
- AskConductor/SubmitPlan (subagent) + Answer/Review (conductor)
- SessionRecorder DAG integration (question/plan turns)
- Environment context injection and orchestration guidance
- gci import ordering fix across pkg/{tools,agent,session}

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

CLAUDE.md

@@ -19,12 +19,24 @@ Lint: `golangci-lint run`
 - **Interview tool filtering**: `interviewAllowedTools` in `pkg/agent/loop.go` is the single source of truth for tools available during interview/review phases. Both `filterInterviewTools` (strips definitions before LLM call) and `isToolAllowedDuringInterview` (argument-level gating) reference this map.
 - **History clear**: `/plan start clear` wipes session history and summary on transition to executing. The Mini App review UI offers two sliders: standard approve and approve-with-clear.
 
-## Subagent Orchestration (実装済み部分)
+## Subagent Orchestration (実装済み)
 
 - **Startup flag**: `--orchestration` で on/off。`SubagentsConfig.Enabled` で gate。
 - **Conductor identity**: orchestration 有効時に conductor identity + spawn/subagent guidance を system prompt へ注入。
 - **Sandbox/Spawn**: `pkg/tools/sandbox.go`, `pkg/tools/spawn.go` 実装済み。
 - **AgentReporter**: `orch.AgentReporter` / `orch.Noop` / `orch.Broadcaster` で統一。main/heartbeat/subagent 全セッションが同一 Broadcaster に発火。Mini App は `agentLoop.GetOrchBroadcaster()` → `handler.SetOrchBroadcaster()` で受信。
+- **Container Model (Q&A escalation)**:
+  - `ContainerMessage` + `inCh`/`outCh` channels on `SubagentTask` — deliberate preset (coder/worker/coordinator) のみ
+  - `ask_conductor` tool — subagent → conductor question (blocking)
+  - `answer_subagent` tool — conductor → subagent answer
+  - `submit_plan` tool — subagent → conductor plan review (blocking)
+  - `review_subagent_plan` tool — conductor → subagent approve/reject
+  - `PendingQuestions()` で conductor LLM loop に question/plan_review を注入
+- **Deliberate Plan Mode**: `SubagentPlanState` (Clarifying → Review → Executing → Completed)
+  - `runDeliberateTask()`: clarifying phase (ask_conductor + submit_plan のみ) → executing phase (全ツール)
+  - `runExploratoryTask()`: exploratory preset の single-phase loop
+- **Environment injection**: `extractPlanContext()` で MEMORY.md から Context/Commands/Orchestration セクションを抽出 → subagent system prompt に注入
+- **SessionRecorder 拡張**: `RecordQuestion()` / `RecordPlanSubmit()` + `TurnQuestion` / `TurnPlanSubmit` TurnKind
 
 ## Session DAG (Phase 0–3 実装済み)
 
@@ -61,7 +73,7 @@ Lint: `golangci-lint run`
 | ファイル | 概要 |
 |---|---|
 | [`todo/TASKS-1.md`](todo/TASKS-1.md) | ~~**Memory & Performance Optimization**~~ ✅ 実装済み（MemoryStore キャッシュ＋パース済み state、FunctionCall.Arguments map統一、ToolDefinition.Parameters RawMessage化、検索結果フォーマット共通化、stats 定期フラッシュ） |
-| [`todo/TASKS-2.md`](todo/TASKS-2.md) | **Subagent Orchestration (Container Model)** — SubagentContainer、Orchestrator、Presets enforcement、Subagent Plan Mode（TASKS-1 の型変更前提メモ追記済み） |
+| [`todo/TASKS-2.md`](todo/TASKS-2.md) | ~~**Subagent Orchestration (Container Model)**~~ ✅ 実装済み（Container Q&A escalation、Deliberate Plan Mode、Environment injection、SessionRecorder 拡張） |
 | [`todo/TASKS-3.md`](todo/TASKS-3.md) | ~~**Session DAG (SQLite Store)**~~ ✅ 実装済み（Phase 0–3: SQLite SessionStore、LegacyAdapter、Fork/Report、CompactOldTurns、`/session` CLI コマンド、Mini App グラフ UI） |
 | [`todo/TASKS-4.md`](todo/TASKS-4.md) | ~~**Mini App & Static Serving**~~ ✅ 実装済み（`http.FileServer` 統合、テンプレート注入、Bun ビルド導線、frontend unit test + CI `pnpm test`） |
 | [`todo/TASKS-5.md`](todo/TASKS-5.md) | ~~**Heartbeat Worktree Management**~~ ✅ 実装済み（`/plan worktrees` の `list/inspect/merge/dispose`、安全化した `PruneOrphaned`、Mini App `/miniapp/api/worktrees` + Git タブ UI） |

pkg/agent/context_cache_test.go

@@ -12,70 +12,103 @@ import (
 )
 
 // setupWorkspace creates a temporary workspace with standard directories and optional files.
+
 // Returns the tmpDir path; caller should defer os.RemoveAll(tmpDir).
+
 func setupWorkspace(t *testing.T, files map[string]string) string {
 	t.Helper()
+
 	tmpDir, err := os.MkdirTemp("", "picoclaw-test-*")
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	os.MkdirAll(filepath.Join(tmpDir, "memory"), 0o755)
+
 	os.MkdirAll(filepath.Join(tmpDir, "skills"), 0o755)
+
 	for name, content := range files {
 		dir := filepath.Dir(filepath.Join(tmpDir, name))
+
 		os.MkdirAll(dir, 0o755)
+
 		if err := os.WriteFile(filepath.Join(tmpDir, name), []byte(content), 0o644); err != nil {
 			t.Fatal(err)
 		}
 	}
+
 	return tmpDir
 }
 
 // TestSingleSystemMessage verifies that BuildMessages always produces exactly one
+
 // system message regardless of summary/history variations.
+
 // Fix: multiple system messages break Anthropic (top-level system param) and
+
 // Codex (only reads last system message as instructions).
+
 func TestSingleSystemMessage(t *testing.T) {
 	tmpDir := setupWorkspace(t, map[string]string{
 		"IDENTITY.md": "# Identity\nTest agent.",
 	})
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	tests := []struct {
-		name    string
+		name string
+
 		history []providers.Message
+
 		summary string
+
 		message string
 	}{
 		{
-			name:    "no summary, no history",
+			name: "no summary, no history",
+
 			summary: "",
+
 			message: "hello",
 		},
+
 		{
-			name:    "with summary",
+			name: "with summary",
+
 			summary: "Previous conversation discussed X",
+
 			message: "hello",
 		},
+
 		{
 			name: "with history and summary",
+
 			history: []providers.Message{
 				{Role: "user", Content: "hi"},
+
 				{Role: "assistant", Content: "hello"},
 			},
+
 			summary: strings.Repeat("Long summary text. ", 50),
+
 			message: "new message",
 		},
+
 		{
 			name: "system message in history is filtered",
+
 			history: []providers.Message{
 				{Role: "system", Content: "stale system prompt from previous session"},
+
 				{Role: "user", Content: "hi"},
+
 				{Role: "assistant", Content: "hello"},
 			},
+
 			summary: "",
+
 			message: "new message",
 		},
 	}
@@ -85,35 +118,44 @@ func TestSingleSystemMessage(t *testing.T) {
 			msgs := cb.BuildMessages(tt.history, tt.summary, tt.message, nil, "test", "chat1")
 
 			systemCount := 0
+
 			for _, m := range msgs {
 				if m.Role == "system" {
 					systemCount++
 				}
 			}
+
 			if systemCount != 1 {
 				t.Errorf("expected exactly 1 system message, got %d", systemCount)
 			}
+
 			if msgs[0].Role != "system" {
 				t.Errorf("first message should be system, got %s", msgs[0].Role)
 			}
+
 			if msgs[len(msgs)-1].Role != "user" {
 				t.Errorf("last message should be user, got %s", msgs[len(msgs)-1].Role)
 			}
 
 			// System message must contain identity (static) and time (dynamic)
+
 			sys := msgs[0].Content
+
 			if !strings.Contains(sys, "picoclaw") {
 				t.Error("system message missing identity")
 			}
+
 			if !strings.Contains(sys, "Current Time") {
 				t.Error("system message missing dynamic time context")
 			}
 
 			// Summary handling
+
 			if tt.summary != "" {
 				if !strings.Contains(sys, "CONTEXT_SUMMARY:") {
 					t.Error("summary present but CONTEXT_SUMMARY prefix missing")
 				}
+
 				if !strings.Contains(sys, tt.summary[:20]) {
 					t.Error("summary content not found in system message")
 				}
@@ -127,29 +169,46 @@ func TestSingleSystemMessage(t *testing.T) {
 }
 
 // TestMtimeAutoInvalidation verifies that the cache detects source file changes
+
 // via mtime without requiring explicit InvalidateCache().
+
 // Fix: original implementation had no auto-invalidation — edits to bootstrap files,
+
 // memory, or skills were invisible until process restart.
+
 func TestMtimeAutoInvalidation(t *testing.T) {
 	tests := []struct {
-		name       string
-		file       string // relative path inside workspace
-		contentV1  string
-		contentV2  string
+		name string
+
+		file string // relative path inside workspace
+
+		contentV1 string
+
+		contentV2 string
+
 		checkField string // substring to verify in rebuilt prompt
 	}{
 		{
-			name:       "bootstrap file change",
-			file:       "IDENTITY.md",
-			contentV1:  "# Original Identity",
-			contentV2:  "# Updated Identity",
+			name: "bootstrap file change",
+
+			file: "IDENTITY.md",
+
+			contentV1: "# Original Identity",
+
+			contentV2: "# Updated Identity",
+
 			checkField: "Updated Identity",
 		},
+
 		{
-			name:       "memory file change",
-			file:       "memory/MEMORY.md",
-			contentV1:  "# Memory\nUser likes Go.",
-			contentV2:  "# Memory\nUser likes Rust.",
+			name: "memory file change",
+
+			file: "memory/MEMORY.md",
+
+			contentV1: "# Memory\nUser likes Go.",
+
+			contentV2: "# Memory\nUser likes Rust.",
+
 			checkField: "User likes Rust",
 		},
 	}
@@ -157,6 +216,7 @@ func TestMtimeAutoInvalidation(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			tmpDir := setupWorkspace(t, map[string]string{tt.file: tt.contentV1})
+
 			defer os.RemoveAll(tmpDir)
 
 			cb := NewContextBuilder(tmpDir)
@@ -164,26 +224,39 @@ func TestMtimeAutoInvalidation(t *testing.T) {
 			sp1 := cb.BuildSystemPromptWithCache()
 
 			// Overwrite file and set future mtime to ensure detection.
+
 			// Use 2s offset for filesystem mtime resolution safety (some FS
+
 			// have 1s or coarser granularity, especially in CI containers).
+
 			fullPath := filepath.Join(tmpDir, tt.file)
+
 			os.WriteFile(fullPath, []byte(tt.contentV2), 0o644)
+
 			future := time.Now().Add(2 * time.Second)
+
 			os.Chtimes(fullPath, future, future)
 
 			// Verify sourceFilesChangedLocked detects the mtime change
+
 			cb.systemPromptMutex.RLock()
+
 			changed := cb.sourceFilesChangedLocked()
+
 			cb.systemPromptMutex.RUnlock()
+
 			if !changed {
 				t.Fatalf("sourceFilesChangedLocked() should detect %s change", tt.file)
 			}
 
 			// Should auto-rebuild without explicit InvalidateCache()
+
 			sp2 := cb.BuildSystemPromptWithCache()
+
 			if sp1 == sp2 {
 				t.Errorf("cache not rebuilt after %s change", tt.file)
 			}
+
 			if !strings.Contains(sp2, tt.checkField) {
 				t.Errorf("rebuilt prompt missing expected content %q", tt.checkField)
 			}
@@ -191,23 +264,34 @@ func TestMtimeAutoInvalidation(t *testing.T) {
 	}
 
 	// Skills directory mtime change
+
 	t.Run("skills dir change", func(t *testing.T) {
 		tmpDir := setupWorkspace(t, nil)
+
 		defer os.RemoveAll(tmpDir)
 
 		cb := NewContextBuilder(tmpDir)
+
 		_ = cb.BuildSystemPromptWithCache() // populate cache
 
 		// Touch skills directory (simulate new skill installed)
+
 		skillsDir := filepath.Join(tmpDir, "skills")
+
 		future := time.Now().Add(2 * time.Second)
+
 		os.Chtimes(skillsDir, future, future)
 
 		// Verify sourceFilesChangedLocked detects it (cache is rebuilt)
+
 		// We confirm by checking internal state: a second call should rebuild.
+
 		cb.systemPromptMutex.RLock()
+
 		changed := cb.sourceFilesChangedLocked()
+
 		cb.systemPromptMutex.RUnlock()
+
 		if !changed {
 			t.Error("sourceFilesChangedLocked() should detect skills dir mtime change")
 		}
@@ -215,17 +299,22 @@ func TestMtimeAutoInvalidation(t *testing.T) {
 }
 
 // TestExplicitInvalidateCache verifies that InvalidateCache() forces a rebuild
+
 // even when source files haven't changed (useful for tests and reload commands).
+
 func TestExplicitInvalidateCache(t *testing.T) {
 	tmpDir := setupWorkspace(t, map[string]string{
 		"IDENTITY.md": "# Test Identity",
 	})
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	sp1 := cb.BuildSystemPromptWithCache()
+
 	cb.InvalidateCache()
+
 	sp2 := cb.BuildSystemPromptWithCache()
 
 	if sp1 != sp2 {
@@ -233,29 +322,39 @@ func TestExplicitInvalidateCache(t *testing.T) {
 	}
 
 	// Verify cachedAt was reset
+
 	cb.InvalidateCache()
+
 	cb.systemPromptMutex.RLock()
+
 	if !cb.cachedAt.IsZero() {
 		t.Error("cachedAt should be zero after InvalidateCache()")
 	}
+
 	cb.systemPromptMutex.RUnlock()
 }
 
 // TestCacheStability verifies that the static prompt is stable across repeated calls
+
 // when no files change (regression test for issue #607).
+
 func TestCacheStability(t *testing.T) {
 	tmpDir := setupWorkspace(t, map[string]string{
 		"IDENTITY.md": "# Identity\nContent",
-		"SOUL.md":     "# Soul\nContent",
+
+		"SOUL.md": "# Soul\nContent",
 	})
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	results := make([]string, 5)
+
 	for i := range results {
 		results[i] = cb.BuildSystemPromptWithCache()
 	}
+
 	for i := 1; i < len(results); i++ {
 		if results[i] != results[0] {
 			t.Errorf("cached prompt changed between call 0 and %d", i)
@@ -263,32 +362,47 @@ func TestCacheStability(t *testing.T) {
 	}
 
 	// Static prompt must NOT contain per-request data
+
 	if strings.Contains(results[0], "Current Time") {
 		t.Error("static cached prompt should not contain time (added dynamically)")
 	}
 }
 
 // TestNewFileCreationInvalidatesCache verifies that creating a source file that
+
 // did not exist when the cache was built triggers a cache rebuild.
+
 // This catches the "from nothing to something" edge case that the old
+
 // modifiedSince (return false on stat error) would miss.
+
 func TestNewFileCreationInvalidatesCache(t *testing.T) {
 	tests := []struct {
-		name       string
-		file       string // relative path inside workspace
-		content    string
+		name string
+
+		file string // relative path inside workspace
+
+		content string
+
 		checkField string // substring to verify in rebuilt prompt
 	}{
 		{
-			name:       "new bootstrap file",
-			file:       "SOUL.md",
-			content:    "# Soul\nBe kind and helpful.",
+			name: "new bootstrap file",
+
+			file: "SOUL.md",
+
+			content: "# Soul\nBe kind and helpful.",
+
 			checkField: "Be kind and helpful",
 		},
+
 		{
-			name:       "new memory file",
-			file:       "memory/MEMORY.md",
-			content:    "# Memory\nUser prefers dark mode.",
+			name: "new memory file",
+
+			file: "memory/MEMORY.md",
+
+			content: "# Memory\nUser prefers dark mode.",
+
 			checkField: "User prefers dark mode",
 		},
 	}
@@ -296,29 +410,41 @@ func TestNewFileCreationInvalidatesCache(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			// Start with an empty workspace (no bootstrap/memory files)
+
 			tmpDir := setupWorkspace(t, nil)
+
 			defer os.RemoveAll(tmpDir)
 
 			cb := NewContextBuilder(tmpDir)
 
 			// Populate cache — file does not exist yet
+
 			sp1 := cb.BuildSystemPromptWithCache()
+
 			if strings.Contains(sp1, tt.checkField) {
 				t.Fatalf("prompt should not contain %q before file is created", tt.checkField)
 			}
 
 			// Create the file after cache was built
+
 			fullPath := filepath.Join(tmpDir, tt.file)
+
 			os.MkdirAll(filepath.Dir(fullPath), 0o755)
+
 			if err := os.WriteFile(fullPath, []byte(tt.content), 0o644); err != nil {
 				t.Fatal(err)
 			}
+
 			// Set future mtime to guarantee detection
+
 			future := time.Now().Add(2 * time.Second)
+
 			os.Chtimes(fullPath, future, future)
 
 			// Cache should auto-invalidate because file went from absent -> present
+
 			sp2 := cb.BuildSystemPromptWithCache()
+
 			if !strings.Contains(sp2, tt.checkField) {
 				t.Errorf("cache not invalidated on new file creation: expected %q in prompt", tt.checkField)
 			}
@@ -327,110 +453,163 @@ func TestNewFileCreationInvalidatesCache(t *testing.T) {
 }
 
 // TestSkillFileContentChange verifies that modifying a skill file's content
+
 // (not just the directory structure) invalidates the cache.
+
 // This is the scenario where directory mtime alone is insufficient — on most
+
 // filesystems, editing a file inside a directory does NOT update the parent
+
 // directory's mtime.
+
 func TestSkillFileContentChange(t *testing.T) {
 	skillMD := `---
+
 name: test-skill
+
 description: "A test skill"
+
 ---
+
 # Test Skill v1
+
 Original content.`
 
 	tmpDir := setupWorkspace(t, map[string]string{
 		"skills/test-skill/SKILL.md": skillMD,
 	})
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	// Populate cache
+
 	sp1 := cb.BuildSystemPromptWithCache()
+
 	_ = sp1 // cache is warm
 
 	// Modify the skill file content (without touching the skills/ directory)
+
 	updatedSkillMD := `---
+
 name: test-skill
+
 description: "An updated test skill"
+
 ---
+
 # Test Skill v2
+
 Updated content.`
 
 	skillPath := filepath.Join(tmpDir, "skills", "test-skill", "SKILL.md")
+
 	if err := os.WriteFile(skillPath, []byte(updatedSkillMD), 0o644); err != nil {
 		t.Fatal(err)
 	}
+
 	// Set future mtime on the skill file only (NOT the directory)
+
 	future := time.Now().Add(2 * time.Second)
+
 	os.Chtimes(skillPath, future, future)
 
 	// Verify that sourceFilesChangedLocked detects the content change
+
 	cb.systemPromptMutex.RLock()
+
 	changed := cb.sourceFilesChangedLocked()
+
 	cb.systemPromptMutex.RUnlock()
+
 	if !changed {
 		t.Error("sourceFilesChangedLocked() should detect skill file content change")
 	}
 
 	// Verify cache is actually rebuilt with new content
+
 	sp2 := cb.BuildSystemPromptWithCache()
+
 	if sp1 == sp2 && strings.Contains(sp1, "test-skill") {
 		// If the skill appeared in the prompt and the prompt didn't change,
+
 		// the cache was not invalidated.
+
 		t.Error("cache should be invalidated when skill file content changes")
 	}
 }
 
 // TestConcurrentBuildSystemPromptWithCache verifies that multiple goroutines
+
 // can safely call BuildSystemPromptWithCache concurrently without producing
+
 // empty results, panics, or data races.
+
 // Run with: go test -race ./pkg/agent/ -run TestConcurrentBuildSystemPromptWithCache
+
 func TestConcurrentBuildSystemPromptWithCache(t *testing.T) {
 	tmpDir := setupWorkspace(t, map[string]string{
-		"IDENTITY.md":          "# Identity\nConcurrency test agent.",
-		"SOUL.md":              "# Soul\nBe helpful.",
-		"memory/MEMORY.md":     "# Memory\nUser prefers Go.",
+		"IDENTITY.md": "# Identity\nConcurrency test agent.",
+
+		"SOUL.md": "# Soul\nBe helpful.",
+
+		"memory/MEMORY.md": "# Memory\nUser prefers Go.",
+
 		"skills/demo/SKILL.md": "---\nname: demo\ndescription: \"demo skill\"\n---\n# Demo",
 	})
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	const goroutines = 20
+
 	const iterations = 50
 
 	var wg sync.WaitGroup
+
 	errs := make(chan string, goroutines*iterations)
 
 	for g := range goroutines {
 		wg.Add(1)
+
 		go func(id int) {
 			defer wg.Done()
+
 			for i := range iterations {
 				result := cb.BuildSystemPromptWithCache()
+
 				if result == "" {
 					errs <- "empty prompt returned"
+
 					return
 				}
+
 				if !strings.Contains(result, "picoclaw") {
 					errs <- "prompt missing identity"
+
 					return
 				}
 
 				// Also exercise BuildMessages concurrently
+
 				msgs := cb.BuildMessages(nil, "", "hello", nil, "test", "chat")
+
 				if len(msgs) < 2 {
 					errs <- "BuildMessages returned fewer than 2 messages"
+
 					return
 				}
+
 				if msgs[0].Role != "system" {
 					errs <- "first message not system"
+
 					return
 				}
 
 				// Occasionally invalidate to exercise the write path
+
 				if i%10 == 0 {
 					cb.InvalidateCache()
 				}
@@ -439,6 +618,7 @@ func TestConcurrentBuildSystemPromptWithCache(t *testing.T) {
 	}
 
 	wg.Wait()
+
 	close(errs)
 
 	for errMsg := range errs {
@@ -449,64 +629,90 @@ func TestConcurrentBuildSystemPromptWithCache(t *testing.T) {
 // BenchmarkBuildMessagesWithCache measures caching performance.
 
 // TestEmptyWorkspaceBaselineDetectsNewFiles verifies that when the cache is
+
 // built on an empty workspace (no tracked files exist), creating a file
+
 // afterwards still triggers cache invalidation. This validates the
+
 // time.Unix(1, 0) fallback for maxMtime: any real file's mtime is after epoch,
+
 // so fileChangedSince correctly detects the absent -> present transition AND
+
 // the mtime comparison succeeds even without artificially inflated Chtimes.
+
 func TestEmptyWorkspaceBaselineDetectsNewFiles(t *testing.T) {
 	// Empty workspace: no bootstrap files, no memory, no skills content.
+
 	tmpDir := setupWorkspace(t, nil)
+
 	defer os.RemoveAll(tmpDir)
 
 	cb := NewContextBuilder(tmpDir)
 
 	// Build cache — all tracked files are absent, maxMtime falls back to epoch.
+
 	sp1 := cb.BuildSystemPromptWithCache()
 
 	// Create a bootstrap file with natural mtime (no Chtimes manipulation).
+
 	// The file's mtime should be the current wall-clock time, which is
+
 	// strictly after time.Unix(1, 0).
+
 	soulPath := filepath.Join(tmpDir, "SOUL.md")
+
 	if err := os.WriteFile(soulPath, []byte("# Soul\nNewly created."), 0o644); err != nil {
 		t.Fatal(err)
 	}
 
 	// Cache should detect the new file via existedAtCache (absent -> present).
+
 	cb.systemPromptMutex.RLock()
+
 	changed := cb.sourceFilesChangedLocked()
+
 	cb.systemPromptMutex.RUnlock()
+
 	if !changed {
 		t.Fatal("sourceFilesChangedLocked should detect newly created file on empty workspace")
 	}
 
 	sp2 := cb.BuildSystemPromptWithCache()
+
 	if !strings.Contains(sp2, "Newly created") {
 		t.Error("rebuilt prompt should contain new file content")
 	}
+
 	if sp1 == sp2 {
 		t.Error("cache should have been invalidated after file creation")
 	}
 }
 
 // BenchmarkBuildMessagesWithCache measures caching performance.
+
 func BenchmarkBuildMessagesWithCache(b *testing.B) {
 	tmpDir, _ := os.MkdirTemp("", "picoclaw-bench-*")
+
 	defer os.RemoveAll(tmpDir)
 
 	os.MkdirAll(filepath.Join(tmpDir, "memory"), 0o755)
+
 	os.MkdirAll(filepath.Join(tmpDir, "skills"), 0o755)
+
 	for _, name := range []string{"IDENTITY.md", "SOUL.md", "USER.md"} {
 		os.WriteFile(filepath.Join(tmpDir, name), []byte(strings.Repeat("Content.\n", 10)), 0o644)
 	}
 
 	cb := NewContextBuilder(tmpDir)
+
 	history := []providers.Message{
 		{Role: "user", Content: "previous message"},
+
 		{Role: "assistant", Content: "previous response"},
 	}
 
 	b.ResetTimer()
+
 	for i := 0; i < b.N; i++ {
 		_ = cb.BuildMessages(history, "summary", "new message", nil, "cli", "test")
 	}

pkg/agent/context_test.go

@@ -12,9 +12,11 @@ func msg(role, content string) providers.Message {
 
 func assistantWithTools(toolIDs ...string) providers.Message {
 	calls := make([]providers.ToolCall, len(toolIDs))
+
 	for i, id := range toolIDs {
 		calls[i] = providers.ToolCall{ID: id, Type: "function"}
 	}
+
 	return providers.Message{Role: "assistant", ToolCalls: calls}
 }
 
@@ -24,11 +26,13 @@ func toolResult(id string) providers.Message {
 
 func TestSanitizeHistoryForProvider_EmptyHistory(t *testing.T) {
 	result := sanitizeHistoryForProvider(nil)
+
 	if len(result) != 0 {
 		t.Fatalf("expected empty, got %d messages", len(result))
 	}
 
 	result = sanitizeHistoryForProvider([]providers.Message{})
+
 	if len(result) != 0 {
 		t.Fatalf("expected empty, got %d messages", len(result))
 	}
@@ -37,170 +41,228 @@ func TestSanitizeHistoryForProvider_EmptyHistory(t *testing.T) {
 func TestSanitizeHistoryForProvider_SingleToolCall(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "hello"),
+
 		assistantWithTools("A"),
+
 		toolResult("A"),
+
 		msg("assistant", "done"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 4 {
 		t.Fatalf("expected 4 messages, got %d", len(result))
 	}
+
 	assertRoles(t, result, "user", "assistant", "tool", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_MultiToolCalls(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "do two things"),
+
 		assistantWithTools("A", "B"),
+
 		toolResult("A"),
+
 		toolResult("B"),
+
 		msg("assistant", "both done"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 5 {
 		t.Fatalf("expected 5 messages, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user", "assistant", "tool", "tool", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_AssistantToolCallAfterPlainAssistant(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "hi"),
+
 		msg("assistant", "thinking"),
+
 		assistantWithTools("A"),
+
 		toolResult("A"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 2 {
 		t.Fatalf("expected 2 messages, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_OrphanedLeadingTool(t *testing.T) {
 	history := []providers.Message{
 		toolResult("A"),
+
 		msg("user", "hello"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 1 {
 		t.Fatalf("expected 1 message, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user")
 }
 
 func TestSanitizeHistoryForProvider_ToolAfterUserDropped(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "hello"),
+
 		toolResult("A"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 1 {
 		t.Fatalf("expected 1 message, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user")
 }
 
 func TestSanitizeHistoryForProvider_ToolAfterAssistantNoToolCalls(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "hello"),
+
 		msg("assistant", "hi"),
+
 		toolResult("A"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 2 {
 		t.Fatalf("expected 2 messages, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_AssistantToolCallAtStart(t *testing.T) {
 	history := []providers.Message{
 		assistantWithTools("A"),
+
 		toolResult("A"),
+
 		msg("user", "hello"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 1 {
 		t.Fatalf("expected 1 message, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user")
 }
 
 func TestSanitizeHistoryForProvider_MultiToolCallsThenNewRound(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "do two things"),
+
 		assistantWithTools("A", "B"),
+
 		toolResult("A"),
+
 		toolResult("B"),
+
 		msg("assistant", "done"),
+
 		msg("user", "hi"),
+
 		assistantWithTools("C"),
+
 		toolResult("C"),
+
 		msg("assistant", "done again"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 9 {
 		t.Fatalf("expected 9 messages, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user", "assistant", "tool", "tool", "assistant", "user", "assistant", "tool", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_ConsecutiveMultiToolRounds(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "start"),
+
 		assistantWithTools("A", "B"),
+
 		toolResult("A"),
+
 		toolResult("B"),
+
 		assistantWithTools("C", "D"),
+
 		toolResult("C"),
+
 		toolResult("D"),
+
 		msg("assistant", "all done"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 8 {
 		t.Fatalf("expected 8 messages, got %d: %+v", len(result), roles(result))
 	}
+
 	assertRoles(t, result, "user", "assistant", "tool", "tool", "assistant", "tool", "tool", "assistant")
 }
 
 func TestSanitizeHistoryForProvider_PlainConversation(t *testing.T) {
 	history := []providers.Message{
 		msg("user", "hello"),
+
 		msg("assistant", "hi"),
+
 		msg("user", "how are you"),
+
 		msg("assistant", "fine"),
 	}
 
 	result := sanitizeHistoryForProvider(history)
+
 	if len(result) != 4 {
 		t.Fatalf("expected 4 messages, got %d", len(result))
 	}
+
 	assertRoles(t, result, "user", "assistant", "user", "assistant")
 }
 
 func roles(msgs []providers.Message) []string {
 	r := make([]string, len(msgs))
+
 	for i, m := range msgs {
 		r[i] = m.Role
 	}
+
 	return r
 }
 
 func assertRoles(t *testing.T, msgs []providers.Message, expected ...string) {
 	t.Helper()
+
 	if len(msgs) != len(expected) {
 		t.Fatalf("role count mismatch: got %v, want %v", roles(msgs), expected)
 	}
+
 	for i, exp := range expected {
 		if msgs[i].Role != exp {
 			t.Errorf("message[%d]: got role %q, want %q", i, msgs[i].Role, exp)

pkg/agent/instance_test.go

@@ -12,28 +12,35 @@ func TestNewAgentInstance_UsesDefaultsTemperatureAndMaxTokens(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
-				Workspace:         tmpDir,
-				Model:             "test-model",
-				MaxTokens:         1234,
+				Workspace: tmpDir,
+
+				Model: "test-model",
+
+				MaxTokens: 1234,
+
 				MaxToolIterations: 5,
 			},
 		},
 	}
 
 	configuredTemp := 1.0
+
 	cfg.Agents.Defaults.Temperature = &configuredTemp
 
 	provider := &mockProvider{}
+
 	agent := NewAgentInstance(nil, &cfg.Agents.Defaults, cfg, provider)
 
 	if agent.MaxTokens != 1234 {
 		t.Fatalf("MaxTokens = %d, want %d", agent.MaxTokens, 1234)
 	}
+
 	if agent.Temperature != 1.0 {
 		t.Fatalf("Temperature = %f, want %f", agent.Temperature, 1.0)
 	}
@@ -44,23 +51,29 @@ func TestNewAgentInstance_DefaultsTemperatureWhenZero(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
-				Workspace:         tmpDir,
-				Model:             "test-model",
-				MaxTokens:         1234,
+				Workspace: tmpDir,
+
+				Model: "test-model",
+
+				MaxTokens: 1234,
+
 				MaxToolIterations: 5,
 			},
 		},
 	}
 
 	configuredTemp := 0.0
+
 	cfg.Agents.Defaults.Temperature = &configuredTemp
 
 	provider := &mockProvider{}
+
 	agent := NewAgentInstance(nil, &cfg.Agents.Defaults, cfg, provider)
 
 	if agent.Temperature != 0.0 {
@@ -73,20 +86,25 @@ func TestNewAgentInstance_DefaultsTemperatureWhenUnset(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
-				Workspace:         tmpDir,
-				Model:             "test-model",
-				MaxTokens:         1234,
+				Workspace: tmpDir,
+
+				Model: "test-model",
+
+				MaxTokens: 1234,
+
 				MaxToolIterations: 5,
 			},
 		},
 	}
 
 	provider := &mockProvider{}
+
 	agent := NewAgentInstance(nil, &cfg.Agents.Defaults, cfg, provider)
 
 	if agent.Temperature != 0.7 {
@@ -99,33 +117,41 @@ func TestNewAgentInstance_ResolveCandidatesFromModelListAlias(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
 				Workspace: tmpDir,
-				Model:     "step-3.5-flash",
+
+				Model: "step-3.5-flash",
 			},
 		},
+
 		ModelList: []config.ModelConfig{
 			{
 				ModelName: "step-3.5-flash",
-				Model:     "openrouter/stepfun/step-3.5-flash:free",
-				APIBase:   "https://openrouter.ai/api/v1",
+
+				Model: "openrouter/stepfun/step-3.5-flash:free",
+
+				APIBase: "https://openrouter.ai/api/v1",
 			},
 		},
 	}
 
 	provider := &mockProvider{}
+
 	agent := NewAgentInstance(nil, &cfg.Agents.Defaults, cfg, provider)
 
 	if len(agent.Candidates) != 1 {
 		t.Fatalf("len(Candidates) = %d, want 1", len(agent.Candidates))
 	}
+
 	if agent.Candidates[0].Provider != "openrouter" {
 		t.Fatalf("candidate provider = %q, want %q", agent.Candidates[0].Provider, "openrouter")
 	}
+
 	if agent.Candidates[0].Model != "stepfun/step-3.5-flash:free" {
 		t.Fatalf("candidate model = %q, want %q", agent.Candidates[0].Model, "stepfun/step-3.5-flash:free")
 	}
@@ -136,33 +162,41 @@ func TestNewAgentInstance_ResolveCandidatesFromModelListAliasWithoutProtocol(t *
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
 				Workspace: tmpDir,
-				Model:     "glm-5",
+
+				Model: "glm-5",
 			},
 		},
+
 		ModelList: []config.ModelConfig{
 			{
 				ModelName: "glm-5",
-				Model:     "glm-5",
-				APIBase:   "https://api.z.ai/api/coding/paas/v4",
+
+				Model: "glm-5",
+
+				APIBase: "https://api.z.ai/api/coding/paas/v4",
 			},
 		},
 	}
 
 	provider := &mockProvider{}
+
 	agent := NewAgentInstance(nil, &cfg.Agents.Defaults, cfg, provider)
 
 	if len(agent.Candidates) != 1 {
 		t.Fatalf("len(Candidates) = %d, want 1", len(agent.Candidates))
 	}
+
 	if agent.Candidates[0].Provider != "openai" {
 		t.Fatalf("candidate provider = %q, want %q", agent.Candidates[0].Provider, "openai")
 	}
+
 	if agent.Candidates[0].Model != "glm-5" {
 		t.Fatalf("candidate model = %q, want %q", agent.Candidates[0].Model, "glm-5")
 	}

pkg/agent/loop_reporter_test.go

@@ -12,63 +12,93 @@ import (
 )
 
 // makeOrchTestLoop creates a minimal AgentLoop with a temp workspace and
+
 // a real Broadcaster wired as the reporter.
+
 // Returns the loop, the broadcaster, and a cleanup function.
+
 func makeOrchTestLoop(t *testing.T) (*AgentLoop, *orch.Broadcaster) {
 	t.Helper()
+
 	tmpDir, err := os.MkdirTemp("", "agent-orch-test-*")
 	if err != nil {
 		t.Fatalf("MkdirTemp: %v", err)
 	}
+
 	t.Cleanup(func() { os.RemoveAll(tmpDir) })
 
 	cfg := &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
-				Workspace:         tmpDir,
-				Model:             "test-model",
-				MaxTokens:         512,
+				Workspace: tmpDir,
+
+				Model: "test-model",
+
+				MaxTokens: 512,
+
 				MaxToolIterations: 5,
 			},
 		},
 	}
+
 	al := NewAgentLoop(cfg, bus.NewMessageBus(), &mockProvider{})
+
 	b := orch.NewBroadcaster()
+
 	al.SetOrchReporter(b)
+
 	return al, b
 }
 
 // collectOrchEvents drains the subscriber channel until an agent_gc event
+
 // arrives or the deadline is exceeded.
+
 func collectOrchEvents(t *testing.T, ch <-chan orch.Event, timeout time.Duration) []orch.Event {
 	t.Helper()
+
 	var events []orch.Event
+
 	deadline := time.After(timeout)
+
 	for {
 		select {
 		case ev := <-ch:
+
 			events = append(events, ev)
+
 			if ev.Type == "agent_gc" {
 				return events
 			}
+
 		case <-deadline:
+
 			t.Fatalf("timed out waiting for agent_gc; events so far: %+v", events)
 		}
 	}
 }
 
 // TestAgentLoop_ProcessDirect_EmitsSpawnWaitingGC verifies that a main
+
 // session processed via ProcessDirect emits the full lifecycle:
+
 //
+
 //	agent_spawn(sessionKey) → agent_state(waiting) → agent_gc(completed)
+
 //
+
 // and that the Broadcaster snapshot is empty after the call returns.
+
 func TestAgentLoop_ProcessDirect_EmitsSpawnWaitingGC(t *testing.T) {
 	al, b := makeOrchTestLoop(t)
+
 	sub := b.Subscribe()
+
 	defer b.Unsubscribe(sub)
 
 	const sessionKey = "orch-test-session"
+
 	_, err := al.ProcessDirect(context.Background(), "hello", sessionKey)
 	if err != nil {
 		t.Fatalf("ProcessDirect: %v", err)
@@ -77,39 +107,51 @@ func TestAgentLoop_ProcessDirect_EmitsSpawnWaitingGC(t *testing.T) {
 	events := collectOrchEvents(t, sub.Ch, 5*time.Second)
 
 	// First event: agent_spawn with correct ID.
+
 	if events[0].Type != "agent_spawn" || events[0].ID != sessionKey {
 		t.Errorf("first event must be agent_spawn(%s), got: %+v", sessionKey, events[0])
 	}
 
 	// At least one agent_state(waiting) for this session.
+
 	var hasWaiting bool
+
 	for _, ev := range events {
 		if ev.Type == "agent_state" && ev.ID == sessionKey && ev.State == "waiting" {
 			hasWaiting = true
+
 			break
 		}
 	}
+
 	if !hasWaiting {
 		t.Errorf("missing agent_state(waiting) for %s; events: %+v", sessionKey, events)
 	}
 
 	// Last event: agent_gc(completed) for this session.
+
 	last := events[len(events)-1]
+
 	if last.Type != "agent_gc" || last.ID != sessionKey || last.Reason != "completed" {
 		t.Errorf("last event must be agent_gc(completed,%s), got: %+v", sessionKey, last)
 	}
 
 	// Snapshot must be empty — session removed on GC.
+
 	if snap := b.Snapshot(); len(snap) != 0 {
 		t.Errorf("snapshot must be empty after GC, got: %v", snap)
 	}
 }
 
 // TestAgentLoop_ProcessHeartbeat_EmitsSpawnAndGC verifies that heartbeat
+
 // sessions appear on canvas with sessionKey = "heartbeat".
+
 func TestAgentLoop_ProcessHeartbeat_EmitsSpawnAndGC(t *testing.T) {
 	al, b := makeOrchTestLoop(t)
+
 	sub := b.Subscribe()
+
 	defer b.Unsubscribe(sub)
 
 	_, err := al.ProcessHeartbeat(context.Background(), "check system", "heartbeat-chan", "none")
@@ -120,12 +162,15 @@ func TestAgentLoop_ProcessHeartbeat_EmitsSpawnAndGC(t *testing.T) {
 	events := collectOrchEvents(t, sub.Ch, 5*time.Second)
 
 	// ProcessHeartbeat always uses sessionKey = "heartbeat".
+
 	const want = "heartbeat"
+
 	if events[0].Type != "agent_spawn" || events[0].ID != want {
 		t.Errorf("first event must be agent_spawn(%s), got: %+v", want, events[0])
 	}
 
 	last := events[len(events)-1]
+
 	if last.Type != "agent_gc" || last.ID != want || last.Reason != "completed" {
 		t.Errorf("last event must be agent_gc(completed,%s), got: %+v", want, last)
 	}

pkg/agent/memory_test.go

@@ -9,101 +9,172 @@ import (
 
 func newTestMemoryStore(t *testing.T) (*MemoryStore, func()) {
 	t.Helper()
+
 	tmpDir, err := os.MkdirTemp("", "memory-test-*")
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	ms := NewMemoryStore(tmpDir)
+
 	return ms, func() { os.RemoveAll(tmpDir) }
 }
 
 const testPlanInterviewing = `# Active Plan
 
+
+
 > Task: Set up server monitoring
+
 > Status: interviewing
+
 > Phase: 1
+
 `
 
 const testPlanExecuting = `# Active Plan
 
+
+
 > Task: Set up server monitoring
+
 > Status: executing
+
 > Phase: 2
 
+
+
 ## Phase 1: Prometheus Install
+
 - [x] Install Prometheus
+
 - [x] Configure node_exporter
 
+
+
 ## Phase 2: Grafana Setup
+
 - [ ] Install Grafana
+
 - [ ] Create dashboard
 
+
+
 ## Phase 3: Alert Configuration
+
 - [ ] Set up alert rules
+
 - [ ] Configure Telegram notifications
 
+
+
 ## Commands
+
 build: go build ./...
+
 test: go test ./pkg/... -count=1
+
 lint: golangci-lint run
 
+
+
 ## Context
+
 Pi: Debian Bookworm arm64, ports: 3000/9090
+
 `
 
 const testPlanPhase1Complete = `# Active Plan
 
+
+
 > Task: Set up server monitoring
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Prometheus Install
+
 - [x] Install Prometheus
+
 - [x] Configure node_exporter
 
+
+
 ## Phase 2: Grafana Setup
+
 - [ ] Install Grafana
+
 - [ ] Create dashboard
 
+
+
 ## Context
+
 Pi: Debian Bookworm arm64
+
 `
 
 const testPlanAllComplete = `# Active Plan
 
+
+
 > Task: Set up server monitoring
+
 > Status: executing
+
 > Phase: 2
 
+
+
 ## Phase 1: Prometheus Install
+
 - [x] Install Prometheus
+
 - [x] Configure node_exporter
 
+
+
 ## Phase 2: Grafana Setup
+
 - [x] Install Grafana
+
 - [x] Create dashboard
 
+
+
 ## Context
+
 Pi: Debian Bookworm arm64
+
 `
 
 func TestHasActivePlan(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// No plan
+
 	if ms.HasActivePlan() {
 		t.Error("expected no active plan for empty memory")
 	}
 
 	// With regular content
+
 	ms.WriteLongTerm("Some random notes")
+
 	if ms.HasActivePlan() {
 		t.Error("expected no active plan for regular content")
 	}
 
 	// With active plan
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if !ms.HasActivePlan() {
 		t.Error("expected active plan to be detected")
 	}
@@ -111,21 +182,27 @@ func TestHasActivePlan(t *testing.T) {
 
 func TestGetPlanStatus(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// No plan
+
 	if status := ms.GetPlanStatus(); status != "" {
 		t.Errorf("expected empty status, got %q", status)
 	}
 
 	// Interviewing
+
 	ms.WriteLongTerm(testPlanInterviewing)
+
 	if status := ms.GetPlanStatus(); status != "interviewing" {
 		t.Errorf("expected 'interviewing', got %q", status)
 	}
 
 	// Executing
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if status := ms.GetPlanStatus(); status != "executing" {
 		t.Errorf("expected 'executing', got %q", status)
 	}
@@ -133,15 +210,19 @@ func TestGetPlanStatus(t *testing.T) {
 
 func TestGetCurrentPhase(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// No plan
+
 	if phase := ms.GetCurrentPhase(); phase != 0 {
 		t.Errorf("expected phase 0, got %d", phase)
 	}
 
 	// Phase 2
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if phase := ms.GetCurrentPhase(); phase != 2 {
 		t.Errorf("expected phase 2, got %d", phase)
 	}
@@ -149,15 +230,19 @@ func TestGetCurrentPhase(t *testing.T) {
 
 func TestGetTotalPhases(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// No plan
+
 	if total := ms.GetTotalPhases(); total != 0 {
 		t.Errorf("expected 0 phases, got %d", total)
 	}
 
 	// 3 phases
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if total := ms.GetTotalPhases(); total != 3 {
 		t.Errorf("expected 3 phases, got %d", total)
 	}
@@ -165,22 +250,29 @@ func TestGetTotalPhases(t *testing.T) {
 
 func TestIsPlanComplete(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// Not complete
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if ms.IsPlanComplete() {
 		t.Error("expected plan to be incomplete")
 	}
 
 	// All complete
+
 	ms.WriteLongTerm(testPlanAllComplete)
+
 	if !ms.IsPlanComplete() {
 		t.Error("expected plan to be complete")
 	}
 
 	// No plan
+
 	ms.ClearLongTerm()
+
 	if ms.IsPlanComplete() {
 		t.Error("expected false when no plan exists")
 	}
@@ -188,16 +280,21 @@ func TestIsPlanComplete(t *testing.T) {
 
 func TestIsCurrentPhaseComplete(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// Phase 2 not complete
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if ms.IsCurrentPhaseComplete() {
 		t.Error("expected current phase to be incomplete")
 	}
 
 	// Phase 1 complete (current=1)
+
 	ms.WriteLongTerm(testPlanPhase1Complete)
+
 	if !ms.IsCurrentPhaseComplete() {
 		t.Error("expected phase 1 to be complete")
 	}
@@ -205,12 +302,15 @@ func TestIsCurrentPhaseComplete(t *testing.T) {
 
 func TestSetStatus(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanInterviewing)
+
 	if err := ms.SetStatus("executing"); err != nil {
 		t.Fatalf("SetStatus failed: %v", err)
 	}
+
 	if status := ms.GetPlanStatus(); status != "executing" {
 		t.Errorf("expected 'executing', got %q", status)
 	}
@@ -218,12 +318,15 @@ func TestSetStatus(t *testing.T) {
 
 func TestAdvancePhase(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanPhase1Complete)
+
 	if err := ms.AdvancePhase(); err != nil {
 		t.Fatalf("AdvancePhase failed: %v", err)
 	}
+
 	if phase := ms.GetCurrentPhase(); phase != 2 {
 		t.Errorf("expected phase 2 after advance, got %d", phase)
 	}
@@ -231,37 +334,49 @@ func TestAdvancePhase(t *testing.T) {
 
 func TestMarkStep(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanExecuting)
 
 	// Mark step 1 in phase 2
+
 	if err := ms.MarkStep(2, 1); err != nil {
 		t.Fatalf("MarkStep failed: %v", err)
 	}
 
 	content := ms.ReadLongTerm()
+
 	// Phase 2 should have first step checked
+
 	lines := strings.Split(content, "\n")
+
 	foundChecked := false
+
 	inPhase2 := false
+
 	for _, line := range lines {
 		if strings.HasPrefix(line, "## Phase 2:") {
 			inPhase2 = true
+
 			continue
 		}
+
 		if inPhase2 && strings.HasPrefix(line, "## ") {
 			break
 		}
+
 		if inPhase2 && strings.HasPrefix(line, "- [x] Install Grafana") {
 			foundChecked = true
 		}
 	}
+
 	if !foundChecked {
 		t.Error("expected 'Install Grafana' to be marked [x]")
 	}
 
 	// Error case: invalid step
+
 	if err := ms.MarkStep(2, 99); err == nil {
 		t.Error("expected error for invalid step number")
 	}
@@ -269,23 +384,29 @@ func TestMarkStep(t *testing.T) {
 
 func TestAddStep(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanExecuting)
 
 	// Add step to phase 2
+
 	if err := ms.AddStep(2, "Test dashboard"); err != nil {
 		t.Fatalf("AddStep failed: %v", err)
 	}
 
 	content := ms.ReadLongTerm()
+
 	if !strings.Contains(content, "- [ ] Test dashboard") {
 		t.Error("expected new step to be added")
 	}
 
 	// Verify it's in the right place (before Phase 3)
+
 	idx := strings.Index(content, "- [ ] Test dashboard")
+
 	phase3Idx := strings.Index(content, "## Phase 3:")
+
 	if idx > phase3Idx {
 		t.Error("expected new step to be before Phase 3")
 	}
@@ -293,17 +414,21 @@ func TestAddStep(t *testing.T) {
 
 func TestClearLongTerm(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanExecuting)
+
 	if err := ms.ClearLongTerm(); err != nil {
 		t.Fatalf("ClearLongTerm failed: %v", err)
 	}
+
 	if content := ms.ReadLongTerm(); content != "" {
 		t.Errorf("expected empty memory after clear, got %q", content)
 	}
 
 	// Clearing again should not error
+
 	if err := ms.ClearLongTerm(); err != nil {
 		t.Fatalf("ClearLongTerm (idempotent) failed: %v", err)
 	}
@@ -311,34 +436,45 @@ func TestClearLongTerm(t *testing.T) {
 
 func TestGetInterviewContext(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanInterviewing)
+
 	ctx := ms.GetInterviewContext()
 
 	if !strings.Contains(ctx, "Active Plan (interviewing)") {
 		t.Error("expected 'Active Plan (interviewing)' header")
 	}
+
 	if !strings.Contains(ctx, "Interview Guide") {
 		t.Error("expected 'Interview Guide' section")
 	}
+
 	if !strings.Contains(ctx, "Target Format") {
 		t.Error("expected 'Target Format' section")
 	}
+
 	if !strings.Contains(ctx, "Set up server monitoring") {
 		t.Error("expected task description in context")
 	}
+
 	// Should guide AI to ask about tooling
+
 	if !strings.Contains(ctx, "test framework") || !strings.Contains(ctx, "linter") {
 		t.Error("expected interview guide to mention test framework and linter")
 	}
+
 	// Target format should include Commands section example
+
 	if !strings.Contains(ctx, "## Commands") {
 		t.Error("expected target format to include ## Commands section")
 	}
+
 	if !strings.Contains(ctx, "project-specific test command") {
 		t.Error("expected target format Commands to include test command placeholder")
 	}
+
 	if !strings.Contains(ctx, "project-specific lint command") {
 		t.Error("expected target format Commands to include lint command placeholder")
 	}
@@ -346,46 +482,57 @@ func TestGetInterviewContext(t *testing.T) {
 
 func TestGetPlanContext(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanExecuting)
+
 	ctx := ms.GetPlanContext()
 
 	// Should have task summary
+
 	if !strings.Contains(ctx, "Phase 2/3") {
 		t.Error("expected 'Phase 2/3' in plan context")
 	}
 
 	// Completed phase should be summarized
+
 	if !strings.Contains(ctx, "Done: Phase 1") {
 		t.Error("expected completed phase summary")
 	}
 
 	// Current phase should have full detail
+
 	if !strings.Contains(ctx, "Current: Phase 2") {
 		t.Error("expected current phase detail")
 	}
+
 	if !strings.Contains(ctx, "Install Grafana") {
 		t.Error("expected current phase steps")
 	}
 
 	// Future phases should NOT appear
+
 	if strings.Contains(ctx, "Phase 3") {
 		t.Error("expected future phases to be omitted")
 	}
 
 	// Commands should be included
+
 	if !strings.Contains(ctx, "### Commands") {
 		t.Error("expected Commands section in plan context")
 	}
+
 	if !strings.Contains(ctx, "go test") {
 		t.Error("expected test command in Commands section")
 	}
+
 	if !strings.Contains(ctx, "golangci-lint") {
 		t.Error("expected lint command in Commands section")
 	}
 
 	// Context should be included
+
 	if !strings.Contains(ctx, "Debian Bookworm") {
 		t.Error("expected Context section")
 	}
@@ -393,20 +540,27 @@ func TestGetPlanContext(t *testing.T) {
 
 func TestGetMemoryContext_PlanActive_SuppressesDailyNotes(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// Write a daily note
+
 	ms.AppendToday("Today's note")
 
 	// Without plan, daily notes should appear
+
 	ctx := ms.GetMemoryContext()
+
 	if !strings.Contains(ctx, "Recent Daily Notes") {
 		t.Error("expected daily notes when no plan active")
 	}
 
 	// With plan, daily notes should be suppressed
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	ctx = ms.GetMemoryContext()
+
 	if strings.Contains(ctx, "Recent Daily Notes") {
 		t.Error("expected daily notes to be suppressed when plan is active")
 	}
@@ -414,14 +568,17 @@ func TestGetMemoryContext_PlanActive_SuppressesDailyNotes(t *testing.T) {
 
 func TestGetMemoryContext_InterviewingMode(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanInterviewing)
+
 	ctx := ms.GetMemoryContext()
 
 	if !strings.Contains(ctx, "interviewing") {
 		t.Error("expected interviewing context")
 	}
+
 	if !strings.Contains(ctx, "Interview Guide") {
 		t.Error("expected interview guide in context")
 	}
@@ -429,14 +586,17 @@ func TestGetMemoryContext_InterviewingMode(t *testing.T) {
 
 func TestGetMemoryContext_ExecutingMode(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm(testPlanExecuting)
+
 	ctx := ms.GetMemoryContext()
 
 	if !strings.Contains(ctx, "Active Plan") {
 		t.Error("expected active plan in context")
 	}
+
 	if !strings.Contains(ctx, "Current: Phase 2") {
 		t.Error("expected current phase in context")
 	}
@@ -444,14 +604,17 @@ func TestGetMemoryContext_ExecutingMode(t *testing.T) {
 
 func TestGetMemoryContext_RegularMemory(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	ms.WriteLongTerm("Some notes about projects")
+
 	ctx := ms.GetMemoryContext()
 
 	if !strings.Contains(ctx, "Long-term Memory") {
 		t.Error("expected regular long-term memory section")
 	}
+
 	if !strings.Contains(ctx, "Some notes about projects") {
 		t.Error("expected memory content")
 	}
@@ -463,12 +626,15 @@ func TestBuildInterviewSeed(t *testing.T) {
 	if !strings.Contains(seed, "# Active Plan") {
 		t.Error("expected '# Active Plan' header")
 	}
+
 	if !strings.Contains(seed, "Deploy monitoring stack") {
 		t.Error("expected task description")
 	}
+
 	if !strings.Contains(seed, "interviewing") {
 		t.Error("expected interviewing status")
 	}
+
 	if !strings.Contains(seed, "> Phase: 1") {
 		t.Error("expected Phase: 1")
 	}
@@ -476,28 +642,37 @@ func TestBuildInterviewSeed(t *testing.T) {
 
 func TestFormatPlanDisplay(t *testing.T) {
 	ms, cleanup := newTestMemoryStore(t)
+
 	defer cleanup()
 
 	// No plan
+
 	display := ms.FormatPlanDisplay()
+
 	if display != "No active plan." {
 		t.Errorf("expected 'No active plan.', got %q", display)
 	}
 
 	// With plan
+
 	ms.WriteLongTerm(testPlanExecuting)
+
 	display = ms.FormatPlanDisplay()
 
 	if !strings.Contains(display, "Set up server monitoring") {
 		t.Error("expected task name in display")
 	}
+
 	if !strings.Contains(display, "Phase 2/3") {
 		t.Error("expected phase count in display")
 	}
+
 	// Commands section should be visible
+
 	if !strings.Contains(display, "Commands:") {
 		t.Error("expected Commands section in display")
 	}
+
 	if !strings.Contains(display, "go test") {
 		t.Error("expected test command in display")
 	}
@@ -505,118 +680,211 @@ func TestFormatPlanDisplay(t *testing.T) {
 
 func TestValidatePlanStructure(t *testing.T) {
 	tests := []struct {
-		name    string
+		name string
+
 		content string
+
 		wantErr string // "" means nil error expected
 	}{
 		{
 			name: "valid plan with 1 phase and 1 step",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 - [ ] Install deps
+
 `,
+
 			wantErr: "",
 		},
+
 		{
-			name:    "missing Active Plan header",
+			name: "missing Active Plan header",
+
 			content: `> Status: executing`,
+
 			wantErr: "missing '# Active Plan' header",
 		},
+
 		{
 			name: "missing Status line",
+
 			content: `# Active Plan
 
+
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 - [ ] Install deps
+
 `,
+
 			wantErr: "missing '> Status:' line",
 		},
+
 		{
 			name: "missing Phase line",
+
 			content: `# Active Plan
 
+
+
 > Status: executing
 
+
+
 ## Phase 1: Setup
+
 - [ ] Install deps
+
 `,
+
 			wantErr: "missing '> Phase:' line",
 		},
+
 		{
 			name: "no Phase sections",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
+
 `,
+
 			wantErr: "no '## Phase N:' sections found",
 		},
+
 		{
 			name: "phase with no checkbox steps",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 Some description without checkboxes
+
 `,
+
 			wantErr: "Phase 1 has no checkbox steps",
 		},
+
 		{
 			name: "all steps done is valid",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 - [x] Install deps
+
 - [x] Configure
+
 `,
+
 			wantErr: "",
 		},
+
 		{
 			name: "multi-phase valid",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 - [ ] Install deps
 
+
+
 ## Phase 2: Build
+
 - [ ] Compile
+
 - [ ] Test
+
 `,
+
 			wantErr: "",
 		},
+
 		{
 			name: "second phase empty steps",
+
 			content: `# Active Plan
 
+
+
 > Task: Do something
+
 > Status: executing
+
 > Phase: 1
 
+
+
 ## Phase 1: Setup
+
 - [ ] Install deps
 
+
+
 ## Phase 2: Build
+
 No checkboxes here
+
 `,
+
 			wantErr: "Phase 2 has no checkbox steps",
 		},
 	}
@@ -624,9 +892,11 @@ No checkboxes here
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			ms, cleanup := newTestMemoryStore(t)
+
 			defer cleanup()
 
 			ms.WriteLongTerm(tt.content)
+
 			err := ms.ValidatePlanStructure()
 
 			if tt.wantErr == "" {
@@ -649,18 +919,23 @@ func TestMemoryStoreCreation(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Failed to create temp dir: %v", err)
 	}
+
 	defer os.RemoveAll(tmpDir)
 
 	ms := NewMemoryStore(tmpDir)
 
 	// Verify memory directory was created
+
 	memoryDir := filepath.Join(tmpDir, "memory")
+
 	if _, err := os.Stat(memoryDir); os.IsNotExist(err) {
 		t.Error("expected memory directory to be created")
 	}
 
 	// Verify memory file path
+
 	expectedFile := filepath.Join(memoryDir, "MEMORY.md")
+
 	if ms.memoryFile != expectedFile {
 		t.Errorf("expected memory file %q, got %q", expectedFile, ms.memoryFile)
 	}

pkg/agent/mock_provider_test.go

@@ -10,13 +10,18 @@ type mockProvider struct{}
 
 func (m *mockProvider) Chat(
 	ctx context.Context,
+
 	messages []providers.Message,
+
 	tools []providers.ToolDefinition,
+
 	model string,
+
 	opts map[string]any,
 ) (*providers.LLMResponse, error) {
 	return &providers.LLMResponse{
-		Content:   "Mock response",
+		Content: "Mock response",
+
 		ToolCalls: []providers.ToolCall{},
 	}, nil
 }

pkg/agent/registry.go

@@ -10,43 +10,62 @@ import (
 )
 
 // AgentRegistry manages multiple agent instances and routes messages to them.
+
 type AgentRegistry struct {
-	agents   map[string]*AgentInstance
+	agents map[string]*AgentInstance
+
 	resolver *routing.RouteResolver
-	mu       sync.RWMutex
+
+	mu sync.RWMutex
 }
 
 // NewAgentRegistry creates a registry from config, instantiating all agents.
+
 func NewAgentRegistry(
 	cfg *config.Config,
+
 	provider providers.LLMProvider,
 ) *AgentRegistry {
 	registry := &AgentRegistry{
-		agents:   make(map[string]*AgentInstance),
+		agents: make(map[string]*AgentInstance),
+
 		resolver: routing.NewRouteResolver(cfg),
 	}
 
 	agentConfigs := cfg.Agents.List
+
 	if len(agentConfigs) == 0 {
 		implicitAgent := &config.AgentConfig{
-			ID:      "main",
+			ID: "main",
+
 			Default: true,
 		}
+
 		instance := NewAgentInstance(implicitAgent, &cfg.Agents.Defaults, cfg, provider)
+
 		registry.agents["main"] = instance
+
 		logger.InfoCF("agent", "Created implicit main agent (no agents.list configured)", nil)
 	} else {
 		for i := range agentConfigs {
 			ac := &agentConfigs[i]
+
 			id := routing.NormalizeAgentID(ac.ID)
+
 			instance := NewAgentInstance(ac, &cfg.Agents.Defaults, cfg, provider)
+
 			registry.agents[id] = instance
+
 			logger.InfoCF("agent", "Registered agent",
+
 				map[string]any{
-					"agent_id":  id,
-					"name":      ac.Name,
+					"agent_id": id,
+
+					"name": ac.Name,
+
 					"workspace": instance.Workspace,
-					"model":     instance.Model,
+
+					"model": instance.Model,
 				})
 		}
 	}
@@ -55,60 +74,83 @@ func NewAgentRegistry(
 }
 
 // GetAgent returns the agent instance for a given ID.
+
 func (r *AgentRegistry) GetAgent(agentID string) (*AgentInstance, bool) {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
+
 	id := routing.NormalizeAgentID(agentID)
+
 	agent, ok := r.agents[id]
+
 	return agent, ok
 }
 
 // ResolveRoute determines which agent handles the message.
+
 func (r *AgentRegistry) ResolveRoute(input routing.RouteInput) routing.ResolvedRoute {
 	return r.resolver.ResolveRoute(input)
 }
 
 // ListAgentIDs returns all registered agent IDs.
+
 func (r *AgentRegistry) ListAgentIDs() []string {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
+
 	ids := make([]string, 0, len(r.agents))
+
 	for id := range r.agents {
 		ids = append(ids, id)
 	}
+
 	return ids
 }
 
 // CanSpawnSubagent checks if parentAgentID is allowed to spawn targetAgentID.
+
 func (r *AgentRegistry) CanSpawnSubagent(parentAgentID, targetAgentID string) bool {
 	parent, ok := r.GetAgent(parentAgentID)
+
 	if !ok {
 		return false
 	}
+
 	if parent.Subagents == nil || parent.Subagents.AllowAgents == nil {
 		return false
 	}
+
 	targetNorm := routing.NormalizeAgentID(targetAgentID)
+
 	for _, allowed := range parent.Subagents.AllowAgents {
 		if allowed == "*" {
 			return true
 		}
+
 		if routing.NormalizeAgentID(allowed) == targetNorm {
 			return true
 		}
 	}
+
 	return false
 }
 
 // GetDefaultAgent returns the default agent instance.
+
 func (r *AgentRegistry) GetDefaultAgent() *AgentInstance {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
+
 	if agent, ok := r.agents["main"]; ok {
 		return agent
 	}
+
 	for _, agent := range r.agents {
 		return agent
 	}
+
 	return nil
 }

pkg/agent/registry_test.go

@@ -12,9 +12,13 @@ type mockRegistryProvider struct{}
 
 func (m *mockRegistryProvider) Chat(
 	ctx context.Context,
+
 	messages []providers.Message,
+
 	tools []providers.ToolDefinition,
+
 	model string,
+
 	options map[string]any,
 ) (*providers.LLMResponse, error) {
 	return &providers.LLMResponse{Content: "mock", FinishReason: "stop"}, nil
@@ -28,11 +32,15 @@ func testCfg(agents []config.AgentConfig) *config.Config {
 	return &config.Config{
 		Agents: config.AgentsConfig{
 			Defaults: config.AgentDefaults{
-				Workspace:         "/tmp/picoclaw-test-registry",
-				Model:             "gpt-4",
-				MaxTokens:         8192,
+				Workspace: "/tmp/picoclaw-test-registry",
+
+				Model: "gpt-4",
+
+				MaxTokens: 8192,
+
 				MaxToolIterations: 10,
 			},
+
 			List: agents,
 		},
 	}
@@ -40,17 +48,21 @@ func testCfg(agents []config.AgentConfig) *config.Config {
 
 func TestNewAgentRegistry_ImplicitMain(t *testing.T) {
 	cfg := testCfg(nil)
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	ids := registry.ListAgentIDs()
+
 	if len(ids) != 1 || ids[0] != "main" {
 		t.Errorf("expected implicit main agent, got %v", ids)
 	}
 
 	agent, ok := registry.GetAgent("main")
+
 	if !ok || agent == nil {
 		t.Fatal("expected to find 'main' agent")
 	}
+
 	if agent.ID != "main" {
 		t.Errorf("agent.ID = %q, want 'main'", agent.ID)
 	}
@@ -59,24 +71,30 @@ func TestNewAgentRegistry_ImplicitMain(t *testing.T) {
 func TestNewAgentRegistry_ExplicitAgents(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "sales", Default: true, Name: "Sales Bot"},
+
 		{ID: "support", Name: "Support Bot"},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	ids := registry.ListAgentIDs()
+
 	if len(ids) != 2 {
 		t.Fatalf("expected 2 agents, got %d: %v", len(ids), ids)
 	}
 
 	sales, ok := registry.GetAgent("sales")
+
 	if !ok || sales == nil {
 		t.Fatal("expected to find 'sales' agent")
 	}
+
 	if sales.Name != "Sales Bot" {
 		t.Errorf("sales.Name = %q, want 'Sales Bot'", sales.Name)
 	}
 
 	support, ok := registry.GetAgent("support")
+
 	if !ok || support == nil {
 		t.Fatal("expected to find 'support' agent")
 	}
@@ -86,12 +104,15 @@ func TestAgentRegistry_GetAgent_Normalize(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "my-agent", Default: true},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	agent, ok := registry.GetAgent("My-Agent")
+
 	if !ok || agent == nil {
 		t.Fatal("expected to find agent with normalized ID")
 	}
+
 	if agent.ID != "my-agent" {
 		t.Errorf("agent.ID = %q, want 'my-agent'", agent.ID)
 	}
@@ -100,12 +121,16 @@ func TestAgentRegistry_GetAgent_Normalize(t *testing.T) {
 func TestAgentRegistry_GetDefaultAgent(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "alpha"},
+
 		{ID: "beta", Default: true},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	// GetDefaultAgent first checks for "main", then returns any
+
 	agent := registry.GetDefaultAgent()
+
 	if agent == nil {
 		t.Fatal("expected a default agent")
 	}
@@ -114,27 +139,36 @@ func TestAgentRegistry_GetDefaultAgent(t *testing.T) {
 func TestAgentRegistry_CanSpawnSubagent(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{
-			ID:      "parent",
+			ID: "parent",
+
 			Default: true,
+
 			Subagents: &config.SubagentsConfig{
 				AllowAgents: []string{"child1", "child2"},
 			},
 		},
+
 		{ID: "child1"},
+
 		{ID: "child2"},
+
 		{ID: "restricted"},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	if !registry.CanSpawnSubagent("parent", "child1") {
 		t.Error("expected parent to be allowed to spawn child1")
 	}
+
 	if !registry.CanSpawnSubagent("parent", "child2") {
 		t.Error("expected parent to be allowed to spawn child2")
 	}
+
 	if registry.CanSpawnSubagent("parent", "restricted") {
 		t.Error("expected parent to NOT be allowed to spawn restricted")
 	}
+
 	if registry.CanSpawnSubagent("child1", "child2") {
 		t.Error("expected child1 to NOT be allowed to spawn (no subagents config)")
 	}
@@ -143,19 +177,24 @@ func TestAgentRegistry_CanSpawnSubagent(t *testing.T) {
 func TestAgentRegistry_CanSpawnSubagent_Wildcard(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{
-			ID:      "admin",
+			ID: "admin",
+
 			Default: true,
+
 			Subagents: &config.SubagentsConfig{
 				AllowAgents: []string{"*"},
 			},
 		},
+
 		{ID: "any-agent"},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	if !registry.CanSpawnSubagent("admin", "any-agent") {
 		t.Error("expected wildcard to allow spawning any agent")
 	}
+
 	if !registry.CanSpawnSubagent("admin", "nonexistent") {
 		t.Error("expected wildcard to allow spawning even nonexistent agents")
 	}
@@ -163,12 +202,15 @@ func TestAgentRegistry_CanSpawnSubagent_Wildcard(t *testing.T) {
 
 func TestAgentInstance_Model(t *testing.T) {
 	model := &config.AgentModelConfig{Primary: "claude-opus"}
+
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "custom", Default: true, Model: model},
 	})
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	agent, _ := registry.GetAgent("custom")
+
 	if agent.Model != "claude-opus" {
 		t.Errorf("agent.Model = %q, want 'claude-opus'", agent.Model)
 	}
@@ -178,10 +220,13 @@ func TestAgentInstance_FallbackInheritance(t *testing.T) {
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "inherit", Default: true},
 	})
+
 	cfg.Agents.Defaults.ModelFallbacks = []string{"openai/gpt-4o-mini", "anthropic/haiku"}
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	agent, _ := registry.GetAgent("inherit")
+
 	if len(agent.Fallbacks) != 2 {
 		t.Errorf("expected 2 fallbacks inherited from defaults, got %d", len(agent.Fallbacks))
 	}
@@ -189,16 +234,22 @@ func TestAgentInstance_FallbackInheritance(t *testing.T) {
 
 func TestAgentInstance_FallbackExplicitEmpty(t *testing.T) {
 	model := &config.AgentModelConfig{
-		Primary:   "gpt-4",
+		Primary: "gpt-4",
+
 		Fallbacks: []string{}, // explicitly empty = disable
+
 	}
+
 	cfg := testCfg([]config.AgentConfig{
 		{ID: "no-fallback", Default: true, Model: model},
 	})
+
 	cfg.Agents.Defaults.ModelFallbacks = []string{"should-not-inherit"}
+
 	registry := NewAgentRegistry(cfg, &mockRegistryProvider{})
 
 	agent, _ := registry.GetAgent("no-fallback")
+
 	if len(agent.Fallbacks) != 0 {
 		t.Errorf("expected 0 fallbacks (explicit empty), got %d: %v", len(agent.Fallbacks), agent.Fallbacks)
 	}

pkg/agent/session_recorder.go

@@ -7,6 +7,7 @@ import (
 )
 
 // sessionRecorderImpl bridges tools.SessionRecorder → session.SessionStore.
+
 type sessionRecorderImpl struct {
 	adapter *session.LegacyAdapter
 }
@@ -19,40 +20,102 @@ func newSessionRecorder(adapter *session.LegacyAdapter) *sessionRecorderImpl {
 
 func (r *sessionRecorderImpl) RecordFork(conductorKey, subagentKey, taskID, label string) error {
 	store := r.adapter.Store()
+
 	return store.Fork(conductorKey, subagentKey, &session.CreateOpts{
 		ForkTurnID: taskID,
-		Label:      label,
+
+		Label: label,
 	})
 }
 
 func (r *sessionRecorderImpl) RecordSubagentTurn(subagentKey string, messages []providers.Message) error {
 	store := r.adapter.Store()
+
 	turn := &session.Turn{
-		Kind:     session.TurnNormal,
+		Kind: session.TurnNormal,
+
 		Messages: messages,
 	}
+
 	return store.Append(subagentKey, turn)
 }
 
 func (r *sessionRecorderImpl) RecordCompletion(subagentKey, status, result string) error {
 	store := r.adapter.Store()
+
 	return store.SetStatus(subagentKey, status)
 }
 
 func (r *sessionRecorderImpl) RecordReport(conductorKey, subagentKey, senderID, content string) error {
 	store := r.adapter.Store()
+
 	turn := &session.Turn{
-		Kind:      session.TurnReport,
+		Kind: session.TurnReport,
+
 		OriginKey: subagentKey,
-		Author:    senderID,
+
+		Author: senderID,
+
 		Messages: []providers.Message{
 			{Role: "user", Content: content},
 		},
 	}
+
 	if err := store.Append(conductorKey, turn); err != nil {
 		return err
 	}
+
 	// Advance LegacyAdapter's stored counter so flush loop doesn't double-write.
+
 	r.adapter.AdvanceStored(conductorKey, 1)
+
+	return nil
+}
+
+func (r *sessionRecorderImpl) RecordQuestion(conductorKey, subagentKey, taskID, question string) error {
+	store := r.adapter.Store()
+
+	turn := &session.Turn{
+		Kind: session.TurnQuestion,
+
+		OriginKey: subagentKey,
+
+		Author: taskID,
+
+		Messages: []providers.Message{
+			{Role: "user", Content: question},
+		},
+	}
+
+	if err := store.Append(conductorKey, turn); err != nil {
+		return err
+	}
+
+	r.adapter.AdvanceStored(conductorKey, 1)
+
+	return nil
+}
+
+func (r *sessionRecorderImpl) RecordPlanSubmit(conductorKey, subagentKey, taskID, planText string) error {
+	store := r.adapter.Store()
+
+	turn := &session.Turn{
+		Kind: session.TurnPlanSubmit,
+
+		OriginKey: subagentKey,
+
+		Author: taskID,
+
+		Messages: []providers.Message{
+			{Role: "user", Content: planText},
+		},
+	}
+
+	if err := store.Append(conductorKey, turn); err != nil {
+		return err
+	}
+
+	r.adapter.AdvanceStored(conductorKey, 1)
+
 	return nil
 }

pkg/agent/session_recorder_test.go

@@ -11,15 +11,22 @@ import (
 
 func newTestRecorder(t *testing.T) (*sessionRecorderImpl, *session.LegacyAdapter, session.SessionStore) {
 	t.Helper()
+
 	tmpDir := t.TempDir()
+
 	dbPath := filepath.Join(tmpDir, "test.db")
+
 	store, err := session.OpenSQLiteStore(dbPath)
 	if err != nil {
 		t.Fatalf("open store: %v", err)
 	}
+
 	adapter := session.NewLegacyAdapter(store)
+
 	t.Cleanup(func() { adapter.Close() })
+
 	recorder := newSessionRecorder(adapter)
+
 	return recorder, adapter, store
 }
 
@@ -27,6 +34,7 @@ func TestRecordFork(t *testing.T) {
 	rec, _, store := newTestRecorder(t)
 
 	// Create conductor session first.
+
 	if err := store.Create("conductor:main", nil); err != nil {
 		t.Fatalf("create conductor session: %v", err)
 	}
@@ -37,31 +45,39 @@ func TestRecordFork(t *testing.T) {
 	}
 
 	// Verify child session exists with correct parent.
+
 	info, err := store.Get("subagent:subagent-1")
 	if err != nil {
 		t.Fatalf("Get child: %v", err)
 	}
+
 	if info == nil {
 		t.Fatal("child session not found")
 	}
+
 	if info.ParentKey != "conductor:main" {
 		t.Errorf("ParentKey = %q, want %q", info.ParentKey, "conductor:main")
 	}
+
 	if info.ForkTurnID != "subagent-1" {
 		t.Errorf("ForkTurnID = %q, want %q", info.ForkTurnID, "subagent-1")
 	}
+
 	if info.Label != "scout" {
 		t.Errorf("Label = %q, want %q", info.Label, "scout")
 	}
 
 	// Verify parent lists child.
+
 	children, err := store.Children("conductor:main")
 	if err != nil {
 		t.Fatalf("Children: %v", err)
 	}
+
 	if len(children) != 1 {
 		t.Fatalf("children count = %d, want 1", len(children))
 	}
+
 	if children[0].Key != "subagent:subagent-1" {
 		t.Errorf("child key = %q, want %q", children[0].Key, "subagent:subagent-1")
 	}
@@ -71,15 +87,19 @@ func TestRecordSubagentTurn(t *testing.T) {
 	rec, _, store := newTestRecorder(t)
 
 	// Create subagent session.
+
 	if err := store.Create("subagent:subagent-1", nil); err != nil {
 		t.Fatalf("create: %v", err)
 	}
 
 	msgs := []providers.Message{
 		{Role: "system", Content: "You are a scout."},
+
 		{Role: "user", Content: "Investigate X."},
+
 		{Role: "assistant", Content: "Found Y."},
 	}
+
 	if err := rec.RecordSubagentTurn("subagent:subagent-1", msgs); err != nil {
 		t.Fatalf("RecordSubagentTurn: %v", err)
 	}
@@ -88,15 +108,19 @@ func TestRecordSubagentTurn(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Turns: %v", err)
 	}
+
 	if len(turns) != 1 {
 		t.Fatalf("turns count = %d, want 1", len(turns))
 	}
+
 	if turns[0].Kind != session.TurnNormal {
 		t.Errorf("Kind = %d, want TurnNormal", turns[0].Kind)
 	}
+
 	if len(turns[0].Messages) != 3 {
 		t.Errorf("messages count = %d, want 3", len(turns[0].Messages))
 	}
+
 	if turns[0].Messages[2].Content != "Found Y." {
 		t.Errorf("last message = %q, want %q", turns[0].Messages[2].Content, "Found Y.")
 	}
@@ -117,18 +141,23 @@ func TestRecordCompletion(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Get: %v", err)
 	}
+
 	if info.Status != "completed" {
 		t.Errorf("Status = %q, want %q", info.Status, "completed")
 	}
 
 	// Test failed status.
+
 	if err := store.Create("subagent:subagent-2", nil); err != nil {
 		t.Fatalf("create: %v", err)
 	}
+
 	if err := rec.RecordCompletion("subagent:subagent-2", "failed", "error"); err != nil {
 		t.Fatalf("RecordCompletion failed: %v", err)
 	}
+
 	info2, _ := store.Get("subagent:subagent-2")
+
 	if info2.Status != "failed" {
 		t.Errorf("Status = %q, want %q", info2.Status, "failed")
 	}
@@ -138,6 +167,7 @@ func TestRecordReport(t *testing.T) {
 	rec, adapter, store := newTestRecorder(t)
 
 	// Create conductor session via adapter so it's in cache.
+
 	_ = adapter.GetOrCreate("conductor:main")
 
 	if err := store.Create("subagent:subagent-1", nil); err != nil {
@@ -145,27 +175,34 @@ func TestRecordReport(t *testing.T) {
 	}
 
 	content := "[System: subagent:subagent-1] Task 'scout' completed.\n\nResult:\nFound Y."
+
 	if err := rec.RecordReport("conductor:main", "subagent:subagent-1", "subagent:subagent-1", content); err != nil {
 		t.Fatalf("RecordReport: %v", err)
 	}
 
 	// Verify TurnReport in store.
+
 	turns, err := store.Turns("conductor:main", 0)
 	if err != nil {
 		t.Fatalf("Turns: %v", err)
 	}
+
 	if len(turns) != 1 {
 		t.Fatalf("turns count = %d, want 1", len(turns))
 	}
+
 	if turns[0].Kind != session.TurnReport {
 		t.Errorf("Kind = %d, want TurnReport(%d)", turns[0].Kind, session.TurnReport)
 	}
+
 	if turns[0].OriginKey != "subagent:subagent-1" {
 		t.Errorf("OriginKey = %q, want %q", turns[0].OriginKey, "subagent:subagent-1")
 	}
+
 	if turns[0].Author != "subagent:subagent-1" {
 		t.Errorf("Author = %q, want %q", turns[0].Author, "subagent:subagent-1")
 	}
+
 	if len(turns[0].Messages) != 1 || turns[0].Messages[0].Role != "user" {
 		t.Errorf("unexpected messages: %v", turns[0].Messages)
 	}
@@ -175,6 +212,7 @@ func TestAdvanceStoredPreventsDoubleWrite(t *testing.T) {
 	rec, adapter, store := newTestRecorder(t)
 
 	// Create conductor session via adapter.
+
 	_ = adapter.GetOrCreate("conductor:main")
 
 	if err := store.Create("subagent:subagent-1", nil); err != nil {
@@ -182,57 +220,157 @@ func TestAdvanceStoredPreventsDoubleWrite(t *testing.T) {
 	}
 
 	// Simulate: conductor has 2 messages already flushed.
+
 	adapter.AddMessage("conductor:main", "user", "hello")
+
 	adapter.AddMessage("conductor:main", "assistant", "hi")
+
 	if err := adapter.Save("conductor:main"); err != nil {
 		t.Fatalf("Save: %v", err)
 	}
 
 	// RecordReport writes directly to store and advances stored counter.
+
 	content := "[System: subagent:subagent-1] result"
+
 	if err := rec.RecordReport("conductor:main", "subagent:subagent-1", "subagent:subagent-1", content); err != nil {
 		t.Fatalf("RecordReport: %v", err)
 	}
 
 	// The in-memory cache should also be updated (by loop.go calling AddFullMessage).
+
 	// Simulate what loop.go does after RecordReport succeeds.
+
 	adapter.AddFullMessage("conductor:main", providers.Message{Role: "user", Content: content})
+
 	// AdvanceStored was already called by RecordReport, so stored = 3 + 1 = 4
+
 	// but we added 1 message to cache making it len=4 as well. No double write.
 
 	// Save should NOT re-write the report turn.
+
 	if err := adapter.Save("conductor:main"); err != nil {
 		t.Fatalf("Save after report: %v", err)
 	}
 
 	// Count all turns in store for conductor session.
+
 	turns, err := store.Turns("conductor:main", 0)
 	if err != nil {
 		t.Fatalf("Turns: %v", err)
 	}
 
 	// Expected: turn 1 (initial 2 msgs), turn 2 (TurnReport from RecordReport)
+
 	// NOT turn 3 (duplicate from flush).
+
 	if len(turns) != 2 {
 		t.Errorf("turns count = %d, want 2 (no double-write)", len(turns))
+
 		for i, turn := range turns {
 			t.Logf("  turn[%d]: seq=%d kind=%d msgs=%d", i, turn.Seq, turn.Kind, len(turn.Messages))
 		}
 	}
 }
 
+func TestRecordQuestion(t *testing.T) {
+	rec, adapter, store := newTestRecorder(t)
+
+	// Create conductor session via adapter so it's in cache.
+
+	_ = adapter.GetOrCreate("conductor:main")
+
+	if err := store.Create("subagent:subagent-1", nil); err != nil {
+		t.Fatalf("create subagent: %v", err)
+	}
+
+	question := "What database schema should I use for the users table?"
+
+	if err := rec.RecordQuestion("conductor:main", "subagent:subagent-1", "subagent-1", question); err != nil {
+		t.Fatalf("RecordQuestion: %v", err)
+	}
+
+	turns, err := store.Turns("conductor:main", 0)
+	if err != nil {
+		t.Fatalf("Turns: %v", err)
+	}
+
+	if len(turns) != 1 {
+		t.Fatalf("turns count = %d, want 1", len(turns))
+	}
+
+	if turns[0].Kind != session.TurnQuestion {
+		t.Errorf("Kind = %d, want TurnQuestion(%d)", turns[0].Kind, session.TurnQuestion)
+	}
+
+	if turns[0].OriginKey != "subagent:subagent-1" {
+		t.Errorf("OriginKey = %q, want %q", turns[0].OriginKey, "subagent:subagent-1")
+	}
+
+	if turns[0].Author != "subagent-1" {
+		t.Errorf("Author = %q, want %q", turns[0].Author, "subagent-1")
+	}
+
+	if len(turns[0].Messages) != 1 || turns[0].Messages[0].Content != question {
+		t.Errorf("unexpected messages: %v", turns[0].Messages)
+	}
+}
+
+func TestRecordPlanSubmit(t *testing.T) {
+	rec, adapter, store := newTestRecorder(t)
+
+	_ = adapter.GetOrCreate("conductor:main")
+
+	if err := store.Create("subagent:subagent-1", nil); err != nil {
+		t.Fatalf("create subagent: %v", err)
+	}
+
+	planText := "Goal: Implement auth\nSteps:\n1. Add middleware\n2. Add JWT validation"
+
+	if err := rec.RecordPlanSubmit("conductor:main", "subagent:subagent-1", "subagent-1", planText); err != nil {
+		t.Fatalf("RecordPlanSubmit: %v", err)
+	}
+
+	turns, err := store.Turns("conductor:main", 0)
+	if err != nil {
+		t.Fatalf("Turns: %v", err)
+	}
+
+	if len(turns) != 1 {
+		t.Fatalf("turns count = %d, want 1", len(turns))
+	}
+
+	if turns[0].Kind != session.TurnPlanSubmit {
+		t.Errorf("Kind = %d, want TurnPlanSubmit(%d)", turns[0].Kind, session.TurnPlanSubmit)
+	}
+
+	if turns[0].OriginKey != "subagent:subagent-1" {
+		t.Errorf("OriginKey = %q, want %q", turns[0].OriginKey, "subagent:subagent-1")
+	}
+
+	if len(turns[0].Messages) != 1 || turns[0].Messages[0].Content != planText {
+		t.Errorf("unexpected messages: %v", turns[0].Messages)
+	}
+}
+
 func TestExtractTaskID(t *testing.T) {
 	tests := []struct {
 		input string
-		want  string
+
+		want string
 	}{
 		{"subagent:subagent-1", "subagent-1"},
+
 		{"subagent:subagent-42", "subagent-42"},
+
 		{"plain-id", "plain-id"},
+
 		{"a:b:c", "c"},
 	}
+
 	for _, tt := range tests {
 		got := extractTaskID(tt.input)
+
 		if got != tt.want {
 			t.Errorf("extractTaskID(%q) = %q, want %q", tt.input, got, tt.want)
 		}
@@ -241,5 +379,6 @@ func TestExtractTaskID(t *testing.T) {
 
 func init() {
 	// Suppress log output in tests.
+
 	os.Setenv("PICOCLAW_LOG_LEVEL", "error")
 }

pkg/agent/session_tracker.go

@@ -8,38 +8,55 @@ import (
 )
 
 // SessionEntry represents an active or recently-active session.
+
 type SessionEntry struct {
-	SessionKey  string    `json:"session_key"`
-	Channel     string    `json:"channel"`
-	ChatID      string    `json:"chat_id"`
-	TouchDir    string    `json:"touch_dir"`
-	ProjectPath string    `json:"project_path,omitempty"` // canonical project path
-	Purpose     string    `json:"purpose,omitempty"`      // 1-line task description
-	Branch      string    `json:"branch,omitempty"`       // git branch name
-	LastSeenAt  time.Time `json:"last_seen_at"`
+	SessionKey string `json:"session_key"`
+
+	Channel string `json:"channel"`
+
+	ChatID string `json:"chat_id"`
+
+	TouchDir string `json:"touch_dir"`
+
+	ProjectPath string `json:"project_path,omitempty"` // canonical project path
+
+	Purpose string `json:"purpose,omitempty"` // 1-line task description
+
+	Branch string `json:"branch,omitempty"` // git branch name
+
+	LastSeenAt time.Time `json:"last_seen_at"`
 }
 
 // TouchMeta carries optional metadata for Touch calls.
+
 type TouchMeta struct {
 	ProjectPath string // canonical project path (always original workspace-relative)
-	Purpose     string // 1-line task description
-	Branch      string // git branch name
+
+	Purpose string // 1-line task description
+
+	Branch string // git branch name
 }
 
 // PeerInfo is the minimal info shared between sessions on the same project.
+
 type PeerInfo struct {
 	SessionKey string
-	Purpose    string
-	Branch     string
+
+	Purpose string
+
+	Branch string
 }
 
 // SessionTracker tracks per-session tool-call activity.
+
 // Thread-safe; used by AgentLoop for plan coordination and by the mini app API for observability.
+
 type SessionTracker struct {
 	entries sync.Map // sessionKey → *SessionEntry
 }
 
 // NewSessionTracker creates a new tracker.
+
 func NewSessionTracker() *SessionTracker {
 	return &SessionTracker{}
 }
@@ -47,121 +64,177 @@ func NewSessionTracker() *SessionTracker {
 const sessionActivityTimeout = 15 * time.Minute
 
 // Touch records a tool-call activity for a session.
+
 // dir is the workspace-relative directory the tool call targeted.
+
 // If dir is empty, only LastSeenAt is updated.
+
 // meta is optional and carries project coordination metadata.
+
 func (st *SessionTracker) Touch(sessionKey, channel, chatID, dir string, meta *TouchMeta) {
 	now := time.Now()
+
 	val, loaded := st.entries.Load(sessionKey)
+
 	if loaded {
 		entry := val.(*SessionEntry)
+
 		entry.LastSeenAt = now
+
 		if dir != "" {
 			entry.TouchDir = dir
 		}
+
 		if channel != "" {
 			entry.Channel = channel
 		}
+
 		if chatID != "" {
 			entry.ChatID = chatID
 		}
+
 		if meta != nil {
 			if meta.ProjectPath != "" {
 				entry.ProjectPath = meta.ProjectPath
 			}
+
 			if meta.Purpose != "" {
 				entry.Purpose = meta.Purpose
 			}
+
 			if meta.Branch != "" {
 				entry.Branch = meta.Branch
 			}
 		}
+
 		return
 	}
+
 	entry := &SessionEntry{
 		SessionKey: sessionKey,
-		Channel:    channel,
-		ChatID:     chatID,
-		TouchDir:   dir,
+
+		Channel: channel,
+
+		ChatID: chatID,
+
+		TouchDir: dir,
+
 		LastSeenAt: now,
 	}
+
 	if meta != nil {
 		entry.ProjectPath = meta.ProjectPath
+
 		entry.Purpose = meta.Purpose
+
 		entry.Branch = meta.Branch
 	}
+
 	st.entries.Store(sessionKey, entry)
 }
 
 // IsActiveInDir returns true if any session (excluding those matching excludeKey)
+
 // has touched a directory overlapping with dir within sessionActivityTimeout.
+
 // Overlap = either is a prefix of the other (parent/child relationship).
+
 func (st *SessionTracker) IsActiveInDir(dir, excludeKey string) bool {
 	cutoff := time.Now().Add(-sessionActivityTimeout)
+
 	active := false
+
 	st.entries.Range(func(key, val any) bool {
 		if key.(string) == excludeKey {
 			return true
 		}
+
 		entry := val.(*SessionEntry)
+
 		if entry.LastSeenAt.After(cutoff) && entry.TouchDir != "" &&
+
 			(strings.HasPrefix(entry.TouchDir, dir) || strings.HasPrefix(dir, entry.TouchDir)) {
 			active = true
+
 			return false
 		}
+
 		return true
 	})
+
 	return active
 }
 
 // ListActive returns all sessions seen within sessionActivityTimeout,
+
 // sorted by LastSeenAt descending (most recent first).
+
 func (st *SessionTracker) ListActive() []SessionEntry {
 	cutoff := time.Now().Add(-sessionActivityTimeout)
+
 	var result []SessionEntry
+
 	st.entries.Range(func(key, val any) bool {
 		entry := val.(*SessionEntry)
+
 		if entry.LastSeenAt.After(cutoff) {
 			result = append(result, *entry) // copy
 		}
+
 		return true
 	})
+
 	sort.Slice(result, func(i, j int) bool {
 		return result[i].LastSeenAt.After(result[j].LastSeenAt)
 	})
+
 	return result
 }
 
 // GetTouchDir returns the TouchDir for a given session key, or "" if not found.
+
 func (st *SessionTracker) GetTouchDir(sessionKey string) string {
 	val, ok := st.entries.Load(sessionKey)
+
 	if !ok {
 		return ""
 	}
+
 	return val.(*SessionEntry).TouchDir
 }
 
 // GetPeerPurposes returns purposes of other active sessions targeting the same project.
+
 // Used for lightweight coordination without context pollution.
+
 func (st *SessionTracker) GetPeerPurposes(sessionKey, projectPath string) []PeerInfo {
 	if projectPath == "" {
 		return nil
 	}
+
 	cutoff := time.Now().Add(-sessionActivityTimeout)
+
 	var result []PeerInfo
+
 	st.entries.Range(func(key, val any) bool {
 		if key.(string) == sessionKey {
 			return true
 		}
+
 		entry := val.(*SessionEntry)
+
 		if entry.LastSeenAt.After(cutoff) && entry.ProjectPath == projectPath {
 			result = append(result, PeerInfo{
 				SessionKey: entry.SessionKey,
-				Purpose:    entry.Purpose,
-				Branch:     entry.Branch,
+
+				Purpose: entry.Purpose,
+
+				Branch: entry.Branch,
 			})
 		}
+
 		return true
 	})
+
 	return result
 }

pkg/agent/session_tracker_test.go

@@ -9,38 +9,53 @@ func TestTouch(t *testing.T) {
 	st := NewSessionTracker()
 
 	// Basic touch creates entry
+
 	st.Touch("sess1", "telegram", "123", "projects/myapp", nil)
+
 	entries := st.ListActive()
+
 	if len(entries) != 1 {
 		t.Fatalf("expected 1 entry, got %d", len(entries))
 	}
+
 	if entries[0].SessionKey != "sess1" {
 		t.Errorf("expected session_key=sess1, got %s", entries[0].SessionKey)
 	}
+
 	if entries[0].Channel != "telegram" {
 		t.Errorf("expected channel=telegram, got %s", entries[0].Channel)
 	}
+
 	if entries[0].TouchDir != "projects/myapp" {
 		t.Errorf("expected touch_dir=projects/myapp, got %s", entries[0].TouchDir)
 	}
 
 	// Touch again with new dir overwrites TouchDir
+
 	st.Touch("sess1", "", "", "projects/other", nil)
+
 	entries = st.ListActive()
+
 	if len(entries) != 1 {
 		t.Fatalf("expected 1 entry, got %d", len(entries))
 	}
+
 	if entries[0].TouchDir != "projects/other" {
 		t.Errorf("expected touch_dir=projects/other, got %s", entries[0].TouchDir)
 	}
+
 	// Channel should remain from first touch
+
 	if entries[0].Channel != "telegram" {
 		t.Errorf("expected channel=telegram (unchanged), got %s", entries[0].Channel)
 	}
 
 	// Touch with empty dir does not overwrite TouchDir
+
 	st.Touch("sess1", "", "", "", nil)
+
 	entries = st.ListActive()
+
 	if entries[0].TouchDir != "projects/other" {
 		t.Errorf("expected touch_dir unchanged, got %s", entries[0].TouchDir)
 	}
@@ -50,36 +65,45 @@ func TestIsActiveInDir(t *testing.T) {
 	st := NewSessionTracker()
 
 	// Setup: sess1 touches "projects/myapp"
+
 	st.Touch("sess1", "telegram", "123", "projects/myapp", nil)
 
 	// Same dir, excluding sess1 → false
+
 	if st.IsActiveInDir("projects/myapp", "sess1") {
 		t.Error("expected false when excluding the only active session")
 	}
 
 	// Same dir, excluding different key → true
+
 	if !st.IsActiveInDir("projects/myapp", "heartbeat") {
 		t.Error("expected true for exact dir match")
 	}
 
 	// Parent dir match: "projects" is prefix of "projects/myapp"
+
 	if !st.IsActiveInDir("projects", "heartbeat") {
 		t.Error("expected true for parent dir match")
 	}
 
 	// Child dir match: "projects/myapp/src" has prefix "projects/myapp"
+
 	if !st.IsActiveInDir("projects/myapp/src", "heartbeat") {
 		t.Error("expected true for child dir match")
 	}
 
 	// Unrelated dir → false
+
 	if st.IsActiveInDir("other/stuff", "heartbeat") {
 		t.Error("expected false for unrelated dir")
 	}
 
 	// Stale entry (manually set LastSeenAt to past)
+
 	val, _ := st.entries.Load("sess1")
+
 	entry := val.(*SessionEntry)
+
 	entry.LastSeenAt = time.Now().Add(-sessionActivityTimeout - time.Minute)
 
 	if st.IsActiveInDir("projects/myapp", "heartbeat") {
@@ -91,32 +115,43 @@ func TestListActive(t *testing.T) {
 	st := NewSessionTracker()
 
 	// Add two sessions
+
 	st.Touch("sess1", "telegram", "123", "projects/a", nil)
+
 	time.Sleep(5 * time.Millisecond) // ensure different timestamps
+
 	st.Touch("sess2", "discord", "456", "projects/b", nil)
 
 	entries := st.ListActive()
+
 	if len(entries) != 2 {
 		t.Fatalf("expected 2 entries, got %d", len(entries))
 	}
 
 	// Most recent first
+
 	if entries[0].SessionKey != "sess2" {
 		t.Errorf("expected sess2 first (most recent), got %s", entries[0].SessionKey)
 	}
+
 	if entries[1].SessionKey != "sess1" {
 		t.Errorf("expected sess1 second, got %s", entries[1].SessionKey)
 	}
 
 	// Make sess1 stale
+
 	val, _ := st.entries.Load("sess1")
+
 	entry := val.(*SessionEntry)
+
 	entry.LastSeenAt = time.Now().Add(-sessionActivityTimeout - time.Minute)
 
 	entries = st.ListActive()
+
 	if len(entries) != 1 {
 		t.Fatalf("expected 1 active entry after stale, got %d", len(entries))
 	}
+
 	if entries[0].SessionKey != "sess2" {
 		t.Errorf("expected only sess2, got %s", entries[0].SessionKey)
 	}

pkg/session/graph.go

@@ -8,96 +8,136 @@ import (
 )
 
 // SessionGraph is a thin wrapper around SessionStore that provides
+
 // structured turn-writing via BeginTurn/TurnWriter.
+
 // It does NOT replace LegacyAdapter — existing call sites remain unchanged.
+
 // Future phases will migrate callers to use SessionGraph directly.
+
 type SessionGraph struct {
 	store SessionStore
 }
 
 // NewSessionGraph creates a SessionGraph backed by the given store.
+
 func NewSessionGraph(store SessionStore) *SessionGraph {
 	return &SessionGraph{store: store}
 }
 
 // Messages returns all messages for the session by reading turns from the store.
+
 func (g *SessionGraph) Messages(sessionKey string) ([]providers.Message, error) {
 	turns, err := g.store.Turns(sessionKey, 0)
 	if err != nil {
 		return nil, err
 	}
+
 	var msgs []providers.Message
+
 	for _, t := range turns {
 		msgs = append(msgs, t.Messages...)
 	}
+
 	if msgs == nil {
 		msgs = []providers.Message{}
 	}
+
 	return msgs, nil
 }
 
 // BeginTurn starts a new turn that can be built up incrementally
+
 // and committed atomically.
+
 func (g *SessionGraph) BeginTurn(sessionKey string, kind TurnKind) *TurnWriter {
 	return &TurnWriter{
-		store:      g.store,
+		store: g.store,
+
 		sessionKey: sessionKey,
+
 		turn: Turn{
 			SessionKey: sessionKey,
-			Kind:       kind,
+
+			Kind: kind,
 		},
 	}
 }
 
 // TurnWriter accumulates messages for a single turn and commits them atomically.
+
 type TurnWriter struct {
-	mu         sync.Mutex
-	store      SessionStore
+	mu sync.Mutex
+
+	store SessionStore
+
 	sessionKey string
-	turn       Turn
-	committed  bool
-	discarded  bool
+
+	turn Turn
+
+	committed bool
+
+	discarded bool
 }
 
 // Add appends a message to the pending turn.
+
 func (tw *TurnWriter) Add(msg providers.Message) {
 	tw.mu.Lock()
+
 	defer tw.mu.Unlock()
+
 	tw.turn.Messages = append(tw.turn.Messages, msg)
 }
 
 // SetOrigin sets the origin session key for this turn (e.g. subagent source).
+
 func (tw *TurnWriter) SetOrigin(sessionKey string) {
 	tw.mu.Lock()
+
 	defer tw.mu.Unlock()
+
 	tw.turn.OriginKey = sessionKey
 }
 
 // SetAuthor sets the author field for this turn.
+
 func (tw *TurnWriter) SetAuthor(author string) {
 	tw.mu.Lock()
+
 	defer tw.mu.Unlock()
+
 	tw.turn.Author = author
 }
 
 // Commit writes the accumulated turn to the store.
+
 // Returns an error if already committed or discarded.
+
 func (tw *TurnWriter) Commit() error {
 	tw.mu.Lock()
+
 	defer tw.mu.Unlock()
+
 	if tw.committed {
 		return errors.New("turn already committed")
 	}
+
 	if tw.discarded {
 		return errors.New("turn already discarded")
 	}
+
 	tw.committed = true
+
 	return tw.store.Append(tw.sessionKey, &tw.turn)
 }
 
 // Discard marks the turn as abandoned — nothing is written.
+
 func (tw *TurnWriter) Discard() {
 	tw.mu.Lock()
+
 	defer tw.mu.Unlock()
+
 	tw.discarded = true
 }

pkg/session/graph_test.go

@@ -8,19 +8,25 @@ import (
 
 func TestSessionGraph_Messages(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("g1", nil); err != nil {
 		t.Fatal(err)
 	}
+
 	if err := store.Append("g1", &Turn{
-		Kind:     TurnNormal,
+		Kind: TurnNormal,
+
 		Messages: []providers.Message{{Role: "user", Content: "hello"}},
 	}); err != nil {
 		t.Fatal(err)
 	}
+
 	if err := store.Append("g1", &Turn{
 		Kind: TurnNormal,
+
 		Messages: []providers.Message{
 			{Role: "assistant", Content: "hi"},
+
 			{Role: "user", Content: "how are you"},
 		},
 	}); err != nil {
@@ -28,13 +34,16 @@ func TestSessionGraph_Messages(t *testing.T) {
 	}
 
 	g := NewSessionGraph(store)
+
 	msgs, err := g.Messages("g1")
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	if len(msgs) != 3 {
 		t.Fatalf("expected 3 messages, got %d", len(msgs))
 	}
+
 	if msgs[0].Content != "hello" || msgs[1].Content != "hi" || msgs[2].Content != "how are you" {
 		t.Errorf("unexpected messages: %+v", msgs)
 	}
@@ -42,14 +51,18 @@ func TestSessionGraph_Messages(t *testing.T) {
 
 func TestSessionGraph_Messages_Empty(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("empty", nil); err != nil {
 		t.Fatal(err)
 	}
+
 	g := NewSessionGraph(store)
+
 	msgs, err := g.Messages("empty")
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	if msgs == nil || len(msgs) != 0 {
 		t.Errorf("expected empty slice, got %v", msgs)
 	}
@@ -57,15 +70,21 @@ func TestSessionGraph_Messages_Empty(t *testing.T) {
 
 func TestTurnWriter_Commit(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("tw1", nil); err != nil {
 		t.Fatal(err)
 	}
 
 	g := NewSessionGraph(store)
+
 	tw := g.BeginTurn("tw1", TurnNormal)
+
 	tw.Add(providers.Message{Role: "user", Content: "msg1"})
+
 	tw.Add(providers.Message{Role: "assistant", Content: "msg2"})
+
 	tw.SetOrigin("parent-key")
+
 	tw.SetAuthor("agent-1")
 
 	if err := tw.Commit(); err != nil {
@@ -76,15 +95,19 @@ func TestTurnWriter_Commit(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	if len(turns) != 1 {
 		t.Fatalf("expected 1 turn, got %d", len(turns))
 	}
+
 	if len(turns[0].Messages) != 2 {
 		t.Fatalf("expected 2 messages, got %d", len(turns[0].Messages))
 	}
+
 	if turns[0].OriginKey != "parent-key" {
 		t.Errorf("expected origin 'parent-key', got %q", turns[0].OriginKey)
 	}
+
 	if turns[0].Author != "agent-1" {
 		t.Errorf("expected author 'agent-1', got %q", turns[0].Author)
 	}
@@ -92,19 +115,24 @@ func TestTurnWriter_Commit(t *testing.T) {
 
 func TestTurnWriter_Discard(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("tw2", nil); err != nil {
 		t.Fatal(err)
 	}
 
 	g := NewSessionGraph(store)
+
 	tw := g.BeginTurn("tw2", TurnNormal)
+
 	tw.Add(providers.Message{Role: "user", Content: "should not persist"})
+
 	tw.Discard()
 
 	turns, err := store.Turns("tw2", 0)
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	if len(turns) != 0 {
 		t.Errorf("expected 0 turns after discard, got %d", len(turns))
 	}
@@ -112,17 +140,21 @@ func TestTurnWriter_Discard(t *testing.T) {
 
 func TestTurnWriter_DoubleCommit(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("tw3", nil); err != nil {
 		t.Fatal(err)
 	}
 
 	g := NewSessionGraph(store)
+
 	tw := g.BeginTurn("tw3", TurnNormal)
+
 	tw.Add(providers.Message{Role: "user", Content: "once"})
 
 	if err := tw.Commit(); err != nil {
 		t.Fatal(err)
 	}
+
 	if err := tw.Commit(); err == nil {
 		t.Error("expected error on double commit")
 	}
@@ -130,13 +162,17 @@ func TestTurnWriter_DoubleCommit(t *testing.T) {
 
 func TestTurnWriter_CommitAfterDiscard(t *testing.T) {
 	store := newTestStore(t)
+
 	if err := store.Create("tw4", nil); err != nil {
 		t.Fatal(err)
 	}
 
 	g := NewSessionGraph(store)
+
 	tw := g.BeginTurn("tw4", TurnNormal)
+
 	tw.Add(providers.Message{Role: "user", Content: "x"})
+
 	tw.Discard()
 
 	if err := tw.Commit(); err == nil {

pkg/session/legacy_adapter.go

@@ -460,94 +460,137 @@ func (la *LegacyAdapter) Save(key string) error {
 }
 
 // DefaultPruneTTL is the default time-to-live for session pruning.
+
 const DefaultPruneTTL = 7 * 24 * time.Hour
 
 // CompactOldTurns flushes pending writes, then compacts SQLite turns
+
 // keeping only the last keepLast messages. Sets session summary to the given value.
+
 func (la *LegacyAdapter) CompactOldTurns(key string, keepLast int, summary string) error {
 	// 1. Flush pending messages to SQLite
+
 	if err := la.Save(key); err != nil {
 		return err
 	}
+
 	// 2. Query all turns
+
 	turns, err := la.store.Turns(key, 0)
 	if err != nil {
 		return err
 	}
+
 	// 3. Count total messages, find cut point
+
 	totalMsgs := 0
+
 	for _, t := range turns {
 		totalMsgs += len(t.Messages)
 	}
+
 	if keepLast >= totalMsgs {
 		// Nothing to compact, just update summary
+
 		if err := la.store.SetSummary(key, summary); err != nil {
 			return err
 		}
+
 		la.mu.Lock()
+
 		if c, ok := la.cache[key]; ok {
 			c.summary = summary
 		}
+
 		la.mu.Unlock()
+
 		return nil
 	}
+
 	dropCount := totalMsgs - keepLast
+
 	accumulated := 0
+
 	cutSeq := 0
+
 	for _, t := range turns {
 		accumulated += len(t.Messages)
+
 		if accumulated <= dropCount {
 			cutSeq = t.Seq
 		} else {
 			break
 		}
 	}
+
 	if cutSeq == 0 {
 		if err := la.store.SetSummary(key, summary); err != nil {
 			return err
 		}
+
 		la.mu.Lock()
+
 		if c, ok := la.cache[key]; ok {
 			c.summary = summary
 		}
+
 		la.mu.Unlock()
+
 		return nil
 	}
+
 	// 4. Compact in SQLite
+
 	if err := la.store.Compact(key, cutSeq, summary); err != nil {
 		return err
 	}
+
 	// 5. Update in-memory cache
+
 	la.mu.Lock()
+
 	defer la.mu.Unlock()
+
 	if c, ok := la.cache[key]; ok {
 		if keepLast < len(c.messages) {
 			c.messages = c.messages[len(c.messages)-keepLast:]
 		}
+
 		c.stored = len(c.messages)
+
 		c.replaced = false
+
 		c.dirty = false
+
 		c.summary = summary
 	}
+
 	return nil
 }
 
 // Store returns the underlying SessionStore for direct DAG operations.
+
 func (la *LegacyAdapter) Store() SessionStore {
 	return la.store
 }
 
 // Graph returns a SessionGraph backed by the underlying store.
+
 func (la *LegacyAdapter) Graph() *SessionGraph {
 	return NewSessionGraph(la.store)
 }
 
 // AdvanceStored increments the stored counter for a session by delta,
+
 // preventing the flush loop from re-persisting messages already written
+
 // directly to the store (e.g. TurnReport).
+
 func (la *LegacyAdapter) AdvanceStored(key string, delta int) {
 	la.mu.Lock()
+
 	defer la.mu.Unlock()
+
 	if c, ok := la.cache[key]; ok {
 		c.stored += delta
 	}
@@ -573,16 +616,25 @@ func (la *LegacyAdapter) Close() {
 
 func (la *LegacyAdapter) flushLoop() {
 	flushTicker := time.NewTicker(5 * time.Minute)
+
 	pruneTicker := time.NewTicker(6 * time.Hour)
+
 	defer flushTicker.Stop()
+
 	defer pruneTicker.Stop()
+
 	for {
 		select {
 		case <-flushTicker.C:
+
 			la.FlushDirty()
+
 		case <-pruneTicker.C:
+
 			_, _ = la.store.Prune(DefaultPruneTTL)
+
 		case <-la.done:
+
 			return
 		}
 	}

pkg/session/legacy_adapter_test.go

@@ -126,7 +126,11 @@ func TestBackend_AddFullMessage(t *testing.T) {
 				Content: "sure",
 
 				ToolCalls: []providers.ToolCall{
-					{ID: "call_1", Type: "function", Function: &providers.FunctionCall{Name: "exec", Arguments: map[string]any{}}},
+					{
+						ID:       "call_1",
+						Type:     "function",
+						Function: &providers.FunctionCall{Name: "exec", Arguments: map[string]any{}},
+					},
 				},
 			})
 
@@ -485,49 +489,72 @@ func TestBackend_IncrementalSave(t *testing.T) {
 
 func TestCompactOldTurns(t *testing.T) {
 	dbPath := filepath.Join(t.TempDir(), "test.db")
+
 	store, err := OpenSQLiteStore(dbPath)
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	la := NewLegacyAdapter(store)
+
 	defer la.Close()
 
 	la.GetOrCreate("k1")
+
 	// Turn 1: 2 messages
+
 	la.AddMessage("k1", "user", "a")
+
 	la.AddMessage("k1", "assistant", "b")
+
 	la.Save("k1")
+
 	// Turn 2: 3 messages
+
 	la.AddMessage("k1", "user", "c")
+
 	la.AddMessage("k1", "assistant", "d")
+
 	la.AddMessage("k1", "user", "e")
+
 	la.Save("k1")
+
 	// Turn 3: 2 messages
+
 	la.AddMessage("k1", "user", "f")
+
 	la.AddMessage("k1", "assistant", "g")
+
 	la.Save("k1")
 
 	// Total: 7 messages across 3 turns. keepLast=2 → drop 5 → compact turns 1+2 (5 msgs)
+
 	if err := la.CompactOldTurns("k1", 2, "test summary"); err != nil {
 		t.Fatalf("CompactOldTurns: %v", err)
 	}
 
 	h := la.GetHistory("k1")
+
 	if len(h) != 2 {
 		t.Fatalf("expected 2 messages in cache, got %d", len(h))
 	}
+
 	if h[0].Content != "f" || h[1].Content != "g" {
 		t.Errorf("unexpected messages: %+v", h)
 	}
+
 	if s := la.GetSummary("k1"); s != "test summary" {
 		t.Errorf("expected summary 'test summary', got %q", s)
 	}
 
 	// Verify in SQLite: only turn 3 remains
+
 	turns, _ := store.Turns("k1", 0)
+
 	if len(turns) != 1 {
 		t.Fatalf("expected 1 turn in SQLite, got %d", len(turns))
 	}
+
 	if len(turns[0].Messages) != 2 {
 		t.Errorf("expected 2 messages in remaining turn, got %d", len(turns[0].Messages))
 	}
@@ -535,27 +562,36 @@ func TestCompactOldTurns(t *testing.T) {
 
 func TestCompactOldTurns_NothingToCompact(t *testing.T) {
 	dbPath := filepath.Join(t.TempDir(), "test.db")
+
 	store, err := OpenSQLiteStore(dbPath)
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	la := NewLegacyAdapter(store)
+
 	defer la.Close()
 
 	la.GetOrCreate("k1")
+
 	la.AddMessage("k1", "user", "a")
+
 	la.AddMessage("k1", "assistant", "b")
+
 	la.Save("k1")
 
 	// keepLast=10 >= total 2 → nothing compacted, summary still updated
+
 	if err := la.CompactOldTurns("k1", 10, "new summary"); err != nil {
 		t.Fatalf("CompactOldTurns: %v", err)
 	}
 
 	h := la.GetHistory("k1")
+
 	if len(h) != 2 {
 		t.Fatalf("expected 2 messages, got %d", len(h))
 	}
+
 	if s := la.GetSummary("k1"); s != "new summary" {
 		t.Errorf("expected 'new summary', got %q", s)
 	}
@@ -563,29 +599,40 @@ func TestCompactOldTurns_NothingToCompact(t *testing.T) {
 
 func TestCompactOldTurns_SingleTurn(t *testing.T) {
 	dbPath := filepath.Join(t.TempDir(), "test.db")
+
 	store, err := OpenSQLiteStore(dbPath)
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	la := NewLegacyAdapter(store)
+
 	defer la.Close()
 
 	la.GetOrCreate("k1")
+
 	la.AddMessage("k1", "user", "a")
+
 	la.AddMessage("k1", "assistant", "b")
+
 	la.AddMessage("k1", "user", "c")
+
 	la.Save("k1")
 
 	// Single turn with 3 messages, keepLast=2 → dropCount=1, but first turn has 3 msgs
+
 	// accumulated(3) > dropCount(1) on first turn → cutSeq=0 → no compaction
+
 	if err := la.CompactOldTurns("k1", 2, "sum"); err != nil {
 		t.Fatalf("CompactOldTurns: %v", err)
 	}
 
 	h := la.GetHistory("k1")
+
 	if len(h) != 3 {
 		t.Fatalf("expected 3 messages (no compaction), got %d", len(h))
 	}
+
 	if s := la.GetSummary("k1"); s != "sum" {
 		t.Errorf("expected 'sum', got %q", s)
 	}
@@ -593,22 +640,29 @@ func TestCompactOldTurns_SingleTurn(t *testing.T) {
 
 func TestCompactOldTurns_Graph(t *testing.T) {
 	dbPath := filepath.Join(t.TempDir(), "test.db")
+
 	store, err := OpenSQLiteStore(dbPath)
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	la := NewLegacyAdapter(store)
+
 	defer la.Close()
 
 	la.GetOrCreate("k1")
+
 	la.AddMessage("k1", "user", "hello")
+
 	la.Save("k1")
 
 	g := la.Graph()
+
 	msgs, err := g.Messages("k1")
 	if err != nil {
 		t.Fatal(err)
 	}
+
 	if len(msgs) != 1 || msgs[0].Content != "hello" {
 		t.Errorf("unexpected graph messages: %+v", msgs)
 	}

pkg/session/manager.go

@@ -12,56 +12,76 @@ import (
 )
 
 type Session struct {
-	Key      string              `json:"key"`
+	Key string `json:"key"`
+
 	Messages []providers.Message `json:"messages"`
-	Summary  string              `json:"summary,omitempty"`
-	Created  time.Time           `json:"created"`
-	Updated  time.Time           `json:"updated"`
+
+	Summary string `json:"summary,omitempty"`
+
+	Created time.Time `json:"created"`
+
+	Updated time.Time `json:"updated"`
 }
 
 type SessionManager struct {
 	sessions map[string]*Session
-	mu       sync.RWMutex
-	storage  string
+
+	mu sync.RWMutex
+
+	storage string
 
 	// Write-behind: dirty keys are flushed periodically to reduce disk writes.
-	dirtyMu   sync.Mutex
+
+	dirtyMu sync.Mutex
+
 	dirtyKeys map[string]bool
-	done      chan struct{}
+
+	done chan struct{}
 }
 
 func NewSessionManager(storage string) *SessionManager {
 	sm := &SessionManager{
-		sessions:  make(map[string]*Session),
-		storage:   storage,
+		sessions: make(map[string]*Session),
+
+		storage: storage,
+
 		dirtyKeys: make(map[string]bool),
-		done:      make(chan struct{}),
+
+		done: make(chan struct{}),
 	}
 
 	if storage != "" {
 		os.MkdirAll(storage, 0o755)
+
 		sm.loadSessions()
 	}
 
 	go sm.flushLoop()
+
 	return sm
 }
 
 func (sm *SessionManager) GetOrCreate(key string) *Session {
 	sm.mu.Lock()
+
 	defer sm.mu.Unlock()
 
 	session, ok := sm.sessions[key]
+
 	if ok {
 		return session
 	}
 
 	session = &Session{
-		Key:      key,
+		Key: key,
+
 		Messages: []providers.Message{},
-		Created:  time.Now(),
-		Updated:  time.Now(),
+
+		Created: time.Now(),
+
+		Updated: time.Now(),
 	}
+
 	sm.sessions[key] = session
 
 	return session
@@ -69,79 +89,102 @@ func (sm *SessionManager) GetOrCreate(key string) *Session {
 
 func (sm *SessionManager) AddMessage(sessionKey, role, content string) {
 	sm.AddFullMessage(sessionKey, providers.Message{
-		Role:    role,
+		Role: role,
+
 		Content: content,
 	})
 }
 
 // AddFullMessage adds a complete message with tool calls and tool call ID to the session.
+
 // This is used to save the full conversation flow including tool calls and tool results.
+
 func (sm *SessionManager) AddFullMessage(sessionKey string, msg providers.Message) {
 	sm.mu.Lock()
+
 	defer sm.mu.Unlock()
 
 	session, ok := sm.sessions[sessionKey]
+
 	if !ok {
 		session = &Session{
-			Key:      sessionKey,
+			Key: sessionKey,
+
 			Messages: []providers.Message{},
-			Created:  time.Now(),
+
+			Created: time.Now(),
 		}
+
 		sm.sessions[sessionKey] = session
 	}
 
 	session.Messages = append(session.Messages, msg)
+
 	session.Updated = time.Now()
 }
 
 func (sm *SessionManager) GetHistory(key string) []providers.Message {
 	sm.mu.RLock()
+
 	defer sm.mu.RUnlock()
 
 	session, ok := sm.sessions[key]
+
 	if !ok {
 		return []providers.Message{}
 	}
 
 	history := make([]providers.Message, len(session.Messages))
+
 	copy(history, session.Messages)
+
 	return history
 }
 
 func (sm *SessionManager) GetSummary(key string) string {
 	sm.mu.RLock()
+
 	defer sm.mu.RUnlock()
 
 	session, ok := sm.sessions[key]
+
 	if !ok {
 		return ""
 	}
+
 	return session.Summary
 }
 
 func (sm *SessionManager) SetSummary(key string, summary string) {
 	sm.mu.Lock()
+
 	defer sm.mu.Unlock()
 
 	session, ok := sm.sessions[key]
+
 	if ok {
 		session.Summary = summary
+
 		session.Updated = time.Now()
 	}
 }
 
 func (sm *SessionManager) TruncateHistory(key string, keepLast int) {
 	sm.mu.Lock()
+
 	defer sm.mu.Unlock()
 
 	session, ok := sm.sessions[key]
+
 	if !ok {
 		return
 	}
 
 	if keepLast <= 0 {
 		session.Messages = []providers.Message{}
+
 		session.Updated = time.Now()
+
 		return
 	}
 
@@ -150,14 +193,20 @@ func (sm *SessionManager) TruncateHistory(key string, keepLast int) {
 	}
 
 	session.Messages = session.Messages[len(session.Messages)-keepLast:]
+
 	session.Updated = time.Now()
 }
 
 // sanitizeFilename converts a session key into a cross-platform safe filename.
+
 // Session keys use "channel:chatID" (e.g. "telegram:123456") but ':' is the
+
 // volume separator on Windows, so filepath.Base would misinterpret the key.
+
 // We replace it with '_'. The original key is preserved inside the JSON file,
+
 // so loadSessions still maps back to the right in-memory key.
+
 func sanitizeFilename(key string) string {
 	return strings.ReplaceAll(key, ":", "_")
 }
@@ -170,33 +219,47 @@ func (sm *SessionManager) Save(key string) error {
 	filename := sanitizeFilename(key)
 
 	// filepath.IsLocal rejects empty names, "..", absolute paths, and
+
 	// OS-reserved device names (NUL, COM1 … on Windows).
+
 	// The extra checks reject "." and any directory separators so that
+
 	// the session file is always written directly inside sm.storage.
+
 	if filename == "." || !filepath.IsLocal(filename) || strings.ContainsAny(filename, `/\`) {
 		return os.ErrInvalid
 	}
 
 	// Snapshot under read lock, then perform slow file I/O after unlock.
+
 	sm.mu.RLock()
+
 	stored, ok := sm.sessions[key]
+
 	if !ok {
 		sm.mu.RUnlock()
+
 		return nil
 	}
 
 	snapshot := Session{
-		Key:     stored.Key,
+		Key: stored.Key,
+
 		Summary: stored.Summary,
+
 		Created: stored.Created,
+
 		Updated: stored.Updated,
 	}
+
 	if len(stored.Messages) > 0 {
 		snapshot.Messages = make([]providers.Message, len(stored.Messages))
+
 		copy(snapshot.Messages, stored.Messages)
 	} else {
 		snapshot.Messages = []providers.Message{}
 	}
+
 	sm.mu.RUnlock()
 
 	data, err := json.MarshalIndent(snapshot, "", "  ")
@@ -205,13 +268,16 @@ func (sm *SessionManager) Save(key string) error {
 	}
 
 	sessionPath := filepath.Join(sm.storage, filename+".json")
+
 	tmpFile, err := os.CreateTemp(sm.storage, "session-*.tmp")
 	if err != nil {
 		return err
 	}
 
 	tmpPath := tmpFile.Name()
+
 	cleanup := true
+
 	defer func() {
 		if cleanup {
 			_ = os.Remove(tmpPath)
@@ -220,16 +286,22 @@ func (sm *SessionManager) Save(key string) error {
 
 	if _, err := tmpFile.Write(data); err != nil {
 		_ = tmpFile.Close()
+
 		return err
 	}
+
 	if err := tmpFile.Chmod(0o644); err != nil {
 		_ = tmpFile.Close()
+
 		return err
 	}
+
 	if err := tmpFile.Sync(); err != nil {
 		_ = tmpFile.Close()
+
 		return err
 	}
+
 	if err := tmpFile.Close(); err != nil {
 		return err
 	}
@@ -237,7 +309,9 @@ func (sm *SessionManager) Save(key string) error {
 	if err := os.Rename(tmpPath, sessionPath); err != nil {
 		return err
 	}
+
 	cleanup = false
+
 	return nil
 }
 
@@ -257,12 +331,14 @@ func (sm *SessionManager) loadSessions() error {
 		}
 
 		sessionPath := filepath.Join(sm.storage, file.Name())
+
 		data, err := os.ReadFile(sessionPath)
 		if err != nil {
 			continue
 		}
 
 		var session Session
+
 		if err := json.Unmarshal(data, &session); err != nil {
 			continue
 		}
@@ -274,57 +350,84 @@ func (sm *SessionManager) loadSessions() error {
 }
 
 // SanitizeHistory rebuilds session history to ensure valid tool-call ordering.
+
 // LLM APIs require that every assistant message with ToolCalls is immediately
+
 // followed by exactly the matching tool-result messages (role="tool"), with no
+
 // other messages in between. Violations can happen from session collisions or
+
 // mid-execution crashes.
+
 //
+
 // The function walks the full history and copies only well-formed groups:
+
 //   - user/system messages are always kept
+
 //   - assistant messages without tool calls are always kept
+
 //   - assistant messages WITH tool calls are kept only if the immediately
+
 //     following messages are the complete set of matching tool results
+
 //
+
 // Returns the sanitized history and the number of messages removed.
+
 func SanitizeHistory(history []providers.Message) ([]providers.Message, int) {
 	if len(history) == 0 {
 		return history, 0
 	}
 
 	result := make([]providers.Message, 0, len(history))
+
 	i := 0
 
 	for i < len(history) {
 		msg := history[i]
 
 		// Non-assistant messages or assistant without tool calls: keep
+
 		if msg.Role != "assistant" || len(msg.ToolCalls) == 0 {
 			// Skip stray tool results not preceded by their assistant
+
 			if msg.Role == "tool" {
 				i++
+
 				continue
 			}
+
 			result = append(result, msg)
+
 			i++
+
 			continue
 		}
 
 		// Assistant with tool calls: validate the immediately following messages
+
 		expectedIDs := make(map[string]bool, len(msg.ToolCalls))
+
 		for _, tc := range msg.ToolCalls {
 			expectedIDs[tc.ID] = true
 		}
+
 		needed := len(expectedIDs)
 
 		// Peek ahead: the next `needed` messages must all be tool results with matching IDs
+
 		groupOK := true
+
 		if i+needed >= len(history) {
 			groupOK = false
 		} else {
 			for j := 0; j < needed; j++ {
 				next := history[i+1+j]
+
 				if next.Role != "tool" || !expectedIDs[next.ToolCallID] {
 					groupOK = false
+
 					break
 				}
 			}
@@ -332,14 +435,19 @@ func SanitizeHistory(history []providers.Message) ([]providers.Message, int) {
 
 		if groupOK {
 			// Copy assistant + all tool results
+
 			result = append(result, msg)
+
 			for j := 0; j < needed; j++ {
 				result = append(result, history[i+1+j])
 			}
+
 			i += 1 + needed
 		} else {
 			// Skip the broken assistant message; tool results will be skipped
+
 			// individually when encountered (the "stray tool result" check above)
+
 			i++
 		}
 	}
@@ -348,37 +456,54 @@ func SanitizeHistory(history []providers.Message) ([]providers.Message, int) {
 }
 
 // SetHistory updates the messages of a session.
+
 func (sm *SessionManager) SetHistory(key string, history []providers.Message) {
 	sm.mu.Lock()
+
 	defer sm.mu.Unlock()
 
 	session, ok := sm.sessions[key]
+
 	if ok {
 		// Create a deep copy to strictly isolate internal state
+
 		// from the caller's slice.
+
 		msgs := make([]providers.Message, len(history))
+
 		copy(msgs, history)
+
 		session.Messages = msgs
+
 		session.Updated = time.Now()
 	}
 }
 
 // MarkDirty marks a session key for deferred persistence.
+
 // The session will be written to disk on the next periodic flush or on Close().
+
 func (sm *SessionManager) MarkDirty(key string) {
 	sm.dirtyMu.Lock()
+
 	sm.dirtyKeys[key] = true
+
 	sm.dirtyMu.Unlock()
 }
 
 // FlushDirty writes all dirty sessions to disk.
+
 func (sm *SessionManager) FlushDirty() {
 	sm.dirtyMu.Lock()
+
 	keys := make([]string, 0, len(sm.dirtyKeys))
+
 	for k := range sm.dirtyKeys {
 		keys = append(keys, k)
 	}
+
 	sm.dirtyKeys = make(map[string]bool)
+
 	sm.dirtyMu.Unlock()
 
 	for _, k := range keys {
@@ -387,24 +512,34 @@ func (sm *SessionManager) FlushDirty() {
 }
 
 // Close stops the background flush goroutine and writes all dirty sessions.
+
 func (sm *SessionManager) Close() {
 	select {
 	case <-sm.done:
+
 		return // already closed
+
 	default:
 	}
+
 	close(sm.done)
+
 	sm.FlushDirty()
 }
 
 func (sm *SessionManager) flushLoop() {
 	ticker := time.NewTicker(5 * time.Minute)
+
 	defer ticker.Stop()
+
 	for {
 		select {
 		case <-ticker.C:
+
 			sm.FlushDirty()
+
 		case <-sm.done:
+
 			return
 		}
 	}

pkg/session/manager_test.go

@@ -10,20 +10,27 @@ import (
 
 func TestSanitizeFilename(t *testing.T) {
 	tests := []struct {
-		input    string
+		input string
+
 		expected string
 	}{
 		{"simple", "simple"},
+
 		{"telegram:123456", "telegram_123456"},
+
 		{"discord:987654321", "discord_987654321"},
+
 		{"slack:C01234", "slack_C01234"},
+
 		{"no-colons-here", "no-colons-here"},
+
 		{"multiple:colons:here", "multiple_colons_here"},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.input, func(t *testing.T) {
 			got := sanitizeFilename(tt.input)
+
 			if got != tt.expected {
 				t.Errorf("sanitizeFilename(%q) = %q, want %q", tt.input, got, tt.expected)
 			}
@@ -33,30 +40,41 @@ func TestSanitizeFilename(t *testing.T) {
 
 func TestSave_WithColonInKey(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	sm := NewSessionManager(tmpDir)
 
 	// Create a session with a key containing colon (typical channel session key).
+
 	key := "telegram:123456"
+
 	sm.GetOrCreate(key)
+
 	sm.AddMessage(key, "user", "hello")
 
 	// Save should succeed even though the key contains ':'
+
 	if err := sm.Save(key); err != nil {
 		t.Fatalf("Save(%q) failed: %v", key, err)
 	}
 
 	// The file on disk should use sanitized name.
+
 	expectedFile := filepath.Join(tmpDir, "telegram_123456.json")
+
 	if _, err := os.Stat(expectedFile); os.IsNotExist(err) {
 		t.Fatalf("expected session file %s to exist", expectedFile)
 	}
 
 	// Load into a fresh manager and verify the session round-trips.
+
 	sm2 := NewSessionManager(tmpDir)
+
 	history := sm2.GetHistory(key)
+
 	if len(history) != 1 {
 		t.Fatalf("expected 1 message after reload, got %d", len(history))
 	}
+
 	if history[0].Content != "hello" {
 		t.Errorf("expected message content %q, got %q", "hello", history[0].Content)
 	}
@@ -65,19 +83,27 @@ func TestSave_WithColonInKey(t *testing.T) {
 func TestSanitizeHistory_OrphanedToolCall(t *testing.T) {
 	history := []providers.Message{
 		{Role: "user", Content: "hello"},
+
 		{Role: "assistant", Content: "sure", ToolCalls: []providers.ToolCall{
 			{ID: "call_1", Name: "exec"},
+
 			{ID: "call_2", Name: "list_dir"},
 		}},
+
 		{Role: "tool", Content: "ok", ToolCallID: "call_1"},
+
 		// Missing tool result for call_2 → orphaned
+
 	}
 
 	sanitized, removed := SanitizeHistory(history)
+
 	if removed == 0 {
 		t.Fatal("expected orphaned messages to be removed")
 	}
+
 	// After sanitization, only the user message should remain
+
 	if len(sanitized) != 1 || sanitized[0].Role != "user" {
 		t.Errorf("expected [user], got %d messages", len(sanitized))
 	}
@@ -85,25 +111,36 @@ func TestSanitizeHistory_OrphanedToolCall(t *testing.T) {
 
 func TestSanitizeHistory_InterleavedMessages(t *testing.T) {
 	// Simulates session collision: a user message got interleaved between
+
 	// an assistant tool call and its tool result
+
 	history := []providers.Message{
 		{Role: "user", Content: "first"},
+
 		{Role: "assistant", Content: "ok", ToolCalls: []providers.ToolCall{
 			{ID: "call_1", Name: "exec"},
 		}},
-		{Role: "user", Content: "collision!"},               // ← interleaved from other session
+
+		{Role: "user", Content: "collision!"}, // ← interleaved from other session
+
 		{Role: "tool", Content: "ok", ToolCallID: "call_1"}, // ← out of order
+
 		{Role: "assistant", Content: "done"},
 	}
 
 	sanitized, removed := SanitizeHistory(history)
+
 	if removed == 0 {
 		t.Fatal("expected interleaved messages to be removed")
 	}
+
 	// Should keep: user("first"), user("collision!"), assistant("done")
+
 	// Should remove: assistant(call_1), tool(call_1)
+
 	if len(sanitized) != 3 {
 		t.Errorf("expected 3 messages, got %d", len(sanitized))
+
 		for i, m := range sanitized {
 			t.Logf("  [%d] role=%s content=%q", i, m.Role, m.Content)
 		}
@@ -113,17 +150,22 @@ func TestSanitizeHistory_InterleavedMessages(t *testing.T) {
 func TestSanitizeHistory_CleanHistory(t *testing.T) {
 	history := []providers.Message{
 		{Role: "user", Content: "hello"},
+
 		{Role: "assistant", Content: "sure", ToolCalls: []providers.ToolCall{
 			{ID: "call_1", Name: "exec"},
 		}},
+
 		{Role: "tool", Content: "ok", ToolCallID: "call_1"},
+
 		{Role: "assistant", Content: "done"},
 	}
 
 	sanitized, removed := SanitizeHistory(history)
+
 	if removed != 0 {
 		t.Errorf("expected 0 removed, got %d", removed)
 	}
+
 	if len(sanitized) != 4 {
 		t.Errorf("expected 4 messages, got %d", len(sanitized))
 	}
@@ -132,19 +174,26 @@ func TestSanitizeHistory_CleanHistory(t *testing.T) {
 func TestSanitizeHistory_MultipleToolCalls(t *testing.T) {
 	history := []providers.Message{
 		{Role: "user", Content: "hello"},
+
 		{Role: "assistant", Content: "", ToolCalls: []providers.ToolCall{
 			{ID: "call_1", Name: "exec"},
+
 			{ID: "call_2", Name: "read_file"},
 		}},
+
 		{Role: "tool", Content: "ok", ToolCallID: "call_1"},
+
 		{Role: "tool", Content: "content", ToolCallID: "call_2"},
+
 		{Role: "assistant", Content: "all done"},
 	}
 
 	sanitized, removed := SanitizeHistory(history)
+
 	if removed != 0 {
 		t.Errorf("expected 0 removed, got %d", removed)
 	}
+
 	if len(sanitized) != 5 {
 		t.Errorf("expected 5 messages, got %d", len(sanitized))
 	}
@@ -152,6 +201,7 @@ func TestSanitizeHistory_MultipleToolCalls(t *testing.T) {
 
 func TestSanitizeHistory_Empty(t *testing.T) {
 	sanitized, removed := SanitizeHistory(nil)
+
 	if removed != 0 || sanitized != nil {
 		t.Errorf("expected nil/0, got %v/%d", sanitized, removed)
 	}
@@ -159,11 +209,14 @@ func TestSanitizeHistory_Empty(t *testing.T) {
 
 func TestSave_RejectsPathTraversal(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	sm := NewSessionManager(tmpDir)
 
 	badKeys := []string{"", ".", "..", "foo/bar", "foo\\bar"}
+
 	for _, key := range badKeys {
 		sm.GetOrCreate(key)
+
 		if err := sm.Save(key); err == nil {
 			t.Errorf("Save(%q) should have failed but didn't", key)
 		}

pkg/session/sqlite.go

@@ -14,54 +14,104 @@ const sqliteDriver = "sqlite"
 
 const schema = `
 
+
+
 CREATE TABLE IF NOT EXISTS sessions (
 
+
+
 	key         TEXT PRIMARY KEY,
 
+
+
 	parent_key  TEXT NOT NULL DEFAULT '',
 
+
+
 	fork_turn_id TEXT NOT NULL DEFAULT '',
 
+
+
 	status      TEXT NOT NULL DEFAULT 'active',
 
+
+
 	label       TEXT NOT NULL DEFAULT '',
 
+
+
 	summary     TEXT NOT NULL DEFAULT '',
 
+
+
 	created_at  TEXT NOT NULL,
 
+
+
 	updated_at  TEXT NOT NULL
 
+
+
 );
 
+
+
 CREATE TABLE IF NOT EXISTS turns (
 
+
+
 	id          TEXT PRIMARY KEY,
 
+
+
 	session_key TEXT NOT NULL REFERENCES sessions(key) ON DELETE CASCADE,
 
+
+
 	seq         INTEGER NOT NULL,
 
+
+
 	kind        INTEGER NOT NULL DEFAULT 0,
 
+
+
 	messages    TEXT NOT NULL DEFAULT '[]',
 
+
+
 	origin_key  TEXT NOT NULL DEFAULT '',
 
+
+
 	summary     TEXT NOT NULL DEFAULT '',
 
+
+
 	author      TEXT NOT NULL DEFAULT '',
 
+
+
 	created_at  TEXT NOT NULL,
 
+
+
 	meta        TEXT NOT NULL DEFAULT '{}'
 
+
+
 );
 
+
+
 CREATE INDEX IF NOT EXISTS idx_turns_session_seq ON turns(session_key, seq);
 
+
+
 CREATE INDEX IF NOT EXISTS idx_sessions_parent ON sessions(parent_key);
 
+
+
 `
 
 // SQLiteStore implements SessionStore backed by a single SQLite file.
@@ -130,6 +180,8 @@ func (s *SQLiteStore) Create(key string, opts *CreateOpts) error {
 
 		`INSERT INTO sessions (key, parent_key, fork_turn_id, label, created_at, updated_at)
 
+
+
 		 VALUES (?, ?, ?, ?, ?, ?)`,
 
 		key, parentKey, forkTurnID, label, now, now,
@@ -143,6 +195,8 @@ func (s *SQLiteStore) Get(key string) (*SessionInfo, error) {
 
 		`SELECT key, parent_key, fork_turn_id, status, label, summary, created_at, updated_at
 
+
+
 		 FROM sessions WHERE key = ?`, key,
 	)
 
@@ -291,6 +345,8 @@ func (s *SQLiteStore) Append(sessionKey string, turn *Turn) error {
 
 		`INSERT INTO turns (id, session_key, seq, kind, messages, origin_key, summary, author, created_at, meta)
 
+
+
 		 VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
 
 		turn.ID, sessionKey, turn.Seq, int(turn.Kind),
@@ -315,6 +371,8 @@ func (s *SQLiteStore) Turns(sessionKey string, sinceSeq int) ([]*Turn, error) {
 
 		`SELECT id, session_key, seq, kind, messages, origin_key, summary, author, created_at, meta
 
+
+
 		 FROM turns WHERE session_key = ? AND seq > ? ORDER BY seq`,
 
 		sessionKey, sinceSeq,
@@ -379,6 +437,8 @@ func (s *SQLiteStore) LastTurn(sessionKey string) (*Turn, error) {
 
 		`SELECT id, session_key, seq, kind, messages, origin_key, summary, author, created_at, meta
 
+
+
 		 FROM turns WHERE session_key = ? ORDER BY seq DESC LIMIT 1`,
 
 		sessionKey,

pkg/session/types.go

@@ -19,6 +19,15 @@ const (
 
 )
 
+// Escalation turn kinds — explicit values to keep stable across versions.
+
+const (
+	TurnQuestion TurnKind = 10 // Subagent → conductor question (escalation)
+
+	TurnPlanSubmit TurnKind = 11 // Subagent plan submission for review
+
+)
+
 // Turn represents a single conversation turn persisted in the store.
 
 type Turn struct {

pkg/tools/answer_subagent.go

@@ -0,0 +1,138 @@
+package tools
+
+import (
+	"context"
+	"fmt"
+)
+
+// AnswerSubagentTool allows the conductor to answer a subagent's question.
+
+type AnswerSubagentTool struct {
+	manager *SubagentManager
+}
+
+func NewAnswerSubagentTool(manager *SubagentManager) *AnswerSubagentTool {
+	return &AnswerSubagentTool{manager: manager}
+}
+
+func (t *AnswerSubagentTool) Name() string { return "answer_subagent" }
+
+func (t *AnswerSubagentTool) Description() string {
+	return "Answer a subagent's question or escalation. The subagent is blocked waiting for your response."
+}
+
+func (t *AnswerSubagentTool) Parameters() map[string]any {
+	return map[string]any{
+		"type": "object",
+
+		"properties": map[string]any{
+			"task_id": map[string]any{
+				"type": "string",
+
+				"description": "The task ID of the subagent (e.g. subagent-1)",
+			},
+
+			"answer": map[string]any{
+				"type": "string",
+
+				"description": "Your answer to the subagent's question",
+			},
+		},
+
+		"required": []string{"task_id", "answer"},
+	}
+}
+
+func (t *AnswerSubagentTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
+	taskID, _ := args["task_id"].(string)
+
+	if taskID == "" {
+		return ErrorResult("required parameter \"task_id\" (string) is missing")
+	}
+
+	answer, _ := args["answer"].(string)
+
+	if answer == "" {
+		return ErrorResult("required parameter \"answer\" (string) is missing")
+	}
+
+	if t.manager == nil {
+		return ErrorResult("subagent manager not available")
+	}
+
+	if err := t.manager.AnswerQuestion(taskID, answer); err != nil {
+		return ErrorResult(fmt.Sprintf("failed to answer subagent: %v", err))
+	}
+
+	return &ToolResult{
+		ForLLM: fmt.Sprintf("Answer sent to %s.", taskID),
+
+		ForUser: fmt.Sprintf("Answered %s", taskID),
+	}
+}
+
+// ReviewSubagentPlanTool allows the conductor to approve/reject a subagent's plan.
+
+type ReviewSubagentPlanTool struct {
+	manager *SubagentManager
+}
+
+func NewReviewSubagentPlanTool(manager *SubagentManager) *ReviewSubagentPlanTool {
+	return &ReviewSubagentPlanTool{manager: manager}
+}
+
+func (t *ReviewSubagentPlanTool) Name() string { return "review_subagent_plan" }
+
+func (t *ReviewSubagentPlanTool) Description() string {
+	return "Approve or reject a subagent's execution plan. Use decision 'approved' to approve, or provide rejection feedback."
+}
+
+func (t *ReviewSubagentPlanTool) Parameters() map[string]any {
+	return map[string]any{
+		"type": "object",
+
+		"properties": map[string]any{
+			"task_id": map[string]any{
+				"type": "string",
+
+				"description": "The task ID of the subagent (e.g. subagent-1)",
+			},
+
+			"decision": map[string]any{
+				"type": "string",
+
+				"description": "Decision: 'approved' to approve, or rejection feedback text",
+			},
+		},
+
+		"required": []string{"task_id", "decision"},
+	}
+}
+
+func (t *ReviewSubagentPlanTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
+	taskID, _ := args["task_id"].(string)
+
+	if taskID == "" {
+		return ErrorResult("required parameter \"task_id\" (string) is missing")
+	}
+
+	decision, _ := args["decision"].(string)
+
+	if decision == "" {
+		return ErrorResult("required parameter \"decision\" (string) is missing")
+	}
+
+	if t.manager == nil {
+		return ErrorResult("subagent manager not available")
+	}
+
+	if err := t.manager.AnswerQuestion(taskID, decision); err != nil {
+		return ErrorResult(fmt.Sprintf("failed to send review decision: %v", err))
+	}
+
+	return &ToolResult{
+		ForLLM: fmt.Sprintf("Review decision '%s' sent to %s.", decision, taskID),
+
+		ForUser: fmt.Sprintf("Reviewed %s: %s", taskID, decision),
+	}
+}

pkg/tools/ask_conductor.go

@@ -0,0 +1,110 @@
+package tools
+
+import (
+	"context"
+	"fmt"
+)
+
+// AskConductorTool allows a subagent to ask the conductor a question.
+
+// The subagent blocks until the conductor answers via AnswerSubagentTool.
+
+type AskConductorTool struct {
+	taskID string
+
+	conductorKey string
+
+	subagentKey string
+
+	outCh chan<- ContainerMessage
+
+	inCh <-chan string
+
+	recorder SessionRecorder
+}
+
+func NewAskConductorTool(
+	taskID, conductorKey, subagentKey string,
+
+	outCh chan<- ContainerMessage,
+
+	inCh <-chan string,
+
+	recorder SessionRecorder,
+) *AskConductorTool {
+	return &AskConductorTool{
+		taskID: taskID,
+
+		conductorKey: conductorKey,
+
+		subagentKey: subagentKey,
+
+		outCh: outCh,
+
+		inCh: inCh,
+
+		recorder: recorder,
+	}
+}
+
+func (t *AskConductorTool) Name() string { return "ask_conductor" }
+
+func (t *AskConductorTool) Description() string {
+	return "Ask the conductor a clarifying question. Blocks until the conductor responds. Use when you need guidance or a decision before proceeding."
+}
+
+func (t *AskConductorTool) Parameters() map[string]any {
+	return map[string]any{
+		"type": "object",
+
+		"properties": map[string]any{
+			"question": map[string]any{
+				"type": "string",
+
+				"description": "The question to ask the conductor",
+			},
+		},
+
+		"required": []string{"question"},
+	}
+}
+
+func (t *AskConductorTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
+	question, ok := args["question"].(string)
+
+	if !ok || question == "" {
+		return ErrorResult("required parameter \"question\" (string) is missing")
+	}
+
+	// Fire-and-forget: record question in session DAG.
+
+	if t.recorder != nil {
+		_ = t.recorder.RecordQuestion(t.conductorKey, t.subagentKey, t.taskID, question)
+	}
+
+	// Send question to conductor (blocking with ctx).
+
+	select {
+	case t.outCh <- ContainerMessage{Type: "question", Content: question, TaskID: t.taskID}:
+
+	case <-ctx.Done():
+
+		return ErrorResult(fmt.Sprintf("context canceled while sending question: %v", ctx.Err()))
+	}
+
+	// Wait for conductor's answer.
+
+	select {
+	case answer := <-t.inCh:
+
+		return &ToolResult{
+			ForLLM: fmt.Sprintf("Conductor answered: %s", answer),
+
+			ForUser: answer,
+		}
+
+	case <-ctx.Done():
+
+		return ErrorResult(fmt.Sprintf("context canceled while waiting for answer: %v", ctx.Err()))
+	}
+}

pkg/tools/ask_conductor_test.go

@@ -0,0 +1,77 @@
+package tools
+
+import (
+	"context"
+	"testing"
+	"time"
+)
+
+func TestAskConductorTool_Execute(t *testing.T) {
+	outCh := make(chan ContainerMessage, 4)
+
+	inCh := make(chan string, 1)
+
+	tool := NewAskConductorTool("subagent-1", "conductor:main", "subagent:subagent-1", outCh, inCh, nil)
+
+	if tool.Name() != "ask_conductor" {
+		t.Errorf("Name() = %q, want %q", tool.Name(), "ask_conductor")
+	}
+
+	// Simulate conductor answering in background.
+
+	go func() {
+		msg := <-outCh
+
+		if msg.Type != "question" {
+			t.Errorf("msg.Type = %q, want %q", msg.Type, "question")
+		}
+
+		if msg.Content != "What port?" {
+			t.Errorf("msg.Content = %q, want %q", msg.Content, "What port?")
+		}
+
+		inCh <- "Use port 8080"
+	}()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+
+	defer cancel()
+
+	result := tool.Execute(ctx, map[string]any{"question": "What port?"})
+
+	if result.IsError {
+		t.Fatalf("unexpected error: %s", result.ForLLM)
+	}
+
+	if result.ForUser != "Use port 8080" {
+		t.Errorf("ForUser = %q, want %q", result.ForUser, "Use port 8080")
+	}
+}
+
+func TestAskConductorTool_MissingQuestion(t *testing.T) {
+	tool := NewAskConductorTool("subagent-1", "conductor:main", "subagent:subagent-1", nil, nil, nil)
+
+	result := tool.Execute(context.Background(), map[string]any{})
+
+	if !result.IsError {
+		t.Error("expected error for missing question")
+	}
+}
+
+func TestAskConductorTool_ContextCanceled(t *testing.T) {
+	outCh := make(chan ContainerMessage) // unbuffered, will block
+
+	inCh := make(chan string)
+
+	tool := NewAskConductorTool("subagent-1", "conductor:main", "subagent:subagent-1", outCh, inCh, nil)
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	cancel() // cancel immediately
+
+	result := tool.Execute(ctx, map[string]any{"question": "test?"})
+
+	if !result.IsError {
+		t.Error("expected error on canceled context")
+	}
+}

pkg/tools/bg_monitor.go

@@ -10,17 +10,21 @@ import (
 )
 
 const (
-	bgWatchPollInterval   = 100 * time.Millisecond
+	bgWatchPollInterval = 100 * time.Millisecond
+
 	bgWatchDefaultTimeout = 30 * time.Second
-	bgTailDefaultLines    = 20
+
+	bgTailDefaultLines = 20
 )
 
 // BgMonitorTool monitors and inspects background processes managed by ExecTool.
+
 type BgMonitorTool struct {
 	exec *ExecTool
 }
 
 // NewBgMonitorTool creates a new BgMonitorTool that accesses bg processes from the given ExecTool.
+
 func NewBgMonitorTool(exec *ExecTool) *BgMonitorTool {
 	return &BgMonitorTool{exec: exec}
 }
@@ -36,95 +40,130 @@ func (t *BgMonitorTool) Description() string {
 func (t *BgMonitorTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"action": map[string]any{
-				"type":        "string",
-				"enum":        []string{"list", "watch", "tail"},
+				"type": "string",
+
+				"enum": []string{"list", "watch", "tail"},
+
 				"description": "Action: 'list' all bg processes, 'watch' for a pattern in output, 'tail' recent output lines.",
 			},
+
 			"bg_id": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Background process ID (e.g. 'bg-1'). Required for watch and tail.",
 			},
+
 			"pattern": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Regex pattern to watch for in output (used with action='watch').",
 			},
+
 			"lines": map[string]any{
-				"type":        "number",
+				"type": "number",
+
 				"description": "Number of recent lines to return (used with action='tail', default 20).",
 			},
+
 			"watch_timeout": map[string]any{
-				"type":        "number",
+				"type": "number",
+
 				"description": "Timeout in seconds for watch action (default 30).",
 			},
 		},
+
 		"required": []string{"action"},
 	}
 }
 
 func (t *BgMonitorTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	action, _ := args["action"].(string)
+
 	switch action {
 	case "list":
+
 		return t.actionList()
+
 	case "watch":
+
 		return t.actionWatch(ctx, args)
+
 	case "tail":
+
 		return t.actionTail(args)
+
 	default:
+
 		return ErrorResult(fmt.Sprintf("unknown action %q (use 'list', 'watch', or 'tail')", action))
 	}
 }
 
 func (t *BgMonitorTool) actionList() *ToolResult {
 	procs := t.exec.BgProcesses()
+
 	if len(procs) == 0 {
 		return &ToolResult{
-			ForLLM:  "No background processes.",
+			ForLLM: "No background processes.",
+
 			ForUser: "No background processes.",
 		}
 	}
 
 	ids := make([]string, 0, len(procs))
+
 	for id := range procs {
 		ids = append(ids, id)
 	}
+
 	sort.Strings(ids)
 
 	var sb strings.Builder
+
 	sb.WriteString("Background Processes:\n\n")
+
 	for _, id := range ids {
 		bp := procs[id]
+
 		if bp.isRunning() {
 			uptime := time.Since(bp.startedAt).Truncate(time.Second)
+
 			fmt.Fprintf(&sb, "  [%s] pid=%d running  (uptime: %s, max: %s) %s\n",
+
 				id, bp.pid, uptime, getBgMaxLifetime(), bp.command)
 		} else {
 			ran := time.Since(bp.startedAt).Truncate(time.Second)
+
 			if bp.exitErr != nil {
 				fmt.Fprintf(&sb, "  [%s] pid=%d exited=err (ran: %s) %s\n",
+
 					id, bp.pid, ran, bp.command)
 			} else {
 				fmt.Fprintf(&sb, "  [%s] pid=%d exited=0 (ran: %s) %s\n",
+
 					id, bp.pid, ran, bp.command)
 			}
 		}
 	}
 
 	return &ToolResult{
-		ForLLM:  sb.String(),
+		ForLLM: sb.String(),
+
 		ForUser: sb.String(),
 	}
 }
 
 func (t *BgMonitorTool) actionWatch(ctx context.Context, args map[string]any) *ToolResult {
 	bgID, _ := args["bg_id"].(string)
+
 	if bgID == "" {
 		return ErrorResult("bg_id is required for watch action")
 	}
 
 	patternStr, _ := args["pattern"].(string)
+
 	if patternStr == "" {
 		return ErrorResult("pattern is required for watch action")
 	}
@@ -135,64 +174,93 @@ func (t *BgMonitorTool) actionWatch(ctx context.Context, args map[string]any) *T
 	}
 
 	timeout := bgWatchDefaultTimeout
+
 	if t, ok := args["watch_timeout"].(float64); ok && t > 0 {
 		timeout = time.Duration(t) * time.Second
 	}
 
 	procs := t.exec.BgProcesses()
+
 	bp, ok := procs[bgID]
+
 	if !ok {
 		return ErrorResult(fmt.Sprintf("background process %q not found", bgID))
 	}
 
 	deadline := time.After(timeout)
+
 	ticker := time.NewTicker(bgWatchPollInterval)
+
 	defer ticker.Stop()
 
 	for {
 		// Check for pattern match
+
 		if match := bp.output.Match(pattern); match != "" {
 			return &ToolResult{
-				ForLLM:  fmt.Sprintf("Match found in [%s]: %s", bgID, match),
+				ForLLM: fmt.Sprintf("Match found in [%s]: %s", bgID, match),
+
 				ForUser: fmt.Sprintf("Match found in [%s]: %s", bgID, match),
 			}
 		}
 
 		// Check if process exited
+
 		if !bp.isRunning() {
 			output := bp.output.String()
+
 			tail := lastNLines(output, 10)
+
 			var sb strings.Builder
+
 			fmt.Fprintf(&sb, "Process %s exited before pattern matched.\n", bgID)
+
 			if bp.exitErr != nil {
 				fmt.Fprintf(&sb, "Exit: %v\n", bp.exitErr)
 			} else {
 				fmt.Fprintf(&sb, "Exit: 0\n")
 			}
+
 			fmt.Fprintf(&sb, "\nLast output:\n%s", tail)
+
 			return &ToolResult{
-				ForLLM:  sb.String(),
+				ForLLM: sb.String(),
+
 				ForUser: sb.String(),
+
 				IsError: true,
 			}
 		}
 
 		select {
 		case <-deadline:
+
 			// Timeout
+
 			output := bp.output.String()
+
 			tail := lastNLines(output, 10)
+
 			var sb strings.Builder
+
 			fmt.Fprintf(&sb, "Watch timed out after %s waiting for pattern %q in [%s].\n", timeout, patternStr, bgID)
+
 			fmt.Fprintf(&sb, "\nLast output:\n%s", tail)
+
 			return &ToolResult{
-				ForLLM:  sb.String(),
+				ForLLM: sb.String(),
+
 				ForUser: sb.String(),
+
 				IsError: true,
 			}
+
 		case <-ctx.Done():
+
 			return ErrorResult("watch canceled")
+
 		case <-ticker.C:
+
 			// Continue polling
 		}
 	}
@@ -200,17 +268,21 @@ func (t *BgMonitorTool) actionWatch(ctx context.Context, args map[string]any) *T
 
 func (t *BgMonitorTool) actionTail(args map[string]any) *ToolResult {
 	bgID, _ := args["bg_id"].(string)
+
 	if bgID == "" {
 		return ErrorResult("bg_id is required for tail action")
 	}
 
 	n := bgTailDefaultLines
+
 	if lines, ok := args["lines"].(float64); ok && lines > 0 {
 		n = int(lines)
 	}
 
 	procs := t.exec.BgProcesses()
+
 	bp, ok := procs[bgID]
+
 	if !ok {
 		return ErrorResult(fmt.Sprintf("background process %q not found", bgID))
 	}
@@ -218,7 +290,9 @@ func (t *BgMonitorTool) actionTail(args map[string]any) *ToolResult {
 	lines := bp.output.Lines(n)
 
 	var sb strings.Builder
+
 	fmt.Fprintf(&sb, "[%s] pid=%d %s\n", bp.id, bp.pid, bp.command)
+
 	if bp.isRunning() {
 		fmt.Fprintf(&sb, "Status: running\n")
 	} else {
@@ -228,7 +302,9 @@ func (t *BgMonitorTool) actionTail(args map[string]any) *ToolResult {
 			fmt.Fprintf(&sb, "Status: exited=0\n")
 		}
 	}
+
 	fmt.Fprintf(&sb, "\nLast %d lines:\n", n)
+
 	for _, line := range lines {
 		fmt.Fprintf(&sb, "%s\n", line)
 	}
@@ -238,19 +314,24 @@ func (t *BgMonitorTool) actionTail(args map[string]any) *ToolResult {
 	}
 
 	return &ToolResult{
-		ForLLM:  sb.String(),
+		ForLLM: sb.String(),
+
 		ForUser: sb.String(),
 	}
 }
 
 // lastNLines returns the last n lines from a string.
+
 func lastNLines(s string, n int) string {
 	lines := strings.Split(s, "\n")
+
 	if len(lines) > 0 && lines[len(lines)-1] == "" {
 		lines = lines[:len(lines)-1]
 	}
+
 	if n >= len(lines) {
 		return strings.Join(lines, "\n")
 	}
+
 	return strings.Join(lines[len(lines)-n:], "\n")
 }

pkg/tools/bg_monitor_test.go

@@ -10,64 +10,83 @@ import (
 
 func TestBgMonitor_List(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	// List with no processes
+
 	result := monitor.Execute(context.Background(), map[string]any{"action": "list"})
+
 	if result.IsError {
 		t.Fatalf("unexpected error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "No background") {
 		t.Errorf("expected 'No background' message, got: %s", result.ForLLM)
 	}
 
 	// Start two bg processes
+
 	var cmd1, cmd2 string
+
 	if runtime.GOOS == "windows" {
 		cmd1 = "Start-Sleep -Seconds 30"
+
 		cmd2 = "Start-Sleep -Seconds 30"
 	} else {
 		cmd1 = "sleep 30"
+
 		cmd2 = "sleep 30"
 	}
 
 	r1 := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd1,
+		"command": cmd1,
+
 		"background": true,
 	})
+
 	if r1.IsError {
 		t.Fatalf("failed to start bg-1: %s", r1.ForLLM)
 	}
 
 	r2 := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd2,
+		"command": cmd2,
+
 		"background": true,
 	})
+
 	if r2.IsError {
 		t.Fatalf("failed to start bg-2: %s", r2.ForLLM)
 	}
 
 	// List should show both
+
 	result = monitor.Execute(context.Background(), map[string]any{"action": "list"})
+
 	if result.IsError {
 		t.Fatalf("unexpected error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "bg-1") {
 		t.Errorf("expected bg-1 in list, got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "bg-2") {
 		t.Errorf("expected bg-2 in list, got: %s", result.ForLLM)
 	}
 
 	// Cleanup
+
 	tool.Shutdown()
 }
 
 func TestBgMonitor_Watch_Match(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	var cmd string
+
 	if runtime.GOOS == "windows" {
 		cmd = "Write-Output 'Server ready on port 3000'; Start-Sleep -Seconds 30"
 	} else {
@@ -75,26 +94,35 @@ func TestBgMonitor_Watch_Match(t *testing.T) {
 	}
 
 	r := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd,
+		"command": cmd,
+
 		"background": true,
 	})
+
 	if r.IsError {
 		t.Fatalf("failed to start bg: %s", r.ForLLM)
 	}
 
 	// Watch for "ready" pattern — should match quickly
+
 	result := monitor.Execute(context.Background(), map[string]any{
-		"action":        "watch",
-		"bg_id":         "bg-1",
-		"pattern":       "ready",
+		"action": "watch",
+
+		"bg_id": "bg-1",
+
+		"pattern": "ready",
+
 		"watch_timeout": float64(10),
 	})
+
 	if result.IsError {
 		t.Fatalf("expected watch to match, got error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "Match found") {
 		t.Errorf("expected 'Match found' message, got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "ready") {
 		t.Errorf("expected match to contain 'ready', got: %s", result.ForLLM)
 	}
@@ -104,9 +132,11 @@ func TestBgMonitor_Watch_Match(t *testing.T) {
 
 func TestBgMonitor_Watch_Timeout(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	var cmd string
+
 	if runtime.GOOS == "windows" {
 		cmd = "Start-Sleep -Seconds 30"
 	} else {
@@ -114,23 +144,31 @@ func TestBgMonitor_Watch_Timeout(t *testing.T) {
 	}
 
 	r := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd,
+		"command": cmd,
+
 		"background": true,
 	})
+
 	if r.IsError {
 		t.Fatalf("failed to start bg: %s", r.ForLLM)
 	}
 
 	// Watch for a pattern that won't appear, with short timeout
+
 	result := monitor.Execute(context.Background(), map[string]any{
-		"action":        "watch",
-		"bg_id":         "bg-1",
-		"pattern":       "never_going_to_match",
+		"action": "watch",
+
+		"bg_id": "bg-1",
+
+		"pattern": "never_going_to_match",
+
 		"watch_timeout": float64(1),
 	})
+
 	if !result.IsError {
 		t.Fatalf("expected watch to timeout with error, got success: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "timed out") {
 		t.Errorf("expected 'timed out' message, got: %s", result.ForLLM)
 	}
@@ -140,9 +178,11 @@ func TestBgMonitor_Watch_Timeout(t *testing.T) {
 
 func TestBgMonitor_Watch_ProcessExit(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	var cmd string
+
 	if runtime.GOOS == "windows" {
 		cmd = "Write-Output 'done quickly'"
 	} else {
@@ -150,26 +190,35 @@ func TestBgMonitor_Watch_ProcessExit(t *testing.T) {
 	}
 
 	r := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd,
+		"command": cmd,
+
 		"background": true,
 	})
+
 	if r.IsError {
 		t.Fatalf("failed to start bg: %s", r.ForLLM)
 	}
 
 	// Wait a bit for the process to exit
+
 	time.Sleep(4 * time.Second)
 
 	// Watch for a pattern that doesn't match — process should have exited
+
 	result := monitor.Execute(context.Background(), map[string]any{
-		"action":        "watch",
-		"bg_id":         "bg-1",
-		"pattern":       "never_match",
+		"action": "watch",
+
+		"bg_id": "bg-1",
+
+		"pattern": "never_match",
+
 		"watch_timeout": float64(5),
 	})
+
 	if !result.IsError {
 		t.Fatalf("expected error when process exits, got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "exited") {
 		t.Errorf("expected 'exited' message, got: %s", result.ForLLM)
 	}
@@ -179,9 +228,11 @@ func TestBgMonitor_Watch_ProcessExit(t *testing.T) {
 
 func TestBgMonitor_Tail(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	var cmd string
+
 	if runtime.GOOS == "windows" {
 		cmd = "1..5 | ForEach-Object { Write-Output \"line $_\" }; Start-Sleep -Seconds 30"
 	} else {
@@ -189,25 +240,33 @@ func TestBgMonitor_Tail(t *testing.T) {
 	}
 
 	r := tool.Execute(context.Background(), map[string]any{
-		"command":    cmd,
+		"command": cmd,
+
 		"background": true,
 	})
+
 	if r.IsError {
 		t.Fatalf("failed to start bg: %s", r.ForLLM)
 	}
 
 	// Wait for initial output to be captured
+
 	time.Sleep(4 * time.Second)
 
 	// Tail last 3 lines
+
 	result := monitor.Execute(context.Background(), map[string]any{
 		"action": "tail",
-		"bg_id":  "bg-1",
-		"lines":  float64(3),
+
+		"bg_id": "bg-1",
+
+		"lines": float64(3),
 	})
+
 	if result.IsError {
 		t.Fatalf("unexpected error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "line 5") {
 		t.Errorf("expected tail to contain 'line 5', got: %s", result.ForLLM)
 	}
@@ -217,12 +276,15 @@ func TestBgMonitor_Tail(t *testing.T) {
 
 func TestBgMonitor_InvalidAction(t *testing.T) {
 	tool, _ := NewExecTool("", false)
+
 	monitor := NewBgMonitorTool(tool)
 
 	result := monitor.Execute(context.Background(), map[string]any{"action": "invalid"})
+
 	if !result.IsError {
 		t.Fatalf("expected error for invalid action")
 	}
+
 	if !strings.Contains(result.ForLLM, "unknown action") {
 		t.Errorf("expected 'unknown action' message, got: %s", result.ForLLM)
 	}

pkg/tools/createpr.go

@@ -10,27 +10,42 @@ import (
 
 const (
 	ciPollInterval = 30 * time.Second
-	ciPollTimeout  = 15 * time.Minute
+
+	ciPollTimeout = 15 * time.Minute
 )
 
 // CreatePRTool creates a GitHub pull request from the current worktree branch.
+
 //
+
 // Safety invariants:
+
 //   - Only works inside a worktree (WorktreeInfo must be in context)
+
 //   - Base branch is auto-detected from WorktreeInfo.BaseBranch
+
 //   - Requires the branch to be already pushed (use git_push first)
+
 //   - Checks for merge conflicts with base before creating
+
 //   - Uses `gh pr create` under the hood
+
 //
+
 // Async behavior:
+
 //   - PR creation itself is synchronous and returns immediately with the PR URL
+
 //   - If CI runs are triggered, a background goroutine polls `gh pr checks`
+
 //     and calls the AsyncCallback when CI completes (pass or fail)
+
 type CreatePRTool struct {
 	callback AsyncCallback
 }
 
 // NewCreatePRTool creates a CreatePRTool.
+
 func NewCreatePRTool() *CreatePRTool {
 	return &CreatePRTool{}
 }
@@ -38,122 +53,187 @@ func NewCreatePRTool() *CreatePRTool {
 func (t *CreatePRTool) Name() string { return "create_pr" }
 
 // SetCallback implements AsyncTool for CI completion notification.
+
 func (t *CreatePRTool) SetCallback(cb AsyncCallback) {
 	t.callback = cb
 }
 
 func (t *CreatePRTool) Description() string {
 	return "Create a GitHub pull request from the current worktree branch. " +
+
 		"The base branch is auto-detected from the worktree's parent branch. " +
+
 		"The branch must be pushed to origin first (use git_push). " +
+
 		"Checks for merge conflicts with the base branch before creating. " +
+
 		"After PR creation, polls CI status in the background and notifies when complete. " +
+
 		"Requires the `gh` CLI to be installed and authenticated."
 }
 
 func (t *CreatePRTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"title": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Pull request title",
 			},
+
 			"body": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Pull request body/description (supports markdown)",
 			},
+
 			"draft": map[string]any{
-				"type":        "boolean",
+				"type": "boolean",
+
 				"description": "Create as draft PR (default: false)",
 			},
 		},
+
 		"required": []string{"title"},
 	}
 }
 
 func (t *CreatePRTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	wt := WorktreeInfoFromCtx(ctx)
+
 	if wt == nil {
 		return ErrorResult(
+
 			"create_pr requires an active worktree.\n" +
+
 				"This tool can only be used during worktree-based sessions " +
+
 				"(e.g., heartbeat tasks or plan executing phase).\n" +
+
 				"The worktree provides the branch name and base branch for the PR.")
 	}
 
 	branch := wt.Branch
+
 	if branch == "" {
 		return ErrorResult(
+
 			"worktree has no branch name.\n" +
+
 				"The WorktreeInfo was set but Branch is empty. " +
+
 				"This is an internal error — the worktree may not have been created correctly.")
 	}
 
 	baseBranch := wt.BaseBranch
+
 	if baseBranch == "" {
 		baseBranch = "main"
 	}
 
 	title, ok := args["title"].(string)
+
 	if !ok || strings.TrimSpace(title) == "" {
 		return ErrorResult(
+
 			"title is required.\n" +
+
 				"Provide a concise PR title describing the change (e.g., \"Add rate limiter to API endpoints\").")
 	}
 
 	// Verify the branch has been pushed by checking if the remote ref exists
+
 	checkCtx, checkCancel := context.WithTimeout(ctx, 15*time.Second)
+
 	defer checkCancel()
+
 	checkCmd := exec.CommandContext(checkCtx, "git", "ls-remote", "--exit-code", "origin", branch)
+
 	checkCmd.Dir = wt.Path
+
 	if err := checkCmd.Run(); err != nil {
 		return ErrorResult(fmt.Sprintf(
+
 			"branch %q not found on origin.\n"+
+
 				"The branch must be pushed before creating a PR. Use the git_push tool first.\n"+
+
 				"git_push will auto-commit uncommitted changes and push the worktree branch to origin.",
+
 			branch))
 	}
 
 	// Fetch latest base branch and check for merge conflicts
+
 	fetchCtx, fetchCancel := context.WithTimeout(ctx, 30*time.Second)
+
 	defer fetchCancel()
+
 	fetchCmd := exec.CommandContext(fetchCtx, "git", "fetch", "origin", baseBranch)
+
 	fetchCmd.Dir = wt.Path
+
 	if out, err := fetchCmd.CombinedOutput(); err != nil {
 		return ErrorResult(fmt.Sprintf(
+
 			"failed to fetch origin/%s: %s\n%s\n"+
+
 				"Cannot verify merge compatibility without the latest base branch. "+
+
 				"Check network connectivity and that the base branch %q exists on origin.",
+
 			baseBranch, err, strings.TrimSpace(string(out)), baseBranch))
 	}
 
 	// Try a merge dry-run to detect conflicts.
+
 	// merge-tree --write-tree is a plumbing command (Git 2.38+) that performs a
+
 	// three-way merge entirely in-memory without touching the working tree.
+
 	// Exit code 0 = clean merge, non-zero = conflicts detected.
+
 	mergeCtx, mergeCancel := context.WithTimeout(ctx, 30*time.Second)
+
 	defer mergeCancel()
+
 	mergeCmd := exec.CommandContext(mergeCtx, "git", "merge-tree",
+
 		"--write-tree", "--no-messages",
+
 		branch, "origin/"+baseBranch)
+
 	mergeCmd.Dir = wt.RepoRoot
+
 	mergeOut, mergeErr := mergeCmd.CombinedOutput()
+
 	if mergeErr != nil {
 		conflictInfo := strings.TrimSpace(string(mergeOut))
+
 		return ErrorResult(fmt.Sprintf(
+
 			"merge conflict detected between %q and %s.\n"+
+
 				"The PR cannot be created cleanly. Resolve the conflicts in the worktree first, "+
+
 				"then use git_push to push the resolution before retrying create_pr.\n"+
+
 				"Conflict details:\n%s",
+
 			branch, baseBranch, conflictInfo))
 	}
 
 	// Build gh pr create command
+
 	ghArgs := []string{
 		"pr", "create",
+
 		"--base", baseBranch,
+
 		"--head", branch,
+
 		"--title", title,
 	}
 
@@ -168,92 +248,143 @@ func (t *CreatePRTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	}
 
 	prCtx, prCancel := context.WithTimeout(ctx, 30*time.Second)
+
 	defer prCancel()
 
 	cmd := exec.CommandContext(prCtx, "gh", ghArgs...)
+
 	cmd.Dir = wt.RepoRoot
+
 	out, err := cmd.CombinedOutput()
+
 	output := strings.TrimSpace(string(out))
 
 	if err != nil {
 		return ErrorResult(fmt.Sprintf(
+
 			"gh pr create failed: %s\n%s\n"+
+
 				"Possible causes:\n"+
+
 				"- gh CLI not installed or not authenticated (run `gh auth login`)\n"+
+
 				"- A PR already exists for branch %q (check with `gh pr list`)\n"+
+
 				"- Repository not configured as a GitHub remote",
+
 			err, output, branch))
 	}
 
 	prURL := output // gh pr create outputs the PR URL
 
 	// Start background CI polling if callback is set
+
 	if t.callback != nil && prURL != "" {
 		cb := t.callback
+
 		repoRoot := wt.RepoRoot
+
 		go pollCIStatus(repoRoot, prURL, cb)
 	}
 
 	return AsyncResult(fmt.Sprintf(
+
 		"Pull request created: %s\n"+
+
 			"Branch: %s -> %s\n"+
+
 			"CI status will be reported asynchronously when checks complete.",
+
 		prURL, branch, baseBranch))
 }
 
 // pollCIStatus polls `gh pr checks` in the background until all checks
+
 // pass, fail, or the timeout is reached. Reports back via AsyncCallback.
+
 func pollCIStatus(repoRoot, prURL string, callback AsyncCallback) {
 	// Detached context with hard timeout — this goroutine outlives the tool call.
+
 	ctx, cancel := context.WithTimeout(context.Background(), ciPollTimeout)
+
 	defer cancel()
 
 	// Initial wait: CI runs take a few seconds to register after PR creation
+
 	select {
 	case <-time.After(10 * time.Second):
+
 	case <-ctx.Done():
+
 		return
 	}
 
 	ticker := time.NewTicker(ciPollInterval)
+
 	defer ticker.Stop()
 
 	for {
 		status, detail := checkPRChecks(ctx, repoRoot, prURL)
+
 		switch status {
 		case ciStatusPass:
+
 			callback(ctx, NewToolResult(fmt.Sprintf(
+
 				"CI passed for %s\n%s",
+
 				prURL, detail)))
+
 			return
+
 		case ciStatusFail:
+
 			callback(ctx, &ToolResult{
 				ForLLM: fmt.Sprintf(
+
 					"CI failed for %s\n%s\n"+
+
 						"Run `gh run view` for detailed logs.",
+
 					prURL, detail),
+
 				IsError: true,
 			})
+
 			return
+
 		case ciStatusNone:
+
 			callback(ctx, NewToolResult(fmt.Sprintf(
+
 				"No CI checks configured for %s. PR is ready for review.",
+
 				prURL)))
+
 			return
+
 		case ciStatusPending:
+
 			// Still running, continue polling
 		}
 
 		select {
 		case <-ticker.C:
+
 		case <-ctx.Done():
+
 			callback(ctx, &ToolResult{
 				ForLLM: fmt.Sprintf(
+
 					"CI polling timed out after %s for %s.\n"+
+
 						"Checks may still be running. Run `gh pr checks %s` to check.",
+
 					ciPollTimeout, prURL, prURL),
+
 				IsError: true,
 			})
+
 			return
 		}
 	}
@@ -263,36 +394,51 @@ type ciStatus int
 
 const (
 	ciStatusPending ciStatus = iota
+
 	ciStatusPass
+
 	ciStatusFail
+
 	ciStatusNone
 )
 
 // checkPRChecks runs `gh pr checks` and parses the result.
+
 // Returns the aggregate status and raw output for the caller to include.
+
 func checkPRChecks(ctx context.Context, repoRoot, prURL string) (ciStatus, string) {
 	checkCtx, cancel := context.WithTimeout(ctx, 15*time.Second)
+
 	defer cancel()
 
 	cmd := exec.CommandContext(checkCtx, "gh", "pr", "checks", prURL)
+
 	cmd.Dir = repoRoot
+
 	out, err := cmd.CombinedOutput()
+
 	output := strings.TrimSpace(string(out))
 
 	if err != nil {
 		// gh pr checks exits 1 when any check has failed
+
 		if strings.Contains(output, "fail") || strings.Contains(output, "X ") {
 			return ciStatusFail, output
 		}
+
 		// "no checks" case
+
 		if strings.Contains(output, "no checks") || output == "" {
 			return ciStatusNone, ""
 		}
+
 		// Transient error or still pending — keep polling
+
 		return ciStatusPending, output
 	}
 
 	// Exit 0: all checks completed. Check for pending.
+
 	if strings.Contains(output, "pending") || strings.Contains(output, "- ") {
 		return ciStatusPending, output
 	}

pkg/tools/createpr_test.go

@@ -9,191 +9,259 @@ import (
 )
 
 // TestCreatePRTool_NoWorktree verifies that create_pr fails without worktree context.
+
 func TestCreatePRTool_NoWorktree(t *testing.T) {
 	tool := NewCreatePRTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"title": "Test PR",
 	})
+
 	if !result.IsError {
 		t.Fatal("expected error when no worktree in context")
 	}
+
 	assertContains(t, result.ForLLM, "worktree")
+
 	assertContains(t, result.ForLLM, "heartbeat")
 }
 
 // TestCreatePRTool_EmptyBranch verifies that empty branch name is rejected.
+
 func TestCreatePRTool_EmptyBranch(t *testing.T) {
 	tool := NewCreatePRTool()
 
 	ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-		Branch:     "",
+		Branch: "",
+
 		BaseBranch: "main",
-		Path:       t.TempDir(),
-		RepoRoot:   t.TempDir(),
+
+		Path: t.TempDir(),
+
+		RepoRoot: t.TempDir(),
 	})
+
 	result := tool.Execute(ctx, map[string]any{
 		"title": "Test PR",
 	})
+
 	if !result.IsError {
 		t.Fatal("expected error for empty branch")
 	}
+
 	assertContains(t, result.ForLLM, "no branch name")
 }
 
 // TestCreatePRTool_MissingTitle verifies that missing title is rejected.
+
 func TestCreatePRTool_MissingTitle(t *testing.T) {
 	tool := NewCreatePRTool()
 
 	ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-		Branch:     "plan/test",
+		Branch: "plan/test",
+
 		BaseBranch: "main",
-		Path:       t.TempDir(),
-		RepoRoot:   t.TempDir(),
+
+		Path: t.TempDir(),
+
+		RepoRoot: t.TempDir(),
 	})
 
 	tests := []struct {
 		name string
+
 		args map[string]any
 	}{
 		{"no title key", map[string]any{}},
+
 		{"empty title", map[string]any{"title": ""}},
+
 		{"whitespace title", map[string]any{"title": "   "}},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			result := tool.Execute(ctx, tt.args)
+
 			if !result.IsError {
 				t.Fatal("expected error for missing/empty title")
 			}
+
 			assertContains(t, result.ForLLM, "title is required")
 		})
 	}
 }
 
 // TestCreatePRTool_BranchNotPushed verifies the tool checks for remote branch existence.
+
 func TestCreatePRTool_BranchNotPushed(t *testing.T) {
 	tool := NewCreatePRTool()
 
 	ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-		Branch:     "plan/not-pushed",
+		Branch: "plan/not-pushed",
+
 		BaseBranch: "main",
-		Path:       t.TempDir(),
-		RepoRoot:   t.TempDir(),
+
+		Path: t.TempDir(),
+
+		RepoRoot: t.TempDir(),
 	})
+
 	result := tool.Execute(ctx, map[string]any{
 		"title": "Test PR",
 	})
+
 	if !result.IsError {
 		t.Fatal("expected error for unpushed branch")
 	}
+
 	// Should mention git_push as the remedy
+
 	assertContains(t, result.ForLLM, "git_push")
 }
 
 // TestCreatePRTool_DefaultBaseBranch verifies fallback to "main" when BaseBranch is empty.
+
 func TestCreatePRTool_DefaultBaseBranch(t *testing.T) {
 	tool := NewCreatePRTool()
 
 	// With empty BaseBranch, tool should default to "main"
+
 	ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-		Branch:     "plan/test",
+		Branch: "plan/test",
+
 		BaseBranch: "",
-		Path:       t.TempDir(),
-		RepoRoot:   t.TempDir(),
+
+		Path: t.TempDir(),
+
+		RepoRoot: t.TempDir(),
 	})
+
 	result := tool.Execute(ctx, map[string]any{
 		"title": "Test PR",
 	})
+
 	// Will fail at ls-remote (no real repo), but should not fail at baseBranch validation
+
 	if result.IsError && strings.Contains(result.ForLLM, "base branch") {
 		t.Fatal("should not fail on base branch when defaulting to main")
 	}
 }
 
 // TestCreatePRTool_Interface verifies the tool satisfies both Tool and AsyncTool interfaces.
+
 func TestCreatePRTool_Interface(t *testing.T) {
 	var _ Tool = (*CreatePRTool)(nil)
+
 	var _ AsyncTool = (*CreatePRTool)(nil)
 
 	tool := NewCreatePRTool()
+
 	if tool.Name() != "create_pr" {
 		t.Errorf("Name: got %q, want %q", tool.Name(), "create_pr")
 	}
+
 	if tool.Description() == "" {
 		t.Error("Description should not be empty")
 	}
+
 	params := tool.Parameters()
+
 	if params == nil {
 		t.Fatal("Parameters should not be nil")
 	}
 
 	// Verify "title" is required
+
 	required, ok := params["required"].([]string)
+
 	if !ok {
 		t.Fatal("required should be []string")
 	}
+
 	foundTitle := false
+
 	for _, r := range required {
 		if r == "title" {
 			foundTitle = true
 		}
 	}
+
 	if !foundTitle {
 		t.Error("title should be in required parameters")
 	}
 }
 
 // TestCreatePRTool_SetCallback verifies callback is stored.
+
 func TestCreatePRTool_SetCallback(t *testing.T) {
 	tool := NewCreatePRTool()
+
 	if tool.callback != nil {
 		t.Fatal("callback should be nil initially")
 	}
 
 	called := false
+
 	tool.SetCallback(func(ctx context.Context, result *ToolResult) {
 		called = true
 	})
+
 	if tool.callback == nil {
 		t.Fatal("callback should be set after SetCallback")
 	}
+
 	// Verify it's callable (doesn't panic)
+
 	tool.callback(context.Background(), NewToolResult("test"))
+
 	if !called {
 		t.Fatal("callback was not invoked")
 	}
 }
 
 // TestCheckPRChecks_ParseResults tests CI status parsing logic.
+
 func TestCheckPRChecks_ParseResults(t *testing.T) {
 	// This tests the parsing logic conceptually — actual `gh` calls
+
 	// would need integration tests. We verify the status constants exist
+
 	// and the type is usable.
+
 	if ciStatusPending != 0 {
 		t.Error("ciStatusPending should be 0 (default)")
 	}
+
 	if ciStatusPass == ciStatusFail {
 		t.Error("ciStatusPass and ciStatusFail should differ")
 	}
+
 	if ciStatusNone == ciStatusPending {
 		t.Error("ciStatusNone and ciStatusPending should differ")
 	}
 }
 
 // TestAllowedToolsForPreset_GitTools checks git tools are correctly assigned to presets.
+
 func TestAllowedToolsForPreset_GitTools(t *testing.T) {
 	tests := []struct {
-		name         string
-		preset       Preset
-		wantGitPush  bool
+		name string
+
+		preset Preset
+
+		wantGitPush bool
+
 		wantCreatePR bool
 	}{
 		{"scout", PresetScout, false, false},
+
 		{"analyst", PresetAnalyst, false, false},
+
 		{"coder", PresetCoder, true, false},
+
 		{"worker", PresetWorker, true, true},
+
 		{"coordinator", PresetCoordinator, true, true},
 	}
 
@@ -204,6 +272,7 @@ func TestAllowedToolsForPreset_GitTools(t *testing.T) {
 			if got := allowed["git_push"]; got != tt.wantGitPush {
 				t.Errorf("git_push: got %v, want %v", got, tt.wantGitPush)
 			}
+
 			if got := allowed["create_pr"]; got != tt.wantCreatePR {
 				t.Errorf("create_pr: got %v, want %v", got, tt.wantCreatePR)
 			}

pkg/tools/cron.go

@@ -14,25 +14,36 @@ import (
 )
 
 // JobExecutor is the interface for executing cron jobs through the agent
+
 type JobExecutor interface {
 	ProcessDirectWithChannel(ctx context.Context, content, sessionKey, channel, chatID string) (string, error)
 }
 
 // CronTool provides scheduling capabilities for the agent
+
 type CronTool struct {
 	cronService *cron.CronService
-	executor    JobExecutor
-	msgBus      *bus.MessageBus
-	execTool    *ExecTool
-	channel     string
-	chatID      string
-	mu          sync.RWMutex
+
+	executor JobExecutor
+
+	msgBus *bus.MessageBus
+
+	execTool *ExecTool
+
+	channel string
+
+	chatID string
+
+	mu sync.RWMutex
 }
 
 // NewCronTool creates a new CronTool
+
 // execTimeout: 0 means no timeout, >0 sets the timeout duration
+
 func NewCronTool(
 	cronService *cron.CronService, executor JobExecutor, msgBus *bus.MessageBus, workspace string, restrict bool,
+
 	execTimeout time.Duration, config *config.Config,
 ) (*CronTool, error) {
 	execTool, err := NewExecToolWithConfig(workspace, restrict, config)
@@ -41,102 +52,147 @@ func NewCronTool(
 	}
 
 	execTool.SetTimeout(execTimeout)
+
 	return &CronTool{
 		cronService: cronService,
-		executor:    executor,
-		msgBus:      msgBus,
-		execTool:    execTool,
+
+		executor: executor,
+
+		msgBus: msgBus,
+
+		execTool: execTool,
 	}, nil
 }
 
 // Name returns the tool name
+
 func (t *CronTool) Name() string {
 	return "cron"
 }
 
 // Description returns the tool description
+
 func (t *CronTool) Description() string {
 	return "Schedule reminders, tasks, or system commands. IMPORTANT: When user asks to be reminded or scheduled, you MUST call this tool. Use 'at_seconds' for one-time reminders (e.g., 'remind me in 10 minutes' → at_seconds=600). Use 'every_seconds' ONLY for recurring tasks (e.g., 'every 2 hours' → every_seconds=7200). Use 'cron_expr' for complex recurring schedules. Use 'command' to execute shell commands directly."
 }
 
 // Parameters returns the tool parameters schema
+
 func (t *CronTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"action": map[string]any{
-				"type":        "string",
-				"enum":        []string{"add", "list", "remove", "enable", "disable"},
+				"type": "string",
+
+				"enum": []string{"add", "list", "remove", "enable", "disable"},
+
 				"description": "Action to perform. Use 'add' when user wants to schedule a reminder or task.",
 			},
+
 			"message": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The reminder/task message to display when triggered. If 'command' is used, this describes what the command does.",
 			},
+
 			"command": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Optional: Shell command to execute directly (e.g., 'df -h'). If set, the agent will run this command and report output instead of just showing the message. 'deliver' will be forced to false for commands.",
 			},
+
 			"at_seconds": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "One-time reminder: seconds from now when to trigger (e.g., 600 for 10 minutes later). Use this for one-time reminders like 'remind me in 10 minutes'.",
 			},
+
 			"every_seconds": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Recurring interval in seconds (e.g., 3600 for every hour). Use this ONLY for recurring tasks like 'every 2 hours' or 'daily reminder'.",
 			},
+
 			"cron_expr": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Cron expression for complex recurring schedules (e.g., '0 9 * * *' for daily at 9am). Use this for complex recurring schedules.",
 			},
+
 			"job_id": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Job ID (for remove/enable/disable)",
 			},
+
 			"deliver": map[string]any{
-				"type":        "boolean",
+				"type": "boolean",
+
 				"description": "If true, send message directly to channel. If false, let agent process message (for complex tasks). Default: true",
 			},
 		},
+
 		"required": []string{"action"},
 	}
 }
 
 // SetContext sets the current session context for job creation
+
 func (t *CronTool) SetContext(channel, chatID string) {
 	t.mu.Lock()
+
 	defer t.mu.Unlock()
+
 	t.channel = channel
+
 	t.chatID = chatID
 }
 
 // Execute runs the tool with the given arguments
+
 func (t *CronTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	action, ok := args["action"].(string)
+
 	if !ok {
 		return ErrorResult("action is required")
 	}
 
 	switch action {
 	case "add":
+
 		return t.addJob(args)
+
 	case "list":
+
 		return t.listJobs()
+
 	case "remove":
+
 		return t.removeJob(args)
+
 	case "enable":
+
 		return t.enableJob(args, true)
+
 	case "disable":
+
 		return t.enableJob(args, false)
+
 	default:
+
 		return ErrorResult(fmt.Sprintf("unknown action: %s", action))
 	}
 }
 
 func (t *CronTool) addJob(args map[string]any) *ToolResult {
 	t.mu.RLock()
+
 	channel := t.channel
+
 	chatID := t.chatID
+
 	t.mu.RUnlock()
 
 	if channel == "" || chatID == "" {
@@ -144,6 +200,7 @@ func (t *CronTool) addJob(args map[string]any) *ToolResult {
 	}
 
 	message, ok := args["message"].(string)
+
 	if !ok || message == "" {
 		return ErrorResult("message is required for add")
 	}
@@ -151,26 +208,35 @@ func (t *CronTool) addJob(args map[string]any) *ToolResult {
 	var schedule cron.CronSchedule
 
 	// Check for at_seconds (one-time), every_seconds (recurring), or cron_expr
+
 	atSeconds, hasAt := args["at_seconds"].(float64)
+
 	everySeconds, hasEvery := args["every_seconds"].(float64)
+
 	cronExpr, hasCron := args["cron_expr"].(string)
 
 	// Priority: at_seconds > every_seconds > cron_expr
+
 	if hasAt {
 		atMS := time.Now().UnixMilli() + int64(atSeconds)*1000
+
 		schedule = cron.CronSchedule{
 			Kind: "at",
+
 			AtMS: &atMS,
 		}
 	} else if hasEvery {
 		everyMS := int64(everySeconds) * 1000
+
 		schedule = cron.CronSchedule{
-			Kind:    "every",
+			Kind: "every",
+
 			EveryMS: &everyMS,
 		}
 	} else if hasCron {
 		schedule = cron.CronSchedule{
 			Kind: "cron",
+
 			Expr: cronExpr,
 		}
 	} else {
@@ -178,29 +244,43 @@ func (t *CronTool) addJob(args map[string]any) *ToolResult {
 	}
 
 	// Read deliver parameter, default to true
+
 	deliver := true
+
 	if d, ok := args["deliver"].(bool); ok {
 		deliver = d
 	}
 
 	command, _ := args["command"].(string)
+
 	if command != "" {
 		// Commands must be processed by agent/exec tool, so deliver must be false (or handled specifically)
+
 		// Actually, let's keep deliver=false to let the system know it's not a simple chat message
+
 		// But for our new logic in ExecuteJob, we can handle it regardless of deliver flag if Payload.Command is set.
+
 		// However, logically, it's not "delivered" to chat directly as is.
+
 		deliver = false
 	}
 
 	// Truncate message for job name (max 30 chars)
+
 	messagePreview := utils.Truncate(message, 30)
 
 	job, err := t.cronService.AddJob(
+
 		messagePreview,
+
 		schedule,
+
 		message,
+
 		deliver,
+
 		channel,
+
 		chatID,
 	)
 	if err != nil {
@@ -209,7 +289,9 @@ func (t *CronTool) addJob(args map[string]any) *ToolResult {
 
 	if command != "" {
 		job.Payload.Command = command
+
 		// Need to save the updated payload
+
 		t.cronService.UpdateJob(job)
 	}
 
@@ -224,9 +306,12 @@ func (t *CronTool) listJobs() *ToolResult {
 	}
 
 	var sb strings.Builder
+
 	sb.WriteString("Scheduled jobs:\n")
+
 	for _, j := range jobs {
 		var scheduleInfo string
+
 		if j.Schedule.Kind == "every" && j.Schedule.EveryMS != nil {
 			scheduleInfo = fmt.Sprintf("every %ds", *j.Schedule.EveryMS/1000)
 		} else if j.Schedule.Kind == "cron" {
@@ -236,6 +321,7 @@ func (t *CronTool) listJobs() *ToolResult {
 		} else {
 			scheduleInfo = "unknown"
 		}
+
 		fmt.Fprintf(&sb, "- %s (id: %s, %s)\n", j.Name, j.ID, scheduleInfo)
 	}
 
@@ -244,6 +330,7 @@ func (t *CronTool) listJobs() *ToolResult {
 
 func (t *CronTool) removeJob(args map[string]any) *ToolResult {
 	jobID, ok := args["job_id"].(string)
+
 	if !ok || jobID == "" {
 		return ErrorResult("job_id is required for remove")
 	}
@@ -251,49 +338,62 @@ func (t *CronTool) removeJob(args map[string]any) *ToolResult {
 	if t.cronService.RemoveJob(jobID) {
 		return SilentResult(fmt.Sprintf("Cron job removed: %s", jobID))
 	}
+
 	return ErrorResult(fmt.Sprintf("Job %s not found", jobID))
 }
 
 func (t *CronTool) enableJob(args map[string]any, enable bool) *ToolResult {
 	jobID, ok := args["job_id"].(string)
+
 	if !ok || jobID == "" {
 		return ErrorResult("job_id is required for enable/disable")
 	}
 
 	job := t.cronService.EnableJob(jobID, enable)
+
 	if job == nil {
 		return ErrorResult(fmt.Sprintf("Job %s not found", jobID))
 	}
 
 	status := "enabled"
+
 	if !enable {
 		status = "disabled"
 	}
+
 	return SilentResult(fmt.Sprintf("Cron job '%s' %s", job.Name, status))
 }
 
 // ExecuteJob executes a cron job through the agent
+
 func (t *CronTool) ExecuteJob(ctx context.Context, job *cron.CronJob) string {
 	// Get channel/chatID from job payload
+
 	channel := job.Payload.Channel
+
 	chatID := job.Payload.To
 
 	// Default values if not set
+
 	if channel == "" {
 		channel = "cli"
 	}
+
 	if chatID == "" {
 		chatID = "direct"
 	}
 
 	// Execute command if present
+
 	if job.Payload.Command != "" {
 		args := map[string]any{
 			"command": job.Payload.Command,
 		}
 
 		result := t.execTool.Execute(ctx, args)
+
 		var output string
+
 		if result.IsError {
 			output = fmt.Sprintf("Error executing scheduled command: %s", result.ForLLM)
 		} else {
@@ -301,36 +401,54 @@ func (t *CronTool) ExecuteJob(ctx context.Context, job *cron.CronJob) string {
 		}
 
 		pubCtx, pubCancel := context.WithTimeout(context.Background(), 5*time.Second)
+
 		defer pubCancel()
+
 		t.msgBus.PublishOutbound(pubCtx, bus.OutboundMessage{
 			Channel: channel,
-			ChatID:  chatID,
+
+			ChatID: chatID,
+
 			Content: output,
 		})
+
 		return "ok"
 	}
 
 	// If deliver=true, send message directly without agent processing
+
 	if job.Payload.Deliver {
 		pubCtx, pubCancel := context.WithTimeout(context.Background(), 5*time.Second)
+
 		defer pubCancel()
+
 		t.msgBus.PublishOutbound(pubCtx, bus.OutboundMessage{
 			Channel: channel,
-			ChatID:  chatID,
+
+			ChatID: chatID,
+
 			Content: job.Payload.Message,
 		})
+
 		return "ok"
 	}
 
 	// For deliver=false, process through agent (for complex tasks)
+
 	sessionKey := fmt.Sprintf("cron-%s", job.ID)
 
 	// Call agent with job's message
+
 	response, err := t.executor.ProcessDirectWithChannel(
+
 		ctx,
+
 		job.Payload.Message,
+
 		sessionKey,
+
 		channel,
+
 		chatID,
 	)
 	if err != nil {
@@ -338,6 +456,8 @@ func (t *CronTool) ExecuteJob(ctx context.Context, job *cron.CronJob) string {
 	}
 
 	// Response is automatically sent via MessageBus by AgentLoop
+
 	_ = response // Will be sent by AgentLoop
+
 	return "ok"
 }

pkg/tools/dev_preview.go

@@ -10,11 +10,13 @@ import (
 )
 
 // DevPreviewTool allows the agent to control the Mini App dev reverse proxy.
+
 type DevPreviewTool struct {
 	manager miniapp.DevTargetManager
 }
 
 // NewDevPreviewTool creates a new DevPreviewTool.
+
 func NewDevPreviewTool(manager miniapp.DevTargetManager) *DevPreviewTool {
 	return &DevPreviewTool{manager: manager}
 }
@@ -28,113 +30,157 @@ func (t *DevPreviewTool) Description() string {
 func (t *DevPreviewTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"action": map[string]any{
-				"type":        "string",
-				"enum":        []string{"start", "stop", "unregister", "status"},
+				"type": "string",
+
+				"enum": []string{"start", "stop", "unregister", "status"},
+
 				"description": "Action to perform: start (register + activate target), stop (deactivate proxy), unregister (remove a registered target), status (list all targets).",
 			},
+
 			"target": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Target URL for the dev server (e.g. http://localhost:3000). Required for 'start' action. Must be a localhost URL.",
 			},
+
 			"name": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Display name for the target (e.g. 'frontend'). Optional for 'start' action; auto-generated from host:port if omitted.",
 			},
+
 			"id": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Target ID. Required for 'unregister' action.",
 			},
 		},
+
 		"required": []string{"action"},
 	}
 }
 
 func (t *DevPreviewTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	action, ok := args["action"].(string)
+
 	if !ok {
 		return ErrorResult("action is required")
 	}
 
 	switch action {
 	case "start":
+
 		target, _ := args["target"].(string)
+
 		if target == "" {
 			return ErrorResult("target is required for start action")
 		}
+
 		name, _ := args["name"].(string)
+
 		if name == "" {
 			name = inferName(target)
 		}
+
 		id, err := t.manager.RegisterDevTarget(name, target)
 		if err != nil {
 			return ErrorResult(fmt.Sprintf("failed to register dev target: %v", err))
 		}
+
 		if err := t.manager.ActivateDevTarget(id); err != nil {
 			return ErrorResult(fmt.Sprintf("failed to activate dev target: %v", err))
 		}
+
 		return SilentResult(
+
 			fmt.Sprintf(
+
 				"Dev preview started (id=%s, name=%s). Target: %s\nUsers can view it in the Mini App Dev tab.",
+
 				id,
+
 				name,
+
 				target,
 			),
 		)
 
 	case "stop":
+
 		if err := t.manager.DeactivateDevTarget(); err != nil {
 			return ErrorResult(fmt.Sprintf("failed to stop dev preview: %v", err))
 		}
+
 		return SilentResult("Dev preview stopped.")
 
 	case "unregister":
+
 		id, _ := args["id"].(string)
+
 		if id == "" {
 			return ErrorResult("id is required for unregister action")
 		}
+
 		if err := t.manager.UnregisterDevTarget(id); err != nil {
 			return ErrorResult(fmt.Sprintf("failed to unregister target: %v", err))
 		}
+
 		return SilentResult(fmt.Sprintf("Dev target %s unregistered.", id))
 
 	case "status":
+
 		targets := t.manager.ListDevTargets()
+
 		active := t.manager.GetDevTarget()
+
 		if len(targets) == 0 {
 			if active == "" {
 				return SilentResult("Dev preview is not active. No targets registered.")
 			}
+
 			return SilentResult(fmt.Sprintf("Dev preview is active. Target: %s\nNo registered targets.", active))
 		}
+
 		var sb strings.Builder
+
 		if active != "" {
 			sb.WriteString(fmt.Sprintf("Dev preview is active. Target: %s\n", active))
 		} else {
 			sb.WriteString("Dev preview is not active.\n")
 		}
+
 		sb.WriteString("Registered targets:\n")
+
 		for _, dt := range targets {
 			sb.WriteString(fmt.Sprintf("  [%s] %s → %s\n", dt.ID, dt.Name, dt.Target))
 		}
+
 		return SilentResult(sb.String())
 
 	default:
+
 		return ErrorResult(fmt.Sprintf("unknown action: %s", action))
 	}
 }
 
 // inferName generates a display name from a target URL (e.g. "localhost:3000").
+
 func inferName(target string) string {
 	u, err := url.Parse(target)
 	if err != nil {
 		return target
 	}
+
 	host := u.Hostname()
+
 	port := u.Port()
+
 	if port != "" {
 		return host + ":" + port
 	}
+
 	return host
 }

pkg/tools/dev_preview_test.go

@@ -10,12 +10,17 @@ import (
 )
 
 // mockDevTargetManager implements miniapp.DevTargetManager for testing.
+
 type mockDevTargetManager struct {
-	targets  map[string]*miniapp.DevTarget
-	nextID   int
+	targets map[string]*miniapp.DevTarget
+
+	nextID int
+
 	activeID string
-	active   string // active target URL
-	regErr   error
+
+	active string // active target URL
+
+	regErr error
 }
 
 func newMockManager() *mockDevTargetManager {
@@ -26,9 +31,13 @@ func (m *mockDevTargetManager) RegisterDevTarget(name, target string) (string, e
 	if m.regErr != nil {
 		return "", m.regErr
 	}
+
 	m.nextID++
+
 	id := fmt.Sprintf("%d", m.nextID)
+
 	m.targets[id] = &miniapp.DevTarget{ID: id, Name: name, Target: target}
+
 	return id, nil
 }
 
@@ -36,27 +45,37 @@ func (m *mockDevTargetManager) UnregisterDevTarget(id string) error {
 	if _, ok := m.targets[id]; !ok {
 		return fmt.Errorf("target %q not found", id)
 	}
+
 	delete(m.targets, id)
+
 	if m.activeID == id {
 		m.activeID = ""
+
 		m.active = ""
 	}
+
 	return nil
 }
 
 func (m *mockDevTargetManager) ActivateDevTarget(id string) error {
 	dt, ok := m.targets[id]
+
 	if !ok {
 		return fmt.Errorf("target %q not found", id)
 	}
+
 	m.activeID = id
+
 	m.active = dt.Target
+
 	return nil
 }
 
 func (m *mockDevTargetManager) DeactivateDevTarget() error {
 	m.activeID = ""
+
 	m.active = ""
+
 	return nil
 }
 
@@ -66,34 +85,43 @@ func (m *mockDevTargetManager) GetDevTarget() string {
 
 func (m *mockDevTargetManager) ListDevTargets() []miniapp.DevTarget {
 	out := make([]miniapp.DevTarget, 0, len(m.targets))
+
 	for _, dt := range m.targets {
 		out = append(out, *dt)
 	}
+
 	return out
 }
 
 func TestDevPreviewTool_Start(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:3000",
-		"name":   "frontend",
+
+		"name": "frontend",
 	})
 
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if len(mgr.targets) != 1 {
 		t.Errorf("expected 1 registered target, got %d", len(mgr.targets))
 	}
+
 	if mgr.active != "http://localhost:3000" {
 		t.Errorf("expected active target http://localhost:3000, got %q", mgr.active)
 	}
+
 	if !strings.Contains(result.ForLLM, "started") {
 		t.Errorf("expected result to contain 'started', got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "frontend") {
 		t.Errorf("expected result to contain 'frontend', got %q", result.ForLLM)
 	}
@@ -101,17 +129,21 @@ func TestDevPreviewTool_Start(t *testing.T) {
 
 func TestDevPreviewTool_StartAutoName(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:3000",
 	})
 
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	// Auto-generated name should be "localhost:3000"
+
 	for _, dt := range mgr.targets {
 		if dt.Name != "localhost:3000" {
 			t.Errorf("expected auto-name 'localhost:3000', got %q", dt.Name)
@@ -121,6 +153,7 @@ func TestDevPreviewTool_StartAutoName(t *testing.T) {
 
 func TestDevPreviewTool_StartMissingTarget(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -134,11 +167,14 @@ func TestDevPreviewTool_StartMissingTarget(t *testing.T) {
 
 func TestDevPreviewTool_StartError(t *testing.T) {
 	mgr := newMockManager()
+
 	mgr.regErr = fmt.Errorf("only localhost")
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://example.com:3000",
 	})
 
@@ -149,7 +185,9 @@ func TestDevPreviewTool_StartError(t *testing.T) {
 
 func TestDevPreviewTool_Stop(t *testing.T) {
 	mgr := newMockManager()
+
 	mgr.active = "http://localhost:3000"
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -159,6 +197,7 @@ func TestDevPreviewTool_Stop(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if mgr.active != "" {
 		t.Errorf("expected empty active target after stop, got %q", mgr.active)
 	}
@@ -166,19 +205,23 @@ func TestDevPreviewTool_Stop(t *testing.T) {
 
 func TestDevPreviewTool_Unregister(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	// Register a target first
+
 	id, _ := mgr.RegisterDevTarget("frontend", "http://localhost:3000")
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "unregister",
-		"id":     id,
+
+		"id": id,
 	})
 
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if len(mgr.targets) != 0 {
 		t.Errorf("expected 0 targets after unregister, got %d", len(mgr.targets))
 	}
@@ -186,6 +229,7 @@ func TestDevPreviewTool_Unregister(t *testing.T) {
 
 func TestDevPreviewTool_UnregisterMissingID(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -199,11 +243,13 @@ func TestDevPreviewTool_UnregisterMissingID(t *testing.T) {
 
 func TestDevPreviewTool_UnregisterNotFound(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "unregister",
-		"id":     "999",
+
+		"id": "999",
 	})
 
 	if !result.IsError {
@@ -213,10 +259,13 @@ func TestDevPreviewTool_UnregisterNotFound(t *testing.T) {
 
 func TestDevPreviewTool_Status(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	mgr.RegisterDevTarget("api", "http://localhost:8080")
+
 	mgr.RegisterDevTarget("frontend", "http://localhost:3000")
+
 	mgr.active = "http://localhost:8080"
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -226,15 +275,19 @@ func TestDevPreviewTool_Status(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "active") {
 		t.Errorf("expected 'active' in result, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "http://localhost:8080") {
 		t.Errorf("expected target URL in result, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "api") {
 		t.Errorf("expected 'api' in result, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "frontend") {
 		t.Errorf("expected 'frontend' in result, got %q", result.ForLLM)
 	}
@@ -242,6 +295,7 @@ func TestDevPreviewTool_Status(t *testing.T) {
 
 func TestDevPreviewTool_StatusInactive(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -251,6 +305,7 @@ func TestDevPreviewTool_StatusInactive(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "not active") {
 		t.Errorf("expected 'not active' in result, got %q", result.ForLLM)
 	}
@@ -258,6 +313,7 @@ func TestDevPreviewTool_StatusInactive(t *testing.T) {
 
 func TestDevPreviewTool_UnknownAction(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -271,6 +327,7 @@ func TestDevPreviewTool_UnknownAction(t *testing.T) {
 
 func TestDevPreviewTool_MissingAction(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{})
@@ -282,15 +339,19 @@ func TestDevPreviewTool_MissingAction(t *testing.T) {
 
 func TestDevPreviewTool_NameAndSchema(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	if tool.Name() != "dev_preview" {
 		t.Errorf("expected name dev_preview, got %q", tool.Name())
 	}
+
 	if tool.Description() == "" {
 		t.Error("expected non-empty description")
 	}
+
 	params := tool.Parameters()
+
 	if params == nil {
 		t.Fatal("expected non-nil parameters")
 	}
@@ -300,26 +361,35 @@ func TestDevPreviewTool_NameAndSchema(t *testing.T) {
 
 func TestDevPreviewTool_StartMultipleTargets(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	r1 := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:8080",
-		"name":   "api",
+
+		"name": "api",
 	})
+
 	r2 := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:3000",
-		"name":   "frontend",
+
+		"name": "frontend",
 	})
 
 	if r1.IsError || r2.IsError {
 		t.Fatalf("expected both starts to succeed, got err1=%v err2=%v", r1.IsError, r2.IsError)
 	}
+
 	if len(mgr.targets) != 2 {
 		t.Errorf("expected 2 registered targets, got %d", len(mgr.targets))
 	}
+
 	// The second start should make the frontend active
+
 	if mgr.active != "http://localhost:3000" {
 		t.Errorf("expected last started target to be active, got %q", mgr.active)
 	}
@@ -327,12 +397,15 @@ func TestDevPreviewTool_StartMultipleTargets(t *testing.T) {
 
 func TestDevPreviewTool_StopPreservesRegistrations(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:3000",
-		"name":   "frontend",
+
+		"name": "frontend",
 	})
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -342,11 +415,15 @@ func TestDevPreviewTool_StopPreservesRegistrations(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("stop failed: %s", result.ForLLM)
 	}
+
 	// Registration should still be there
+
 	if len(mgr.targets) != 1 {
 		t.Errorf("expected 1 registered target after stop, got %d", len(mgr.targets))
 	}
+
 	// But active should be cleared
+
 	if mgr.active != "" {
 		t.Errorf("expected inactive after stop, got %q", mgr.active)
 	}
@@ -354,9 +431,11 @@ func TestDevPreviewTool_StopPreservesRegistrations(t *testing.T) {
 
 func TestDevPreviewTool_StatusWithTargetsButInactive(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	mgr.RegisterDevTarget("api", "http://localhost:8080")
+
 	// active remains empty
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -366,9 +445,11 @@ func TestDevPreviewTool_StatusWithTargetsButInactive(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("status failed: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "not active") {
 		t.Errorf("expected 'not active' in status, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "api") {
 		t.Errorf("expected 'api' listed in status, got %q", result.ForLLM)
 	}
@@ -376,22 +457,29 @@ func TestDevPreviewTool_StatusWithTargetsButInactive(t *testing.T) {
 
 func TestDevPreviewTool_ResultIsSilent(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	cases := []struct {
 		name string
+
 		args map[string]any
 	}{
 		{"start", map[string]any{"action": "start", "target": "http://localhost:3000"}},
+
 		{"stop", map[string]any{"action": "stop"}},
+
 		{"status", map[string]any{"action": "status"}},
 	}
+
 	for _, tc := range cases {
 		t.Run(tc.name, func(t *testing.T) {
 			result := tool.Execute(context.Background(), tc.args)
+
 			if result.IsError {
 				t.Fatalf("expected success, got error: %s", result.ForLLM)
 			}
+
 			if result.Silent != true {
 				t.Errorf("expected SilentResult (IsSilent=true), got IsSilent=%v", result.Silent)
 			}
@@ -401,11 +489,13 @@ func TestDevPreviewTool_ResultIsSilent(t *testing.T) {
 
 func TestDevPreviewTool_ActionTypeNotString(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": 123,
 	})
+
 	if !result.IsError {
 		t.Error("expected error for non-string action")
 	}
@@ -414,17 +504,26 @@ func TestDevPreviewTool_ActionTypeNotString(t *testing.T) {
 func TestDevPreviewTool_InferName(t *testing.T) {
 	cases := []struct {
 		target string
-		want   string
+
+		want string
 	}{
 		{"http://localhost:3000", "localhost:3000"},
+
 		{"http://localhost:8080", "localhost:8080"},
+
 		{"http://127.0.0.1:9000", "127.0.0.1:9000"},
+
 		{"http://localhost", "localhost"},
+
 		{"http://[::1]:5000", "::1:5000"},
+
 		{"not-a-url", ""}, // url.Parse succeeds but Hostname() is empty
+
 	}
+
 	for _, tc := range cases {
 		got := inferName(tc.target)
+
 		if got != tc.want {
 			t.Errorf("inferName(%q) = %q, want %q", tc.target, got, tc.want)
 		}
@@ -433,18 +532,23 @@ func TestDevPreviewTool_InferName(t *testing.T) {
 
 func TestDevPreviewTool_StartEmptyName(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	// Explicitly pass empty name — should auto-infer
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:5000",
-		"name":   "",
+
+		"name": "",
 	})
 
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	for _, dt := range mgr.targets {
 		if dt.Name != "localhost:5000" {
 			t.Errorf("expected auto-name 'localhost:5000', got %q", dt.Name)
@@ -454,32 +558,41 @@ func TestDevPreviewTool_StartEmptyName(t *testing.T) {
 
 func TestDevPreviewTool_UnregisterActiveTarget(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	// Register and activate
+
 	tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "http://localhost:3000",
-		"name":   "frontend",
+
+		"name": "frontend",
 	})
 
 	// Find the registered ID
+
 	var id string
+
 	for k := range mgr.targets {
 		id = k
 	}
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "unregister",
-		"id":     id,
+
+		"id": id,
 	})
 
 	if result.IsError {
 		t.Fatalf("unregister failed: %s", result.ForLLM)
 	}
+
 	if len(mgr.targets) != 0 {
 		t.Errorf("expected 0 targets, got %d", len(mgr.targets))
 	}
+
 	if mgr.active != "" {
 		t.Errorf("expected no active target, got %q", mgr.active)
 	}
@@ -487,10 +600,12 @@ func TestDevPreviewTool_UnregisterActiveTarget(t *testing.T) {
 
 func TestDevPreviewTool_StartTargetEmptyString(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"action": "start",
+
 		"target": "",
 	})
 
@@ -501,8 +616,11 @@ func TestDevPreviewTool_StartTargetEmptyString(t *testing.T) {
 
 func TestDevPreviewTool_StatusActiveNoTargets(t *testing.T) {
 	// Edge case: active proxy but no registered targets (shouldn't normally happen)
+
 	mgr := newMockManager()
+
 	mgr.active = "http://localhost:9999" // active but targets map is empty
+
 	tool := NewDevPreviewTool(mgr)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -512,12 +630,15 @@ func TestDevPreviewTool_StatusActiveNoTargets(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "active") {
 		t.Errorf("expected 'active' in result, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "http://localhost:9999") {
 		t.Errorf("expected target URL in result, got %q", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "No registered targets") {
 		t.Errorf("expected 'No registered targets' in result, got %q", result.ForLLM)
 	}
@@ -525,10 +646,13 @@ func TestDevPreviewTool_StatusActiveNoTargets(t *testing.T) {
 
 func TestDevPreviewTool_StatusOutputFormat(t *testing.T) {
 	mgr := newMockManager()
+
 	tool := NewDevPreviewTool(mgr)
 
 	id1, _ := mgr.RegisterDevTarget("api", "http://localhost:8080")
+
 	mgr.RegisterDevTarget("frontend", "http://localhost:3000")
+
 	mgr.ActivateDevTarget(id1)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -538,15 +662,21 @@ func TestDevPreviewTool_StatusOutputFormat(t *testing.T) {
 	if result.IsError {
 		t.Fatalf("status failed: %s", result.ForLLM)
 	}
+
 	// Should contain IDs in bracket format
+
 	if !strings.Contains(result.ForLLM, "["+id1+"]") {
 		t.Errorf("expected [%s] in output, got %q", id1, result.ForLLM)
 	}
+
 	// Should contain the arrow
+
 	if !strings.Contains(result.ForLLM, "→") {
 		t.Errorf("expected arrow in output, got %q", result.ForLLM)
 	}
+
 	// Should contain "Registered targets:"
+
 	if !strings.Contains(result.ForLLM, "Registered targets:") {
 		t.Errorf("expected 'Registered targets:' header, got %q", result.ForLLM)
 	}

pkg/tools/edit.go

@@ -9,19 +9,24 @@ import (
 )
 
 // EditFileTool edits a file by replacing old_text with new_text.
+
 // The old_text must exist exactly in the file.
+
 type EditFileTool struct {
 	fs fileSystem
 }
 
 // NewEditFileTool creates a new EditFileTool with optional directory restriction.
+
 func NewEditFileTool(workspace string, restrict bool) *EditFileTool {
 	var fs fileSystem
+
 	if restrict {
 		fs = &sandboxFs{workspace: workspace}
 	} else {
 		fs = &hostFs{}
 	}
+
 	return &EditFileTool{fs: fs}
 }
 
@@ -36,36 +41,46 @@ func (t *EditFileTool) Description() string {
 func (t *EditFileTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"path": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The file path to edit",
 			},
+
 			"old_text": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The exact text to find and replace",
 			},
+
 			"new_text": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The text to replace with",
 			},
 		},
+
 		"required": []string{"path", "old_text", "new_text"},
 	}
 }
 
 func (t *EditFileTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	path, ok := args["path"].(string)
+
 	if !ok {
 		return ErrorResult("path is required")
 	}
 
 	oldText, ok := args["old_text"].(string)
+
 	if !ok {
 		return ErrorResult("old_text is required")
 	}
 
 	newText, ok := args["new_text"].(string)
+
 	if !ok {
 		return ErrorResult("new_text is required")
 	}
@@ -73,6 +88,7 @@ func (t *EditFileTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	if err := editFile(resolveFS(ctx, t.fs, path), path, oldText, newText); err != nil {
 		return ErrorResult(err.Error())
 	}
+
 	return SilentResult(fmt.Sprintf("File edited: %s", path))
 }
 
@@ -82,11 +98,13 @@ type AppendFileTool struct {
 
 func NewAppendFileTool(workspace string, restrict bool) *AppendFileTool {
 	var fs fileSystem
+
 	if restrict {
 		fs = &sandboxFs{workspace: workspace}
 	} else {
 		fs = &hostFs{}
 	}
+
 	return &AppendFileTool{fs: fs}
 }
 
@@ -101,27 +119,34 @@ func (t *AppendFileTool) Description() string {
 func (t *AppendFileTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"path": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The file path to append to",
 			},
+
 			"content": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The content to append",
 			},
 		},
+
 		"required": []string{"path", "content"},
 	}
 }
 
 func (t *AppendFileTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	path, ok := args["path"].(string)
+
 	if !ok {
 		return ErrorResult("path is required")
 	}
 
 	content, ok := args["content"].(string)
+
 	if !ok {
 		return ErrorResult("content is required")
 	}
@@ -129,11 +154,14 @@ func (t *AppendFileTool) Execute(ctx context.Context, args map[string]any) *Tool
 	if err := appendFile(resolveFS(ctx, t.fs, path), path, content); err != nil {
 		return ErrorResult(err.Error())
 	}
+
 	return SilentResult(fmt.Sprintf("Appended to %s", path))
 }
 
 // editFile reads the file via sysFs, performs the replacement, and writes back.
+
 // It uses a fileSystem interface, allowing the same logic for both restricted and unrestricted modes.
+
 func editFile(sysFs fileSystem, path, oldText, newText string) error {
 	content, err := sysFs.ReadFile(path)
 	if err != nil {
@@ -149,17 +177,21 @@ func editFile(sysFs fileSystem, path, oldText, newText string) error {
 }
 
 // appendFile reads the existing content (if any) via sysFs, appends new content, and writes back.
+
 func appendFile(sysFs fileSystem, path, appendContent string) error {
 	content, err := sysFs.ReadFile(path)
+
 	if err != nil && !errors.Is(err, fs.ErrNotExist) {
 		return err
 	}
 
 	newContent := append(content, []byte(appendContent)...)
+
 	return sysFs.WriteFile(path, newContent)
 }
 
 // replaceEditContent handles the core logic of finding and replacing a single occurrence of oldText.
+
 func replaceEditContent(content []byte, oldText, newText string) ([]byte, error) {
 	contentStr := string(content)
 
@@ -168,10 +200,12 @@ func replaceEditContent(content []byte, oldText, newText string) ([]byte, error)
 	}
 
 	count := strings.Count(contentStr, oldText)
+
 	if count > 1 {
 		return nil, fmt.Errorf("old_text appears %d times. Please provide more context to make it unique", count)
 	}
 
 	newContent := strings.Replace(contentStr, oldText, newText, 1)
+
 	return []byte(newContent), nil
 }

pkg/tools/edit_test.go

@@ -11,261 +11,349 @@ import (
 )
 
 // TestEditTool_EditFile_Success verifies successful file editing
+
 func TestEditTool_EditFile_Success(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("Hello World\nThis is a test"), 0o644)
 
 	tool := NewEditFileTool(tmpDir, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "World",
+
 		"new_text": "Universe",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// Should return SilentResult
+
 	if !result.Silent {
 		t.Errorf("Expected Silent=true for EditFile, got false")
 	}
 
 	// ForUser should be empty (silent result)
+
 	if result.ForUser != "" {
 		t.Errorf("Expected ForUser to be empty for SilentResult, got: %s", result.ForUser)
 	}
 
 	// Verify file was actually edited
+
 	content, err := os.ReadFile(testFile)
 	if err != nil {
 		t.Fatalf("Failed to read edited file: %v", err)
 	}
+
 	contentStr := string(content)
+
 	if !strings.Contains(contentStr, "Hello Universe") {
 		t.Errorf("Expected file to contain 'Hello Universe', got: %s", contentStr)
 	}
+
 	if strings.Contains(contentStr, "Hello World") {
 		t.Errorf("Expected 'Hello World' to be replaced, got: %s", contentStr)
 	}
 }
 
 // TestEditTool_EditFile_NotFound verifies error handling for non-existent file
+
 func TestEditTool_EditFile_NotFound(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "nonexistent.txt")
 
 	tool := NewEditFileTool(tmpDir, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "old",
+
 		"new_text": "new",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error for non-existent file")
 	}
 
 	// Should mention file not found
+
 	if !strings.Contains(result.ForLLM, "not found") && !strings.Contains(result.ForUser, "not found") {
 		t.Errorf("Expected 'file not found' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestEditTool_EditFile_OldTextNotFound verifies error when old_text doesn't exist
+
 func TestEditTool_EditFile_OldTextNotFound(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("Hello World"), 0o644)
 
 	tool := NewEditFileTool(tmpDir, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "Goodbye",
+
 		"new_text": "Hello",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when old_text not found")
 	}
 
 	// Should mention old_text not found
+
 	if !strings.Contains(result.ForLLM, "not found") && !strings.Contains(result.ForUser, "not found") {
 		t.Errorf("Expected 'not found' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestEditTool_EditFile_MultipleMatches verifies error when old_text appears multiple times
+
 func TestEditTool_EditFile_MultipleMatches(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("test test test"), 0o644)
 
 	tool := NewEditFileTool(tmpDir, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "test",
+
 		"new_text": "done",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when old_text appears multiple times")
 	}
 
 	// Should mention multiple occurrences
+
 	if !strings.Contains(result.ForLLM, "times") && !strings.Contains(result.ForUser, "times") {
 		t.Errorf("Expected 'multiple times' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestEditTool_EditFile_OutsideAllowedDir verifies error when path is outside allowed directory
+
 func TestEditTool_EditFile_OutsideAllowedDir(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	otherDir := t.TempDir()
+
 	testFile := filepath.Join(otherDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("content"), 0o644)
 
 	tool := NewEditFileTool(tmpDir, true) // Restrict to tmpDir
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "content",
+
 		"new_text": "new",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	assert.True(t, result.IsError, "Expected error when path is outside allowed directory")
 
 	// Should mention outside allowed directory
+
 	// Note: ErrorResult only sets ForLLM by default, so ForUser might be empty.
+
 	// We check ForLLM as it's the primary error channel.
+
 	assert.True(
+
 		t,
+
 		strings.Contains(result.ForLLM, "outside") || strings.Contains(result.ForLLM, "access denied") ||
+
 			strings.Contains(result.ForLLM, "escapes"),
+
 		"Expected 'outside allowed' or 'access denied' message, got ForLLM: %s",
+
 		result.ForLLM,
 	)
 }
 
 // TestEditTool_EditFile_MissingPath verifies error handling for missing path
+
 func TestEditTool_EditFile_MissingPath(t *testing.T) {
 	tool := NewEditFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"old_text": "old",
+
 		"new_text": "new",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when path is missing")
 	}
 }
 
 // TestEditTool_EditFile_MissingOldText verifies error handling for missing old_text
+
 func TestEditTool_EditFile_MissingOldText(t *testing.T) {
 	tool := NewEditFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     "/tmp/test.txt",
+		"path": "/tmp/test.txt",
+
 		"new_text": "new",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when old_text is missing")
 	}
 }
 
 // TestEditTool_EditFile_MissingNewText verifies error handling for missing new_text
+
 func TestEditTool_EditFile_MissingNewText(t *testing.T) {
 	tool := NewEditFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     "/tmp/test.txt",
+		"path": "/tmp/test.txt",
+
 		"old_text": "old",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when new_text is missing")
 	}
 }
 
 // TestEditTool_AppendFile_Success verifies successful file appending
+
 func TestEditTool_AppendFile_Success(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("Initial content"), 0o644)
 
 	tool := NewAppendFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":    testFile,
+		"path": testFile,
+
 		"content": "\nAppended content",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// Should return SilentResult
+
 	if !result.Silent {
 		t.Errorf("Expected Silent=true for AppendFile, got false")
 	}
 
 	// ForUser should be empty (silent result)
+
 	if result.ForUser != "" {
 		t.Errorf("Expected ForUser to be empty for SilentResult, got: %s", result.ForUser)
 	}
 
 	// Verify content was actually appended
+
 	content, err := os.ReadFile(testFile)
 	if err != nil {
 		t.Fatalf("Failed to read file: %v", err)
 	}
+
 	contentStr := string(content)
+
 	if !strings.Contains(contentStr, "Initial content") {
 		t.Errorf("Expected original content to remain, got: %s", contentStr)
 	}
+
 	if !strings.Contains(contentStr, "Appended content") {
 		t.Errorf("Expected appended content, got: %s", contentStr)
 	}
 }
 
 // TestEditTool_AppendFile_MissingPath verifies error handling for missing path
+
 func TestEditTool_AppendFile_MissingPath(t *testing.T) {
 	tool := NewAppendFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "test",
 	}
@@ -273,15 +361,19 @@ func TestEditTool_AppendFile_MissingPath(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when path is missing")
 	}
 }
 
 // TestEditTool_AppendFile_MissingContent verifies error handling for missing content
+
 func TestEditTool_AppendFile_MissingContent(t *testing.T) {
 	tool := NewAppendFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": "/tmp/test.txt",
 	}
@@ -289,43 +381,67 @@ func TestEditTool_AppendFile_MissingContent(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when content is missing")
 	}
 }
 
 // TestReplaceEditContent verifies the helper function replaceEditContent
+
 func TestReplaceEditContent(t *testing.T) {
 	tests := []struct {
-		name        string
-		content     []byte
-		oldText     string
-		newText     string
-		expected    []byte
+		name string
+
+		content []byte
+
+		oldText string
+
+		newText string
+
+		expected []byte
+
 		expectError bool
 	}{
 		{
-			name:        "successful replacement",
-			content:     []byte("hello world"),
-			oldText:     "world",
-			newText:     "universe",
-			expected:    []byte("hello universe"),
+			name: "successful replacement",
+
+			content: []byte("hello world"),
+
+			oldText: "world",
+
+			newText: "universe",
+
+			expected: []byte("hello universe"),
+
 			expectError: false,
 		},
+
 		{
-			name:        "old text not found",
-			content:     []byte("hello world"),
-			oldText:     "golang",
-			newText:     "rust",
-			expected:    nil,
+			name: "old text not found",
+
+			content: []byte("hello world"),
+
+			oldText: "golang",
+
+			newText: "rust",
+
+			expected: nil,
+
 			expectError: true,
 		},
+
 		{
-			name:        "multiple matches found",
-			content:     []byte("test text test"),
-			oldText:     "test",
-			newText:     "done",
-			expected:    nil,
+			name: "multiple matches found",
+
+			content: []byte("test text test"),
+
+			oldText: "test",
+
+			newText: "done",
+
+			expected: nil,
+
 			expectError: true,
 		},
 	}
@@ -333,10 +449,12 @@ func TestReplaceEditContent(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			result, err := replaceEditContent(tt.content, tt.oldText, tt.newText)
+
 			if tt.expectError {
 				assert.Error(t, err)
 			} else {
 				assert.NoError(t, err)
+
 				assert.Equal(t, tt.expected, result)
 			}
 		})
@@ -344,94 +462,142 @@ func TestReplaceEditContent(t *testing.T) {
 }
 
 // TestAppendFileTool_AppendToNonExistent_Restricted verifies that AppendFileTool in restricted mode
+
 // can append to a file that does not yet exist — it should silently create the file.
+
 // This exercises the errors.Is(err, fs.ErrNotExist) path in appendFile + sandboxFs.
+
 func TestAppendFileTool_AppendToNonExistent_Restricted(t *testing.T) {
 	workspace := t.TempDir()
+
 	tool := NewAppendFileTool(workspace, true)
+
 	ctx := context.Background()
 
 	args := map[string]any{
-		"path":    "brand_new_file.txt",
+		"path": "brand_new_file.txt",
+
 		"content": "first content",
 	}
 
 	result := tool.Execute(ctx, args)
+
 	assert.False(
+
 		t,
+
 		result.IsError,
+
 		"Expected success when appending to non-existent file in restricted mode, got: %s",
+
 		result.ForLLM,
 	)
 
 	// Verify the file was created with correct content
+
 	data, err := os.ReadFile(filepath.Join(workspace, "brand_new_file.txt"))
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, "first content", string(data))
 }
 
 // TestAppendFileTool_Restricted_Success verifies that AppendFileTool in restricted mode
+
 // correctly appends to an existing file within the sandbox.
+
 func TestAppendFileTool_Restricted_Success(t *testing.T) {
 	workspace := t.TempDir()
+
 	testFile := "existing.txt"
+
 	err := os.WriteFile(filepath.Join(workspace, testFile), []byte("initial"), 0o644)
+
 	assert.NoError(t, err)
 
 	tool := NewAppendFileTool(workspace, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":    testFile,
+		"path": testFile,
+
 		"content": " appended",
 	}
 
 	result := tool.Execute(ctx, args)
+
 	assert.False(t, result.IsError, "Expected success, got: %s", result.ForLLM)
+
 	assert.True(t, result.Silent)
 
 	data, err := os.ReadFile(filepath.Join(workspace, testFile))
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, "initial appended", string(data))
 }
 
 // TestEditFileTool_Restricted_InPlaceEdit verifies that EditFileTool in restricted mode
+
 // correctly edits a file using the sandboxFs path.
+
 func TestEditFileTool_Restricted_InPlaceEdit(t *testing.T) {
 	workspace := t.TempDir()
+
 	testFile := "edit_target.txt"
+
 	err := os.WriteFile(filepath.Join(workspace, testFile), []byte("Hello World"), 0o644)
+
 	assert.NoError(t, err)
 
 	tool := NewEditFileTool(workspace, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     testFile,
+		"path": testFile,
+
 		"old_text": "World",
+
 		"new_text": "Go",
 	}
 
 	result := tool.Execute(ctx, args)
+
 	assert.False(t, result.IsError, "Expected success, got: %s", result.ForLLM)
+
 	assert.True(t, result.Silent)
 
 	data, err := os.ReadFile(filepath.Join(workspace, testFile))
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, "Hello Go", string(data))
 }
 
 // TestEditFileTool_Restricted_FileNotFound verifies that editFile returns a proper
+
 // error message when the target file does not exist.
+
 func TestEditFileTool_Restricted_FileNotFound(t *testing.T) {
 	workspace := t.TempDir()
+
 	tool := NewEditFileTool(workspace, true)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":     "no_such_file.txt",
+		"path": "no_such_file.txt",
+
 		"old_text": "old",
+
 		"new_text": "new",
 	}
 
 	result := tool.Execute(ctx, args)
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "not found")
 }

pkg/tools/filesystem.go

@@ -13,7 +13,9 @@ import (
 )
 
 // validatePath ensures the given path is within the workspace if restrict is true.
+
 // Used by shell.go for working directory validation.
+
 func validatePath(path, workspace string, restrict bool) (string, error) {
 	if workspace == "" {
 		return path, fmt.Errorf("workspace is not defined")
@@ -25,6 +27,7 @@ func validatePath(path, workspace string, restrict bool) (string, error) {
 	}
 
 	var absPath string
+
 	if filepath.IsAbs(path) {
 		absPath = filepath.Clean(path)
 	} else {
@@ -40,7 +43,9 @@ func validatePath(path, workspace string, restrict bool) (string, error) {
 		}
 
 		var resolved string
+
 		workspaceReal := absWorkspace
+
 		if resolved, err = filepath.EvalSymlinks(absWorkspace); err == nil {
 			workspaceReal = resolved
 		}
@@ -51,6 +56,7 @@ func validatePath(path, workspace string, restrict bool) (string, error) {
 			}
 		} else if os.IsNotExist(err) {
 			var parentResolved string
+
 			if parentResolved, err = resolveExistingAncestor(filepath.Dir(absPath)); err == nil {
 				if !isWithinWorkspace(parentResolved, workspaceReal) {
 					return "", fmt.Errorf("access denied: symlink resolves outside workspace")
@@ -73,6 +79,7 @@ func resolveExistingAncestor(path string) (string, error) {
 		} else if !os.IsNotExist(err) {
 			return "", err
 		}
+
 		if filepath.Dir(current) == current {
 			return "", os.ErrNotExist
 		}
@@ -81,6 +88,7 @@ func resolveExistingAncestor(path string) (string, error) {
 
 func isWithinWorkspace(candidate, workspace string) bool {
 	rel, err := filepath.Rel(filepath.Clean(workspace), filepath.Clean(candidate))
+
 	return err == nil && filepath.IsLocal(rel)
 }
 
@@ -90,11 +98,13 @@ type ReadFileTool struct {
 
 func NewReadFileTool(workspace string, restrict bool) *ReadFileTool {
 	var fs fileSystem
+
 	if restrict {
 		fs = &sandboxFs{workspace: workspace}
 	} else {
 		fs = &hostFs{}
 	}
+
 	return &ReadFileTool{fs: fs}
 }
 
@@ -109,18 +119,22 @@ func (t *ReadFileTool) Description() string {
 func (t *ReadFileTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"path": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Path to the file to read",
 			},
 		},
+
 		"required": []string{"path"},
 	}
 }
 
 func (t *ReadFileTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	path, ok := args["path"].(string)
+
 	if !ok {
 		return ErrorResult("path is required")
 	}
@@ -129,6 +143,7 @@ func (t *ReadFileTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	if err != nil {
 		return ErrorResult(err.Error())
 	}
+
 	return NewToolResult(string(content))
 }
 
@@ -138,11 +153,13 @@ type WriteFileTool struct {
 
 func NewWriteFileTool(workspace string, restrict bool) *WriteFileTool {
 	var fs fileSystem
+
 	if restrict {
 		fs = &sandboxFs{workspace: workspace}
 	} else {
 		fs = &hostFs{}
 	}
+
 	return &WriteFileTool{fs: fs}
 }
 
@@ -157,27 +174,34 @@ func (t *WriteFileTool) Description() string {
 func (t *WriteFileTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"path": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Path to the file to write",
 			},
+
 			"content": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Content to write to the file",
 			},
 		},
+
 		"required": []string{"path", "content"},
 	}
 }
 
 func (t *WriteFileTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	path, ok := args["path"].(string)
+
 	if !ok {
 		return ErrorResult("path is required")
 	}
 
 	content, ok := args["content"].(string)
+
 	if !ok {
 		return ErrorResult("content is required")
 	}
@@ -195,11 +219,13 @@ type ListDirTool struct {
 
 func NewListDirTool(workspace string, restrict bool) *ListDirTool {
 	var fs fileSystem
+
 	if restrict {
 		fs = &sandboxFs{workspace: workspace}
 	} else {
 		fs = &hostFs{}
 	}
+
 	return &ListDirTool{fs: fs}
 }
 
@@ -214,18 +240,22 @@ func (t *ListDirTool) Description() string {
 func (t *ListDirTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"path": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Path to list",
 			},
 		},
+
 		"required": []string{"path"},
 	}
 }
 
 func (t *ListDirTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	path, ok := args["path"].(string)
+
 	if !ok {
 		path = "."
 	}
@@ -234,32 +264,42 @@ func (t *ListDirTool) Execute(ctx context.Context, args map[string]any) *ToolRes
 	if err != nil {
 		return ErrorResult(err.Error())
 	}
+
 	return formatDirEntries(entries)
 }
 
 func formatDirEntries(entries []os.DirEntry) *ToolResult {
 	var result strings.Builder
+
 	for _, entry := range entries {
 		if entry.IsDir() {
 			result.WriteString("DIR:  ")
 		} else {
 			result.WriteString("FILE: ")
 		}
+
 		result.WriteString(entry.Name())
+
 		result.WriteByte('\n')
 	}
+
 	return NewToolResult(result.String())
 }
 
 // fileSystem abstracts reading, writing, and listing files, allowing both
+
 // unrestricted (host filesystem) and sandbox (os.Root) implementations to share the same polymorphic interface.
+
 type fileSystem interface {
 	ReadFile(path string) ([]byte, error)
+
 	WriteFile(path string, data []byte) error
+
 	ReadDir(path string) ([]os.DirEntry, error)
 }
 
 // hostFs is an unrestricted fileReadWriter that operates directly on the host filesystem.
+
 type hostFs struct{}
 
 func (h *hostFs) ReadFile(path string) ([]byte, error) {
@@ -268,11 +308,14 @@ func (h *hostFs) ReadFile(path string) ([]byte, error) {
 		if os.IsNotExist(err) {
 			return nil, fmt.Errorf("failed to read file: file not found: %w", err)
 		}
+
 		if os.IsPermission(err) {
 			return nil, fmt.Errorf("failed to read file: access denied: %w", err)
 		}
+
 		return nil, fmt.Errorf("failed to read file: %w", err)
 	}
+
 	return content, nil
 }
 
@@ -281,16 +324,20 @@ func (h *hostFs) ReadDir(path string) ([]os.DirEntry, error) {
 	if err != nil {
 		return nil, fmt.Errorf("failed to read directory: %w", err)
 	}
+
 	return entries, nil
 }
 
 func (h *hostFs) WriteFile(path string, data []byte) error {
 	// Use unified atomic write utility with explicit sync for flash storage reliability.
+
 	// Using 0o600 (owner read/write only) for secure default permissions.
+
 	return fileutil.WriteFileAtomic(path, data, 0o600)
 }
 
 // sandboxFs is a sandboxed fileSystem that operates within a strictly defined workspace using os.Root.
+
 type sandboxFs struct {
 	workspace string
 }
@@ -304,6 +351,7 @@ func (r *sandboxFs) execute(path string, fn func(root *os.Root, relPath string)
 	if err != nil {
 		return fmt.Errorf("failed to open workspace: %w", err)
 	}
+
 	defer root.Close()
 
 	relPath, err := getSafeRelPath(r.workspace, path)
@@ -316,28 +364,37 @@ func (r *sandboxFs) execute(path string, fn func(root *os.Root, relPath string)
 
 func (r *sandboxFs) ReadFile(path string) ([]byte, error) {
 	var content []byte
+
 	err := r.execute(path, func(root *os.Root, relPath string) error {
 		fileContent, err := root.ReadFile(relPath)
 		if err != nil {
 			if os.IsNotExist(err) {
 				return fmt.Errorf("failed to read file: file not found: %w", err)
 			}
+
 			// os.Root returns "escapes from parent" for paths outside the root
+
 			if os.IsPermission(err) || strings.Contains(err.Error(), "escapes from parent") ||
+
 				strings.Contains(err.Error(), "permission denied") {
 				return fmt.Errorf("failed to read file: access denied: %w", err)
 			}
+
 			return fmt.Errorf("failed to read file: %w", err)
 		}
+
 		content = fileContent
+
 		return nil
 	})
+
 	return content, err
 }
 
 func (r *sandboxFs) WriteFile(path string, data []byte) error {
 	return r.execute(path, func(root *os.Root, relPath string) error {
 		dir := filepath.Dir(relPath)
+
 		if dir != "." && dir != "/" {
 			if err := root.MkdirAll(dir, 0o755); err != nil {
 				return fmt.Errorf("failed to create parent directories: %w", err)
@@ -345,42 +402,55 @@ func (r *sandboxFs) WriteFile(path string, data []byte) error {
 		}
 
 		// Use atomic write pattern with explicit sync for flash storage reliability.
+
 		// Using 0o600 (owner read/write only) for secure default permissions.
+
 		tmpRelPath := fmt.Sprintf(".tmp-%d-%d", os.Getpid(), time.Now().UnixNano())
 
 		tmpFile, err := root.OpenFile(tmpRelPath, os.O_WRONLY|os.O_CREATE|os.O_EXCL, 0o600)
 		if err != nil {
 			root.Remove(tmpRelPath)
+
 			return fmt.Errorf("failed to open temp file: %w", err)
 		}
 
 		if _, err := tmpFile.Write(data); err != nil {
 			tmpFile.Close()
+
 			root.Remove(tmpRelPath)
+
 			return fmt.Errorf("failed to write temp file: %w", err)
 		}
 
 		// CRITICAL: Force sync to storage medium before rename.
+
 		// This ensures data is physically written to disk, not just cached.
+
 		if err := tmpFile.Sync(); err != nil {
 			tmpFile.Close()
+
 			root.Remove(tmpRelPath)
+
 			return fmt.Errorf("failed to sync temp file: %w", err)
 		}
 
 		if err := tmpFile.Close(); err != nil {
 			root.Remove(tmpRelPath)
+
 			return fmt.Errorf("failed to close temp file: %w", err)
 		}
 
 		if err := root.Rename(tmpRelPath, relPath); err != nil {
 			root.Remove(tmpRelPath)
+
 			return fmt.Errorf("failed to rename temp file over target: %w", err)
 		}
 
 		// Sync directory to ensure rename is durable
+
 		if dirFile, err := root.Open("."); err == nil {
 			_ = dirFile.Sync()
+
 			dirFile.Close()
 		}
 
@@ -390,26 +460,33 @@ func (r *sandboxFs) WriteFile(path string, data []byte) error {
 
 func (r *sandboxFs) ReadDir(path string) ([]os.DirEntry, error) {
 	var entries []os.DirEntry
+
 	err := r.execute(path, func(root *os.Root, relPath string) error {
 		dirEntries, err := fs.ReadDir(root.FS(), relPath)
 		if err != nil {
 			return err
 		}
+
 		entries = dirEntries
+
 		return nil
 	})
+
 	return entries, err
 }
 
 // Helper to get a safe relative path for os.Root usage
+
 func getSafeRelPath(workspace, path string) (string, error) {
 	if workspace == "" {
 		return "", fmt.Errorf("workspace is not defined")
 	}
 
 	rel := filepath.Clean(path)
+
 	if filepath.IsAbs(rel) {
 		var err error
+
 		rel, err = filepath.Rel(workspace, rel)
 		if err != nil {
 			return "", fmt.Errorf("failed to calculate relative path: %w", err)

pkg/tools/filesystem_test.go

@@ -12,13 +12,18 @@ import (
 )
 
 // TestFilesystemTool_ReadFile_Success verifies successful file reading
+
 func TestFilesystemTool_ReadFile_Success(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "test.txt")
+
 	os.WriteFile(testFile, []byte("test content"), 0o644)
 
 	tool := NewReadFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": testFile,
 	}
@@ -26,26 +31,33 @@ func TestFilesystemTool_ReadFile_Success(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForLLM should contain file content
+
 	if !strings.Contains(result.ForLLM, "test content") {
 		t.Errorf("Expected ForLLM to contain 'test content', got: %s", result.ForLLM)
 	}
 
 	// ReadFile returns NewToolResult which only sets ForLLM, not ForUser
+
 	// This is the expected behavior - file content goes to LLM, not directly to user
+
 	if result.ForUser != "" {
 		t.Errorf("Expected ForUser to be empty for NewToolResult, got: %s", result.ForUser)
 	}
 }
 
 // TestFilesystemTool_ReadFile_NotFound verifies error handling for missing file
+
 func TestFilesystemTool_ReadFile_NotFound(t *testing.T) {
 	tool := NewReadFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": "/nonexistent_file_12345.txt",
 	}
@@ -53,107 +65,135 @@ func TestFilesystemTool_ReadFile_NotFound(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Failure should be marked as error
+
 	if !result.IsError {
 		t.Errorf("Expected error for missing file, got IsError=false")
 	}
 
 	// Should contain error message
+
 	if !strings.Contains(result.ForLLM, "failed to read") && !strings.Contains(result.ForUser, "failed to read") {
 		t.Errorf("Expected error message, got ForLLM: %s, ForUser: %s", result.ForLLM, result.ForUser)
 	}
 }
 
 // TestFilesystemTool_ReadFile_MissingPath verifies error handling for missing path
+
 func TestFilesystemTool_ReadFile_MissingPath(t *testing.T) {
 	tool := &ReadFileTool{}
+
 	ctx := context.Background()
+
 	args := map[string]any{}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when path is missing")
 	}
 
 	// Should mention required parameter
+
 	if !strings.Contains(result.ForLLM, "path is required") && !strings.Contains(result.ForUser, "path is required") {
 		t.Errorf("Expected 'path is required' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestFilesystemTool_WriteFile_Success verifies successful file writing
+
 func TestFilesystemTool_WriteFile_Success(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "newfile.txt")
 
 	tool := NewWriteFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":    testFile,
+		"path": testFile,
+
 		"content": "hello world",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// WriteFile returns SilentResult
+
 	if !result.Silent {
 		t.Errorf("Expected Silent=true for WriteFile, got false")
 	}
 
 	// ForUser should be empty (silent result)
+
 	if result.ForUser != "" {
 		t.Errorf("Expected ForUser to be empty for SilentResult, got: %s", result.ForUser)
 	}
 
 	// Verify file was actually written
+
 	content, err := os.ReadFile(testFile)
 	if err != nil {
 		t.Fatalf("Failed to read written file: %v", err)
 	}
+
 	if string(content) != "hello world" {
 		t.Errorf("Expected file content 'hello world', got: %s", string(content))
 	}
 }
 
 // TestFilesystemTool_WriteFile_CreateDir verifies directory creation
+
 func TestFilesystemTool_WriteFile_CreateDir(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "subdir", "newfile.txt")
 
 	tool := NewWriteFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
-		"path":    testFile,
+		"path": testFile,
+
 		"content": "test",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success with directory creation, got IsError=true: %s", result.ForLLM)
 	}
 
 	// Verify directory was created and file written
+
 	content, err := os.ReadFile(testFile)
 	if err != nil {
 		t.Fatalf("Failed to read written file: %v", err)
 	}
+
 	if string(content) != "test" {
 		t.Errorf("Expected file content 'test', got: %s", string(content))
 	}
 }
 
 // TestFilesystemTool_WriteFile_MissingPath verifies error handling for missing path
+
 func TestFilesystemTool_WriteFile_MissingPath(t *testing.T) {
 	tool := NewWriteFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "test",
 	}
@@ -161,15 +201,19 @@ func TestFilesystemTool_WriteFile_MissingPath(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when path is missing")
 	}
 }
 
 // TestFilesystemTool_WriteFile_MissingContent verifies error handling for missing content
+
 func TestFilesystemTool_WriteFile_MissingContent(t *testing.T) {
 	tool := NewWriteFileTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": "/tmp/test.txt",
 	}
@@ -177,26 +221,35 @@ func TestFilesystemTool_WriteFile_MissingContent(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when content is missing")
 	}
 
 	// Should mention required parameter
+
 	if !strings.Contains(result.ForLLM, "content is required") &&
+
 		!strings.Contains(result.ForUser, "content is required") {
 		t.Errorf("Expected 'content is required' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestFilesystemTool_ListDir_Success verifies successful directory listing
+
 func TestFilesystemTool_ListDir_Success(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	os.WriteFile(filepath.Join(tmpDir, "file1.txt"), []byte("content"), 0o644)
+
 	os.WriteFile(filepath.Join(tmpDir, "file2.txt"), []byte("content"), 0o644)
+
 	os.Mkdir(filepath.Join(tmpDir, "subdir"), 0o755)
 
 	tool := NewListDirTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": tmpDir,
 	}
@@ -204,23 +257,29 @@ func TestFilesystemTool_ListDir_Success(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// Should list files and directories
+
 	if !strings.Contains(result.ForLLM, "file1.txt") || !strings.Contains(result.ForLLM, "file2.txt") {
 		t.Errorf("Expected files in listing, got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "subdir") {
 		t.Errorf("Expected subdir in listing, got: %s", result.ForLLM)
 	}
 }
 
 // TestFilesystemTool_ListDir_NotFound verifies error handling for non-existent directory
+
 func TestFilesystemTool_ListDir_NotFound(t *testing.T) {
 	tool := NewListDirTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"path": "/nonexistent_directory_12345",
 	}
@@ -228,49 +287,61 @@ func TestFilesystemTool_ListDir_NotFound(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Failure should be marked as error
+
 	if !result.IsError {
 		t.Errorf("Expected error for non-existent directory, got IsError=false")
 	}
 
 	// Should contain error message
+
 	if !strings.Contains(result.ForLLM, "failed to read") && !strings.Contains(result.ForUser, "failed to read") {
 		t.Errorf("Expected error message, got ForLLM: %s, ForUser: %s", result.ForLLM, result.ForUser)
 	}
 }
 
 // TestFilesystemTool_ListDir_DefaultPath verifies default to current directory
+
 func TestFilesystemTool_ListDir_DefaultPath(t *testing.T) {
 	tool := NewListDirTool("", false)
+
 	ctx := context.Background()
+
 	args := map[string]any{}
 
 	result := tool.Execute(ctx, args)
 
 	// Should use "." as default path
+
 	if result.IsError {
 		t.Errorf("Expected success with default path '.', got IsError=true: %s", result.ForLLM)
 	}
 }
 
 // Block paths that look inside workspace but point outside via symlink.
+
 func TestFilesystemTool_ReadFile_RejectsSymlinkEscape(t *testing.T) {
 	root := t.TempDir()
+
 	workspace := filepath.Join(root, "workspace")
+
 	if err := os.MkdirAll(workspace, 0o755); err != nil {
 		t.Fatalf("failed to create workspace: %v", err)
 	}
 
 	secret := filepath.Join(root, "secret.txt")
+
 	if err := os.WriteFile(secret, []byte("top secret"), 0o644); err != nil {
 		t.Fatalf("failed to write secret file: %v", err)
 	}
 
 	link := filepath.Join(workspace, "leak.txt")
+
 	if err := os.Symlink(secret, link); err != nil {
 		t.Skipf("symlink not supported in this environment: %v", err)
 	}
 
 	tool := NewReadFileTool(workspace, true)
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"path": link,
 	})
@@ -278,10 +349,15 @@ func TestFilesystemTool_ReadFile_RejectsSymlinkEscape(t *testing.T) {
 	if !result.IsError {
 		t.Fatalf("expected symlink escape to be blocked")
 	}
+
 	// os.Root might return different errors depending on platform/implementation
+
 	// but it definitely should error.
+
 	// Our wrapper returns "access denied or file not found"
+
 	if !strings.Contains(result.ForLLM, "access denied") && !strings.Contains(result.ForLLM, "file not found") &&
+
 		!strings.Contains(result.ForLLM, "no such file") {
 		t.Fatalf("expected symlink escape error, got: %s", result.ForLLM)
 	}
@@ -291,8 +367,11 @@ func TestFilesystemTool_EmptyWorkspace_AccessDenied(t *testing.T) {
 	tool := NewReadFileTool("", true) // restrict=true but workspace=""
 
 	// Try to read a sensitive file (simulated by a temp file outside workspace)
+
 	tmpDir := t.TempDir()
+
 	secretFile := filepath.Join(tmpDir, "shadow")
+
 	os.WriteFile(secretFile, []byte("secret data"), 0o600)
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -300,201 +379,293 @@ func TestFilesystemTool_EmptyWorkspace_AccessDenied(t *testing.T) {
 	})
 
 	// We EXPECT IsError=true (access blocked due to empty workspace)
+
 	assert.True(t, result.IsError, "Security Regression: Empty workspace allowed access! content: %s", result.ForLLM)
 
 	// Verify it failed for the right reason
+
 	assert.Contains(t, result.ForLLM, "workspace is not defined", "Expected 'workspace is not defined' error")
 }
 
 // TestRootMkdirAll verifies that root.MkdirAll (used by sandboxFs.WriteFile) handles all cases:
+
 // single dir, deeply nested dirs, already-existing dirs, and a file blocking a directory path.
+
 func TestRootMkdirAll(t *testing.T) {
 	workspace := t.TempDir()
+
 	root, err := os.OpenRoot(workspace)
 	if err != nil {
 		t.Fatalf("failed to open root: %v", err)
 	}
+
 	defer root.Close()
 
 	// Case 1: Single directory
+
 	err = root.MkdirAll("dir1", 0o755)
+
 	assert.NoError(t, err)
+
 	_, err = os.Stat(filepath.Join(workspace, "dir1"))
+
 	assert.NoError(t, err)
 
 	// Case 2: Deeply nested directory
+
 	err = root.MkdirAll("a/b/c/d", 0o755)
+
 	assert.NoError(t, err)
+
 	_, err = os.Stat(filepath.Join(workspace, "a/b/c/d"))
+
 	assert.NoError(t, err)
 
 	// Case 3: Already exists — must be idempotent
+
 	err = root.MkdirAll("a/b/c/d", 0o755)
+
 	assert.NoError(t, err)
 
 	// Case 4: A regular file blocks directory creation — must error
+
 	err = os.WriteFile(filepath.Join(workspace, "file_exists"), []byte("data"), 0o644)
+
 	assert.NoError(t, err)
+
 	err = root.MkdirAll("file_exists", 0o755)
+
 	assert.Error(t, err, "expected error when a file exists at the directory path")
 }
 
 func TestFilesystemTool_WriteFile_Restricted_CreateDir(t *testing.T) {
 	workspace := t.TempDir()
+
 	tool := NewWriteFileTool(workspace, true)
+
 	ctx := context.Background()
 
 	testFile := "deep/nested/path/to/file.txt"
+
 	content := "deep content"
+
 	args := map[string]any{
-		"path":    testFile,
+		"path": testFile,
+
 		"content": content,
 	}
 
 	result := tool.Execute(ctx, args)
+
 	assert.False(t, result.IsError, "Expected success, got: %s", result.ForLLM)
 
 	// Verify file content
+
 	actualPath := filepath.Join(workspace, testFile)
+
 	data, err := os.ReadFile(actualPath)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, content, string(data))
 }
 
 // TestHostFs_Read_PermissionDenied verifies that hostFs.ReadFile surfaces access denied errors.
+
 func TestHostFs_Read_PermissionDenied(t *testing.T) {
 	if os.Getuid() == 0 {
 		t.Skip("skipping permission test: running as root")
 	}
+
 	tmpDir := t.TempDir()
+
 	protected := filepath.Join(tmpDir, "protected.txt")
+
 	err := os.WriteFile(protected, []byte("secret"), 0o000)
+
 	assert.NoError(t, err)
+
 	defer os.Chmod(protected, 0o644) // ensure cleanup
 
 	_, err = (&hostFs{}).ReadFile(protected)
+
 	assert.Error(t, err)
+
 	assert.Contains(t, err.Error(), "access denied")
 }
 
 // TestHostFs_Read_Directory verifies that hostFs.ReadFile returns an error when given a directory path.
+
 func TestHostFs_Read_Directory(t *testing.T) {
 	tmpDir := t.TempDir()
 
 	_, err := (&hostFs{}).ReadFile(tmpDir)
+
 	assert.Error(t, err, "expected error when reading a directory as a file")
 }
 
 // TestSandboxFs_Read_Directory verifies that sandboxFs.ReadFile returns an error when given a directory.
+
 func TestSandboxFs_Read_Directory(t *testing.T) {
 	workspace := t.TempDir()
+
 	root, err := os.OpenRoot(workspace)
+
 	assert.NoError(t, err)
+
 	defer root.Close()
 
 	// Create a subdirectory
+
 	err = root.Mkdir("subdir", 0o755)
+
 	assert.NoError(t, err)
 
 	_, err = (&sandboxFs{workspace: workspace}).ReadFile("subdir")
+
 	assert.Error(t, err, "expected error when reading a directory as a file")
 }
 
 // TestHostFs_Write_ParentDirMissing verifies that hostFs.WriteFile creates parent dirs automatically.
+
 func TestHostFs_Write_ParentDirMissing(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	target := filepath.Join(tmpDir, "a", "b", "c", "file.txt")
 
 	err := (&hostFs{}).WriteFile(target, []byte("hello"))
+
 	assert.NoError(t, err)
 
 	data, err := os.ReadFile(target)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, "hello", string(data))
 }
 
 // TestSandboxFs_Write_ParentDirMissing verifies that sandboxFs.WriteFile creates
+
 // nested parent directories automatically within the sandbox.
+
 func TestSandboxFs_Write_ParentDirMissing(t *testing.T) {
 	workspace := t.TempDir()
 
 	relPath := "x/y/z/file.txt"
+
 	err := (&sandboxFs{workspace: workspace}).WriteFile(relPath, []byte("nested"))
+
 	assert.NoError(t, err)
 
 	data, err := os.ReadFile(filepath.Join(workspace, relPath))
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, "nested", string(data))
 }
 
 // TestHostFs_Write verifies the hostFs.WriteFile helper function
+
 func TestHostFs_Write(t *testing.T) {
 	tmpDir := t.TempDir()
+
 	testFile := filepath.Join(tmpDir, "atomic_test.txt")
+
 	testData := []byte("atomic test content")
 
 	err := (&hostFs{}).WriteFile(testFile, testData)
+
 	assert.NoError(t, err)
 
 	content, err := os.ReadFile(testFile)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, testData, content)
 
 	// Verify it overwrites correctly
+
 	newData := []byte("new atomic content")
+
 	err = (&hostFs{}).WriteFile(testFile, newData)
+
 	assert.NoError(t, err)
 
 	content, err = os.ReadFile(testFile)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, newData, content)
 }
 
 // TestSandboxFs_Write verifies the sandboxFs.WriteFile helper function
+
 func TestSandboxFs_Write(t *testing.T) {
 	tmpDir := t.TempDir()
 
 	relPath := "atomic_root_test.txt"
+
 	testData := []byte("atomic root test content")
 
 	erw := &sandboxFs{workspace: tmpDir}
+
 	err := erw.WriteFile(relPath, testData)
+
 	assert.NoError(t, err)
 
 	root, err := os.OpenRoot(tmpDir)
+
 	assert.NoError(t, err)
+
 	defer root.Close()
 
 	f, err := root.Open(relPath)
+
 	assert.NoError(t, err)
+
 	defer f.Close()
 
 	content, err := io.ReadAll(f)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, testData, content)
 
 	// Verify it overwrites correctly
+
 	newData := []byte("new root atomic content")
+
 	err = erw.WriteFile(relPath, newData)
+
 	assert.NoError(t, err)
 
 	f2, err := root.Open(relPath)
+
 	assert.NoError(t, err)
+
 	defer f2.Close()
 
 	content, err = io.ReadAll(f2)
+
 	assert.NoError(t, err)
+
 	assert.Equal(t, newData, content)
 }
 
 // TestValidatePath_OutsideWorkspace_IncludesPath verifies that the access
+
 // denied error includes the workspace path so the caller knows the boundary.
+
 func TestValidatePath_OutsideWorkspace_IncludesPath(t *testing.T) {
 	workspace := t.TempDir()
+
 	outsidePath := filepath.Join(t.TempDir(), "secret.txt")
 
 	_, err := validatePath(outsidePath, workspace, true)
+
 	assert.Error(t, err)
+
 	assert.Contains(t, err.Error(), "access denied")
+
 	assert.Contains(t, err.Error(), workspace)
 }

pkg/tools/gitpush.go

@@ -12,35 +12,49 @@ import (
 )
 
 // worktreeInfoKey is the context key for passing WorktreeInfo to tools.
+
 type worktreeInfoKey struct{}
 
 // WithWorktreeInfo returns a context carrying the active WorktreeInfo.
+
 func WithWorktreeInfo(ctx context.Context, wt *git.WorktreeInfo) context.Context {
 	return context.WithValue(ctx, worktreeInfoKey{}, wt)
 }
 
 // WorktreeInfoFromCtx extracts the WorktreeInfo from context, or nil.
+
 func WorktreeInfoFromCtx(ctx context.Context) *git.WorktreeInfo {
 	if v, ok := ctx.Value(worktreeInfoKey{}).(*git.WorktreeInfo); ok {
 		return v
 	}
+
 	return nil
 }
 
 // protectedBranches are branch names that can never be pushed to.
+
 var protectedBranches = regexp.MustCompile(`^(main|master|develop|release/.*)$`)
 
 // GitPushTool implements safe git push restricted to worktree branches.
+
 //
+
 // Safety invariants:
+
 //   - Only works inside a worktree (WorktreeInfo must be in context)
+
 //   - Pushes only the worktree's branch — no arbitrary branch targets
+
 //   - Protected branches (main, master, develop, release/*) are blocked
+
 //   - Force push is never allowed
+
 //   - Auto-commits uncommitted changes before pushing
+
 type GitPushTool struct{}
 
 // NewGitPushTool creates a GitPushTool.
+
 func NewGitPushTool() *GitPushTool {
 	return &GitPushTool{}
 }
@@ -49,94 +63,138 @@ func (t *GitPushTool) Name() string { return "git_push" }
 
 func (t *GitPushTool) Description() string {
 	return "Push the current worktree branch to origin. Only works inside a git worktree. " +
+
 		"Auto-commits uncommitted changes before pushing. " +
+
 		"Protected branches (main, master, develop) cannot be pushed to. Force push is not allowed."
 }
 
 func (t *GitPushTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"commit_message": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Commit message for uncommitted changes. If omitted, uncommitted changes are auto-committed with a default message.",
 			},
 		},
+
 		"required": []string{},
 	}
 }
 
 func (t *GitPushTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	wt := WorktreeInfoFromCtx(ctx)
+
 	if wt == nil {
 		return ErrorResult(
+
 			"git_push requires an active worktree.\n" +
+
 				"This tool can only be used during worktree-based sessions " +
+
 				"(e.g., heartbeat tasks or plan executing phase).\n" +
+
 				"The worktree provides the branch name and isolation boundary — " +
+
 				"without it, git_push cannot determine which branch to push.")
 	}
 
 	branch := wt.Branch
+
 	if branch == "" {
 		return ErrorResult(
+
 			"worktree has no branch name.\n" +
+
 				"The WorktreeInfo was set but Branch is empty. " +
+
 				"This is an internal error — the worktree may not have been created correctly.")
 	}
 
 	// Block protected branches
+
 	if protectedBranches.MatchString(branch) {
 		return ErrorResult(fmt.Sprintf(
+
 			"cannot push to protected branch %q.\n"+
+
 				"Protected branches (main, master, develop, release/*) are blocked to prevent "+
+
 				"accidental overwrites. Work should be done on feature branches created by worktrees.",
+
 			branch))
 	}
 
 	// Auto-commit uncommitted changes
+
 	if git.HasUncommittedChanges(wt.Path) {
 		commitMsg := "auto: save before push"
+
 		if msg, ok := args["commit_message"].(string); ok && msg != "" {
 			commitMsg = msg
 		}
+
 		if err := git.AutoCommit(wt.Path, commitMsg); err != nil {
 			return ErrorResult(fmt.Sprintf(
+
 				"auto-commit failed before push: %v\n"+
+
 					"git_push auto-commits uncommitted changes before pushing. "+
+
 					"The commit failed, so no push was attempted. "+
+
 					"Check if the worktree at %q is in a valid state (e.g., no merge conflicts).",
+
 				err, wt.Path))
 		}
 	}
 
 	// Check there are commits to push
+
 	ahead := git.CommitsAhead(wt.RepoRoot, wt.BaseBranch, branch)
+
 	if ahead == 0 {
 		return NewToolResult(fmt.Sprintf(
+
 			"Nothing to push: branch %q has no commits ahead of %s.\n"+
+
 				"The branch is identical to the base. Make changes and commit before pushing.",
+
 			branch, wt.BaseBranch))
 	}
 
 	// Push with -u (set upstream tracking)
+
 	pushCtx, cancel := context.WithTimeout(ctx, 2*time.Minute)
+
 	defer cancel()
 
 	cmd := exec.CommandContext(pushCtx, "git", "push", "-u", "origin", branch)
+
 	cmd.Dir = wt.Path
+
 	out, err := cmd.CombinedOutput()
+
 	output := strings.TrimSpace(string(out))
 
 	if err != nil {
 		return ErrorResult(fmt.Sprintf(
+
 			"git push failed for branch %q: %s\n%s\n"+
+
 				"Possible causes: network error, authentication failure, or remote rejected the push. "+
+
 				"If the remote branch has diverged, resolve the divergence in the worktree first — "+
+
 				"force push is not available.",
+
 			branch, err, output))
 	}
 
 	return NewToolResult(fmt.Sprintf("Pushed branch %q to origin (%d commit(s) ahead of %s)\n%s",
+
 		branch, ahead, wt.BaseBranch, output))
 }

pkg/tools/gitpush_test.go

@@ -9,22 +9,29 @@ import (
 )
 
 // TestGitPushTool_NoWorktree verifies that git_push fails without worktree context.
+
 func TestGitPushTool_NoWorktree(t *testing.T) {
 	tool := NewGitPushTool()
 
 	result := tool.Execute(context.Background(), map[string]any{})
+
 	if !result.IsError {
 		t.Fatal("expected error when no worktree in context")
 	}
+
 	if result.ForLLM == "" {
 		t.Fatal("error message should not be empty")
 	}
+
 	// Verify helpful guidance is included
+
 	assertContains(t, result.ForLLM, "worktree")
+
 	assertContains(t, result.ForLLM, "heartbeat")
 }
 
 // TestGitPushTool_ProtectedBranch verifies that protected branches are blocked.
+
 func TestGitPushTool_ProtectedBranch(t *testing.T) {
 	tool := NewGitPushTool()
 
@@ -33,40 +40,56 @@ func TestGitPushTool_ProtectedBranch(t *testing.T) {
 	for _, branch := range protectedNames {
 		t.Run(branch, func(t *testing.T) {
 			ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-				Branch:     branch,
+				Branch: branch,
+
 				BaseBranch: "main",
-				Path:       t.TempDir(),
-				RepoRoot:   t.TempDir(),
+
+				Path: t.TempDir(),
+
+				RepoRoot: t.TempDir(),
 			})
+
 			result := tool.Execute(ctx, map[string]any{})
+
 			if !result.IsError {
 				t.Fatalf("expected error for protected branch %q", branch)
 			}
+
 			assertContains(t, result.ForLLM, "protected")
+
 			assertContains(t, result.ForLLM, branch)
 		})
 	}
 }
 
 // TestGitPushTool_EmptyBranch verifies that empty branch name is rejected.
+
 func TestGitPushTool_EmptyBranch(t *testing.T) {
 	tool := NewGitPushTool()
 
 	ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-		Branch:     "",
+		Branch: "",
+
 		BaseBranch: "main",
-		Path:       t.TempDir(),
-		RepoRoot:   t.TempDir(),
+
+		Path: t.TempDir(),
+
+		RepoRoot: t.TempDir(),
 	})
+
 	result := tool.Execute(ctx, map[string]any{})
+
 	if !result.IsError {
 		t.Fatal("expected error for empty branch")
 	}
+
 	assertContains(t, result.ForLLM, "no branch name")
 }
 
 // TestGitPushTool_AllowedBranch verifies that non-protected branches pass the branch check.
+
 // (Push itself will fail because there's no real git repo, but it should get past validation.)
+
 func TestGitPushTool_AllowedBranch(t *testing.T) {
 	tool := NewGitPushTool()
 
@@ -75,13 +98,19 @@ func TestGitPushTool_AllowedBranch(t *testing.T) {
 	for _, branch := range allowedNames {
 		t.Run(branch, func(t *testing.T) {
 			ctx := WithWorktreeInfo(context.Background(), &git.WorktreeInfo{
-				Branch:     branch,
+				Branch: branch,
+
 				BaseBranch: "main",
-				Path:       t.TempDir(),
-				RepoRoot:   t.TempDir(),
+
+				Path: t.TempDir(),
+
+				RepoRoot: t.TempDir(),
 			})
+
 			result := tool.Execute(ctx, map[string]any{})
+
 			// Should NOT fail with "protected branch" error
+
 			if result.IsError && strings.Contains(result.ForLLM, "protected") {
 				t.Fatalf("branch %q should not be blocked as protected", branch)
 			}
@@ -90,25 +119,36 @@ func TestGitPushTool_AllowedBranch(t *testing.T) {
 }
 
 // TestProtectedBranchesRegex tests the regex directly.
+
 func TestProtectedBranchesRegex(t *testing.T) {
 	tests := []struct {
-		branch    string
+		branch string
+
 		protected bool
 	}{
 		{"main", true},
+
 		{"master", true},
+
 		{"develop", true},
+
 		{"release/v1.0", true},
+
 		{"release/2026-03", true},
+
 		{"plan/add-feature", false},
+
 		{"feature/main", false}, // "main" not at start
-		{"main-backup", false},  // "main" followed by suffix
+
+		{"main-backup", false}, // "main" followed by suffix
+
 		{"hotfix/urgent", false},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.branch, func(t *testing.T) {
 			got := protectedBranches.MatchString(tt.branch)
+
 			if got != tt.protected {
 				t.Errorf("branch %q: got protected=%v, want %v", tt.branch, got, tt.protected)
 			}
@@ -117,48 +157,64 @@ func TestProtectedBranchesRegex(t *testing.T) {
 }
 
 // TestWorktreeInfoContext verifies context round-trip.
+
 func TestWorktreeInfoContext(t *testing.T) {
 	wt := &git.WorktreeInfo{
-		Branch:     "plan/test",
+		Branch: "plan/test",
+
 		BaseBranch: "main",
-		Path:       "/tmp/wt",
-		RepoRoot:   "/tmp/repo",
+
+		Path: "/tmp/wt",
+
+		RepoRoot: "/tmp/repo",
 	}
 
 	ctx := WithWorktreeInfo(context.Background(), wt)
+
 	got := WorktreeInfoFromCtx(ctx)
+
 	if got == nil {
 		t.Fatal("expected non-nil WorktreeInfo from context")
 	}
+
 	if got.Branch != wt.Branch {
 		t.Errorf("Branch: got %q, want %q", got.Branch, wt.Branch)
 	}
+
 	if got.BaseBranch != wt.BaseBranch {
 		t.Errorf("BaseBranch: got %q, want %q", got.BaseBranch, wt.BaseBranch)
 	}
 
 	// Nil case
+
 	got2 := WorktreeInfoFromCtx(context.Background())
+
 	if got2 != nil {
 		t.Errorf("expected nil WorktreeInfo from bare context, got %+v", got2)
 	}
 }
 
 // TestGitPushTool_Interface verifies the tool satisfies the Tool interface.
+
 func TestGitPushTool_Interface(t *testing.T) {
 	var _ Tool = (*GitPushTool)(nil)
 
 	tool := NewGitPushTool()
+
 	if tool.Name() != "git_push" {
 		t.Errorf("Name: got %q, want %q", tool.Name(), "git_push")
 	}
+
 	if tool.Description() == "" {
 		t.Error("Description should not be empty")
 	}
+
 	params := tool.Parameters()
+
 	if params == nil {
 		t.Fatal("Parameters should not be nil")
 	}
+
 	if params["type"] != "object" {
 		t.Errorf("Parameters type: got %v, want object", params["type"])
 	}
@@ -166,6 +222,7 @@ func TestGitPushTool_Interface(t *testing.T) {
 
 func assertContains(t *testing.T, s, substr string) {
 	t.Helper()
+
 	if !strings.Contains(s, substr) {
 		t.Errorf("expected %q to contain %q", s, substr)
 	}

pkg/tools/i2c.go

@@ -10,6 +10,7 @@ import (
 )
 
 // I2CTool provides I2C bus interaction for reading sensors and controlling peripherals.
+
 type I2CTool struct{}
 
 func NewI2CTool() *I2CTool {
@@ -27,38 +28,55 @@ func (t *I2CTool) Description() string {
 func (t *I2CTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"action": map[string]any{
-				"type":        "string",
-				"enum":        []string{"detect", "scan", "read", "write"},
+				"type": "string",
+
+				"enum": []string{"detect", "scan", "read", "write"},
+
 				"description": "Action to perform: detect (list available I2C buses), scan (find devices on a bus), read (read bytes from a device), write (send bytes to a device)",
 			},
+
 			"bus": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "I2C bus number (e.g. \"1\" for /dev/i2c-1). Required for scan/read/write.",
 			},
+
 			"address": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "7-bit I2C device address (0x03-0x77). Required for read/write.",
 			},
+
 			"register": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Register address to read from or write to. If set, sends register byte before read/write.",
 			},
+
 			"data": map[string]any{
-				"type":        "array",
-				"items":       map[string]any{"type": "integer"},
+				"type": "array",
+
+				"items": map[string]any{"type": "integer"},
+
 				"description": "Bytes to write (0-255 each). Required for write action.",
 			},
+
 			"length": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Number of bytes to read (1-256). Default: 1. Used with read action.",
 			},
+
 			"confirm": map[string]any{
-				"type":        "boolean",
+				"type": "boolean",
+
 				"description": "Must be true for write operations. Safety guard to prevent accidental writes.",
 			},
 		},
+
 		"required": []string{"action"},
 	}
 }
@@ -69,25 +87,36 @@ func (t *I2CTool) Execute(ctx context.Context, args map[string]any) *ToolResult
 	}
 
 	action, ok := args["action"].(string)
+
 	if !ok {
 		return ErrorResult("action is required")
 	}
 
 	switch action {
 	case "detect":
+
 		return t.detect()
+
 	case "scan":
+
 		return t.scan(args)
+
 	case "read":
+
 		return t.readDevice(args)
+
 	case "write":
+
 		return t.writeDevice(args)
+
 	default:
+
 		return ErrorResult(fmt.Sprintf("unknown action: %s (valid: detect, scan, read, write)", action))
 	}
 }
 
 // detect lists available I2C buses by globbing /dev/i2c-*
+
 func (t *I2CTool) detect() *ToolResult {
 	matches, err := filepath.Glob("/dev/i2c-*")
 	if err != nil {
@@ -102,11 +131,14 @@ func (t *I2CTool) detect() *ToolResult {
 
 	type busInfo struct {
 		Path string `json:"path"`
-		Bus  string `json:"bus"`
+
+		Bus string `json:"bus"`
 	}
 
 	buses := make([]busInfo, 0, len(matches))
+
 	re := regexp.MustCompile(`/dev/i2c-(\d+)`)
+
 	for _, m := range matches {
 		if sub := re.FindStringSubmatch(m); sub != nil {
 			buses = append(buses, busInfo{Path: m, Bus: sub[1]})
@@ -114,44 +146,62 @@ func (t *I2CTool) detect() *ToolResult {
 	}
 
 	result, _ := json.MarshalIndent(buses, "", "  ")
+
 	return SilentResult(fmt.Sprintf("Found %d I2C bus(es):\n%s", len(buses), string(result)))
 }
 
 // Helper functions for I2C operations (used by platform-specific implementations)
 
 // isValidBusID checks that a bus identifier is a simple number (prevents path injection)
+
 //
+
 //nolint:unused // Used by i2c_linux.go
+
 func isValidBusID(id string) bool {
 	matched, _ := regexp.MatchString(`^\d+$`, id)
+
 	return matched
 }
 
 // parseI2CAddress extracts and validates an I2C address from args
+
 //
+
 //nolint:unused // Used by i2c_linux.go
+
 func parseI2CAddress(args map[string]any) (int, *ToolResult) {
 	addrFloat, ok := args["address"].(float64)
+
 	if !ok {
 		return 0, ErrorResult("address is required (e.g. 0x38 for AHT20)")
 	}
+
 	addr := int(addrFloat)
+
 	if addr < 0x03 || addr > 0x77 {
 		return 0, ErrorResult("address must be in valid 7-bit range (0x03-0x77)")
 	}
+
 	return addr, nil
 }
 
 // parseI2CBus extracts and validates an I2C bus from args
+
 //
+
 //nolint:unused // Used by i2c_linux.go
+
 func parseI2CBus(args map[string]any) (string, *ToolResult) {
 	bus, ok := args["bus"].(string)
+
 	if !ok || bus == "" {
 		return "", ErrorResult("bus is required (e.g. \"1\" for /dev/i2c-1)")
 	}
+
 	if !isValidBusID(bus) {
 		return "", ErrorResult("invalid bus identifier: must be a number (e.g. \"1\")")
 	}
+
 	return bus, nil
 }

pkg/tools/i2c_linux.go

@@ -8,279 +8,465 @@ import (
 )
 
 // I2C ioctl constants from Linux kernel headers (<linux/i2c-dev.h>, <linux/i2c.h>)
+
 const (
 	i2cSlave = 0x0703 // Set slave address (fails if in use by driver)
+
 	i2cFuncs = 0x0705 // Query adapter functionality bitmask
+
 	i2cSmbus = 0x0720 // Perform SMBus transaction
 
 	// I2C_FUNC capability bits
-	i2cFuncSmbusQuick    = 0x00010000
+
+	i2cFuncSmbusQuick = 0x00010000
+
 	i2cFuncSmbusReadByte = 0x00020000
 
 	// SMBus transaction types
-	i2cSmbusRead  = 0
+
+	i2cSmbusRead = 0
+
 	i2cSmbusWrite = 1
 
 	// SMBus protocol sizes
+
 	i2cSmbusQuick = 0
-	i2cSmbusByte  = 1
+
+	i2cSmbusByte = 1
 )
 
 // i2cSmbusData matches the kernel union i2c_smbus_data (34 bytes max).
+
 // For quick and byte transactions only the first byte is used (if at all).
+
 type i2cSmbusData [34]byte
 
 // i2cSmbusArgs matches the kernel struct i2c_smbus_ioctl_data.
+
 type i2cSmbusArgs struct {
 	readWrite uint8
-	command   uint8
-	size      uint32
-	data      *i2cSmbusData
+
+	command uint8
+
+	size uint32
+
+	data *i2cSmbusData
 }
 
 // smbusProbe performs a single SMBus probe at the given address.
+
 // Uses SMBus Quick Write (safest) or falls back to SMBus Read Byte for
+
 // EEPROM address ranges where quick write can corrupt AT24RF08 chips.
+
 // This matches i2cdetect's MODE_AUTO behavior.
+
 func smbusProbe(fd int, addr int, hasQuick bool) bool {
+
 	// EEPROM ranges: use read byte (quick write can corrupt AT24RF08)
+
 	useReadByte := (addr >= 0x30 && addr <= 0x37) || (addr >= 0x50 && addr <= 0x5F)
 
 	if !useReadByte && hasQuick {
+
 		// SMBus Quick Write: [START] [ADDR|W] [ACK/NACK] [STOP]
+
 		// Safest probe — no data transferred
+
 		args := i2cSmbusArgs{
+
 			readWrite: i2cSmbusWrite,
-			command:   0,
-			size:      i2cSmbusQuick,
-			data:      nil,
+
+			command: 0,
+
+			size: i2cSmbusQuick,
+
+			data: nil,
 		}
+
 		_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cSmbus, uintptr(unsafe.Pointer(&args)))
+
 		return errno == 0
+
 	}
 
 	// SMBus Read Byte: [START] [ADDR|R] [ACK/NACK] [DATA] [STOP]
+
 	var data i2cSmbusData
+
 	args := i2cSmbusArgs{
+
 		readWrite: i2cSmbusRead,
-		command:   0,
-		size:      i2cSmbusByte,
-		data:      &data,
+
+		command: 0,
+
+		size: i2cSmbusByte,
+
+		data: &data,
 	}
+
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cSmbus, uintptr(unsafe.Pointer(&args)))
+
 	return errno == 0
+
 }
 
 // scan probes valid 7-bit addresses on a bus for connected devices.
+
 // Uses the same hybrid probe strategy as i2cdetect's MODE_AUTO:
+
 // SMBus Quick Write for most addresses, SMBus Read Byte for EEPROM ranges.
+
 func (t *I2CTool) scan(args map[string]any) *ToolResult {
+
 	bus, errResult := parseI2CBus(args)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
 
 	devPath := fmt.Sprintf("/dev/i2c-%s", bus)
+
 	fd, err := syscall.Open(devPath, syscall.O_RDWR, 0)
+
 	if err != nil {
+
 		return ErrorResult(fmt.Sprintf("failed to open %s: %v (check permissions and i2c-dev module)", devPath, err))
+
 	}
+
 	defer syscall.Close(fd)
 
 	// Query adapter capabilities to determine available probe methods.
+
 	// I2C_FUNCS writes an unsigned long, which is word-sized on Linux.
+
 	var funcs uintptr
+
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cFuncs, uintptr(unsafe.Pointer(&funcs)))
+
 	if errno != 0 {
+
 		return ErrorResult(fmt.Sprintf("failed to query I2C adapter capabilities on %s: %v", devPath, errno))
+
 	}
 
 	hasQuick := funcs&i2cFuncSmbusQuick != 0
+
 	hasReadByte := funcs&i2cFuncSmbusReadByte != 0
 
 	if !hasQuick && !hasReadByte {
+
 		return ErrorResult(
+
 			fmt.Sprintf("I2C adapter %s supports neither SMBus Quick nor Read Byte — cannot probe safely", devPath),
 		)
+
 	}
 
 	type deviceEntry struct {
 		Address string `json:"address"`
-		Status  string `json:"status,omitempty"`
+
+		Status string `json:"status,omitempty"`
 	}
 
 	var found []deviceEntry
+
 	// Scan 0x08-0x77, skipping I2C reserved addresses 0x00-0x07
+
 	for addr := 0x08; addr <= 0x77; addr++ {
+
 		// Set slave address — EBUSY means a kernel driver owns this address
+
 		_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cSlave, uintptr(addr))
+
 		if errno != 0 {
+
 			if errno == syscall.EBUSY {
+
 				found = append(found, deviceEntry{
+
 					Address: fmt.Sprintf("0x%02x", addr),
-					Status:  "busy (in use by kernel driver)",
+
+					Status: "busy (in use by kernel driver)",
 				})
+
 			}
+
 			continue
+
 		}
 
 		if smbusProbe(fd, addr, hasQuick) {
+
 			found = append(found, deviceEntry{
+
 				Address: fmt.Sprintf("0x%02x", addr),
 			})
+
 		}
+
 	}
 
 	if len(found) == 0 {
+
 		return SilentResult(fmt.Sprintf("No devices found on %s. Check wiring and pull-up resistors.", devPath))
+
 	}
 
 	result, _ := json.MarshalIndent(map[string]any{
-		"bus":     devPath,
+
+		"bus": devPath,
+
 		"devices": found,
-		"count":   len(found),
+
+		"count": len(found),
 	}, "", "  ")
+
 	return SilentResult(fmt.Sprintf("Scan of %s:\n%s", devPath, string(result)))
+
 }
 
 // readDevice reads bytes from an I2C device, optionally at a specific register
+
 func (t *I2CTool) readDevice(args map[string]any) *ToolResult {
+
 	bus, errResult := parseI2CBus(args)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
 
 	addr, errResult := parseI2CAddress(args)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
 
 	length := 1
+
 	if l, ok := args["length"].(float64); ok {
+
 		length = int(l)
+
 	}
+
 	if length < 1 || length > 256 {
+
 		return ErrorResult("length must be between 1 and 256")
+
 	}
 
 	devPath := fmt.Sprintf("/dev/i2c-%s", bus)
+
 	fd, err := syscall.Open(devPath, syscall.O_RDWR, 0)
+
 	if err != nil {
+
 		return ErrorResult(fmt.Sprintf("failed to open %s: %v", devPath, err))
+
 	}
+
 	defer syscall.Close(fd)
 
 	// Set slave address
+
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cSlave, uintptr(addr))
+
 	if errno != 0 {
+
 		return ErrorResult(fmt.Sprintf("failed to set I2C address 0x%02x: %v", addr, errno))
+
 	}
 
 	// If register is specified, write it first
+
 	if regFloat, ok := args["register"].(float64); ok {
+
 		reg := int(regFloat)
+
 		if reg < 0 || reg > 255 {
+
 			return ErrorResult("register must be between 0x00 and 0xFF")
+
 		}
+
 		_, err = syscall.Write(fd, []byte{byte(reg)})
+
 		if err != nil {
+
 			return ErrorResult(fmt.Sprintf("failed to write register 0x%02x: %v", reg, err))
+
 		}
+
 	}
 
 	// Read data
+
 	buf := make([]byte, length)
+
 	n, err := syscall.Read(fd, buf)
+
 	if err != nil {
+
 		return ErrorResult(fmt.Sprintf("failed to read from device 0x%02x: %v", addr, err))
+
 	}
 
 	// Format as hex bytes
+
 	hexBytes := make([]string, n)
+
 	intBytes := make([]int, n)
+
 	for i := 0; i < n; i++ {
+
 		hexBytes[i] = fmt.Sprintf("0x%02x", buf[i])
+
 		intBytes[i] = int(buf[i])
+
 	}
 
 	result, _ := json.MarshalIndent(map[string]any{
-		"bus":     devPath,
+
+		"bus": devPath,
+
 		"address": fmt.Sprintf("0x%02x", addr),
-		"bytes":   intBytes,
-		"hex":     hexBytes,
-		"length":  n,
+
+		"bytes": intBytes,
+
+		"hex": hexBytes,
+
+		"length": n,
 	}, "", "  ")
+
 	return SilentResult(string(result))
+
 }
 
 // writeDevice writes bytes to an I2C device, optionally at a specific register
+
 func (t *I2CTool) writeDevice(args map[string]any) *ToolResult {
+
 	confirm, _ := args["confirm"].(bool)
+
 	if !confirm {
+
 		return ErrorResult(
+
 			"write operations require confirm: true. Please confirm with the user before writing to I2C devices, as incorrect writes can misconfigure hardware.",
 		)
+
 	}
 
 	bus, errResult := parseI2CBus(args)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
 
 	addr, errResult := parseI2CAddress(args)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
 
 	dataRaw, ok := args["data"].([]any)
+
 	if !ok || len(dataRaw) == 0 {
+
 		return ErrorResult("data is required for write (array of byte values 0-255)")
+
 	}
+
 	if len(dataRaw) > 256 {
+
 		return ErrorResult("data too long: maximum 256 bytes per I2C transaction")
+
 	}
 
 	data := make([]byte, 0, len(dataRaw)+1)
 
 	// If register is specified, prepend it to the data
+
 	if regFloat, ok := args["register"].(float64); ok {
+
 		reg := int(regFloat)
+
 		if reg < 0 || reg > 255 {
+
 			return ErrorResult("register must be between 0x00 and 0xFF")
+
 		}
+
 		data = append(data, byte(reg))
+
 	}
 
 	for i, v := range dataRaw {
+
 		f, ok := v.(float64)
+
 		if !ok {
+
 			return ErrorResult(fmt.Sprintf("data[%d] is not a valid byte value", i))
+
 		}
+
 		b := int(f)
+
 		if b < 0 || b > 255 {
+
 			return ErrorResult(fmt.Sprintf("data[%d] = %d is out of byte range (0-255)", i, b))
+
 		}
+
 		data = append(data, byte(b))
+
 	}
 
 	devPath := fmt.Sprintf("/dev/i2c-%s", bus)
+
 	fd, err := syscall.Open(devPath, syscall.O_RDWR, 0)
+
 	if err != nil {
+
 		return ErrorResult(fmt.Sprintf("failed to open %s: %v", devPath, err))
+
 	}
+
 	defer syscall.Close(fd)
 
 	// Set slave address
+
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), i2cSlave, uintptr(addr))
+
 	if errno != 0 {
+
 		return ErrorResult(fmt.Sprintf("failed to set I2C address 0x%02x: %v", addr, errno))
+
 	}
 
 	// Write data
+
 	n, err := syscall.Write(fd, data)
+
 	if err != nil {
+
 		return ErrorResult(fmt.Sprintf("failed to write to device 0x%02x: %v", addr, err))
+
 	}
 
 	return SilentResult(fmt.Sprintf("Wrote %d byte(s) to device 0x%02x on %s", n, addr, devPath))
+
 }

pkg/tools/logs.go

@@ -9,7 +9,9 @@ import (
 )
 
 // LogsTool provides on-demand access to application logs from the in-memory ring buffer.
+
 // Designed for token-efficient log analysis: defaults to WARN level to exclude noise.
+
 type LogsTool struct{}
 
 func NewLogsTool() *LogsTool {
@@ -20,29 +22,40 @@ func (t *LogsTool) Name() string { return "logs" }
 
 func (t *LogsTool) Description() string {
 	return "Retrieve recent application logs from the in-memory ring buffer. " +
+
 		"Use level filter to minimize token usage (default: WARN). " +
+
 		"Call this when the user asks about errors, issues, or system health."
 }
 
 func (t *LogsTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"level": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Minimum log level: DEBUG, INFO, WARN, ERROR. Default: WARN",
-				"enum":        []string{"DEBUG", "INFO", "WARN", "ERROR"},
+
+				"enum": []string{"DEBUG", "INFO", "WARN", "ERROR"},
 			},
+
 			"component": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Filter by component name (e.g. telegram, discord, slack, agent)",
 			},
+
 			"limit": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Maximum number of log entries to return. Default: 50",
 			},
+
 			"query": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Filter by substring match in log message",
 			},
 		},
@@ -51,38 +64,50 @@ func (t *LogsTool) Parameters() map[string]any {
 
 func (t *LogsTool) Execute(_ context.Context, args map[string]any) *ToolResult {
 	// Parse level (default: WARN)
+
 	level := logger.WARN
+
 	if lvlStr, ok := args["level"].(string); ok && lvlStr != "" {
 		level = logger.ParseLevel(lvlStr)
 	}
 
 	// Parse component
+
 	component, _ := args["component"].(string)
 
 	// Parse limit (default: 50, max: 300)
+
 	limit := 50
+
 	if l, ok := args["limit"].(float64); ok && l > 0 {
 		limit = int(l)
 	}
+
 	if limit > 300 {
 		limit = 300
 	}
 
 	// Parse query
+
 	query, _ := args["query"].(string)
 
 	// Fetch from ring buffer (already sanitized by RecentLogs)
+
 	entries := logger.RecentLogs(level, component, limit)
 
 	// Apply query filter if specified
+
 	if query != "" {
 		filtered := make([]logger.LogEntry, 0, len(entries))
+
 		queryLower := strings.ToLower(query)
+
 		for _, e := range entries {
 			if strings.Contains(strings.ToLower(e.Message), queryLower) {
 				filtered = append(filtered, e)
 			}
 		}
+
 		entries = filtered
 	}
 

pkg/tools/logs_test.go

@@ -11,22 +11,31 @@ import (
 
 func setupTestLogs(t *testing.T) {
 	t.Helper()
+
 	prev := logger.GetLevel()
+
 	t.Cleanup(func() { logger.SetLevel(prev) })
+
 	logger.SetLevel(logger.DEBUG)
 
 	logger.DebugC("agent", "debug message")
+
 	logger.InfoC("telegram", "message received")
+
 	logger.WarnC("telegram", "webhook retry")
+
 	logger.ErrorC("discord", "connection timeout")
+
 	logger.WarnCF("wecom", "signature failed", map[string]any{
 		"token": "secret-value",
+
 		"nonce": "safe-value",
 	})
 }
 
 func TestLogsTool_DefaultLevel(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{})
@@ -36,6 +45,7 @@ func TestLogsTool_DefaultLevel(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -49,6 +59,7 @@ func TestLogsTool_DefaultLevel(t *testing.T) {
 
 func TestLogsTool_LevelFilter(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
@@ -60,6 +71,7 @@ func TestLogsTool_LevelFilter(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -73,10 +85,12 @@ func TestLogsTool_LevelFilter(t *testing.T) {
 
 func TestLogsTool_ComponentFilter(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
-		"level":     "DEBUG",
+		"level": "DEBUG",
+
 		"component": "telegram",
 	})
 
@@ -85,6 +99,7 @@ func TestLogsTool_ComponentFilter(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -98,10 +113,12 @@ func TestLogsTool_ComponentFilter(t *testing.T) {
 
 func TestLogsTool_QueryFilter(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"level": "DEBUG",
+
 		"query": "timeout",
 	})
 
@@ -110,6 +127,7 @@ func TestLogsTool_QueryFilter(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -117,6 +135,7 @@ func TestLogsTool_QueryFilter(t *testing.T) {
 	if len(entries) == 0 {
 		t.Fatal("expected at least one entry matching 'timeout'")
 	}
+
 	for _, e := range entries {
 		if !strings.Contains(strings.ToLower(e.Message), "timeout") {
 			t.Errorf("entry should contain 'timeout': %s", e.Message)
@@ -126,14 +145,17 @@ func TestLogsTool_QueryFilter(t *testing.T) {
 
 func TestLogsTool_QueryCaseInsensitive(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"level": "DEBUG",
+
 		"query": "TIMEOUT",
 	})
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -145,10 +167,12 @@ func TestLogsTool_QueryCaseInsensitive(t *testing.T) {
 
 func TestLogsTool_Limit(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
 		"level": "DEBUG",
+
 		"limit": float64(2),
 	})
 
@@ -157,6 +181,7 @@ func TestLogsTool_Limit(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
@@ -170,12 +195,15 @@ func TestLogsTool_LimitMax(t *testing.T) {
 	tool := NewLogsTool()
 
 	// limit > 300 should be capped
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"level": "DEBUG",
+
 		"limit": float64(999),
 	})
 
 	// Should not error, just cap silently
+
 	if result.IsError {
 		t.Fatalf("unexpected error: %s", result.ForLLM)
 	}
@@ -183,10 +211,12 @@ func TestLogsTool_LimitMax(t *testing.T) {
 
 func TestLogsTool_FieldsSanitized(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
-		"level":     "WARN",
+		"level": "WARN",
+
 		"component": "wecom",
 	})
 
@@ -195,22 +225,27 @@ func TestLogsTool_FieldsSanitized(t *testing.T) {
 	}
 
 	var entries []logger.LogEntry
+
 	if err := json.Unmarshal([]byte(result.ForLLM), &entries); err != nil {
 		t.Fatalf("failed to parse result: %v", err)
 	}
 
 	found := false
+
 	for _, e := range entries {
 		if e.Fields != nil && e.Fields["token"] != nil {
 			found = true
+
 			if e.Fields["token"] != "***" {
 				t.Errorf("token field should be sanitized, got %v", e.Fields["token"])
 			}
+
 			if e.Fields["nonce"] != "safe-value" {
 				t.Errorf("nonce field should be preserved, got %v", e.Fields["nonce"])
 			}
 		}
 	}
+
 	if !found {
 		t.Error("expected to find wecom entry with token field")
 	}
@@ -218,19 +253,23 @@ func TestLogsTool_FieldsSanitized(t *testing.T) {
 
 func TestLogsTool_NoResults(t *testing.T) {
 	prev := logger.GetLevel()
+
 	defer logger.SetLevel(prev)
+
 	logger.SetLevel(logger.DEBUG)
 
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{
-		"level":     "ERROR",
+		"level": "ERROR",
+
 		"component": "nonexistent-component-xyz",
 	})
 
 	if result.IsError {
 		t.Fatalf("should not be an error result: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "No log entries found") {
 		t.Errorf("expected 'No log entries found' message, got: %s", result.ForLLM)
 	}
@@ -238,9 +277,11 @@ func TestLogsTool_NoResults(t *testing.T) {
 
 func TestLogsTool_Silent(t *testing.T) {
 	setupTestLogs(t)
+
 	tool := NewLogsTool()
 
 	result := tool.Execute(context.Background(), map[string]any{})
+
 	if !result.Silent {
 		t.Error("logs tool result should be Silent")
 	}
@@ -252,10 +293,13 @@ func TestLogsTool_ToolInterface(t *testing.T) {
 	if tool.Name() != "logs" {
 		t.Errorf("expected name 'logs', got %q", tool.Name())
 	}
+
 	if tool.Description() == "" {
 		t.Error("description should not be empty")
 	}
+
 	params := tool.Parameters()
+
 	if params == nil {
 		t.Error("parameters should not be nil")
 	}

pkg/tools/message.go

@@ -8,10 +8,13 @@ import (
 type SendCallback func(channel, chatID, content string) error
 
 type MessageTool struct {
-	sendCallback   SendCallback
+	sendCallback SendCallback
+
 	defaultChannel string
-	defaultChatID  string
-	sentInRound    bool // Tracks whether a message was sent in the current processing round
+
+	defaultChatID string
+
+	sentInRound bool // Tracks whether a message was sent in the current processing round
 }
 
 func NewMessageTool() *MessageTool {
@@ -29,31 +32,41 @@ func (t *MessageTool) Description() string {
 func (t *MessageTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"content": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The message content to send",
 			},
+
 			"channel": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Optional: target channel (telegram, whatsapp, etc.)",
 			},
+
 			"chat_id": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Optional: target chat/user ID",
 			},
 		},
+
 		"required": []string{"content"},
 	}
 }
 
 func (t *MessageTool) SetContext(channel, chatID string) {
 	t.defaultChannel = channel
+
 	t.defaultChatID = chatID
+
 	t.sentInRound = false // Reset send tracking for new processing round
 }
 
 // HasSentInRound returns true if the message tool sent a message during the current round.
+
 func (t *MessageTool) HasSentInRound() bool {
 	return t.sentInRound
 }
@@ -64,16 +77,19 @@ func (t *MessageTool) SetSendCallback(callback SendCallback) {
 
 func (t *MessageTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	content, ok := args["content"].(string)
+
 	if !ok {
 		return &ToolResult{ForLLM: "content is required", IsError: true}
 	}
 
 	channel, _ := args["channel"].(string)
+
 	chatID, _ := args["chat_id"].(string)
 
 	if channel == "" {
 		channel = t.defaultChannel
 	}
+
 	if chatID == "" {
 		chatID = t.defaultChatID
 	}
@@ -88,16 +104,21 @@ func (t *MessageTool) Execute(ctx context.Context, args map[string]any) *ToolRes
 
 	if err := t.sendCallback(channel, chatID, content); err != nil {
 		return &ToolResult{
-			ForLLM:  fmt.Sprintf("sending message: %v", err),
+			ForLLM: fmt.Sprintf("sending message: %v", err),
+
 			IsError: true,
-			Err:     err,
+
+			Err: err,
 		}
 	}
 
 	t.sentInRound = true
+
 	// Silent: user already received the message directly
+
 	return &ToolResult{
 		ForLLM: fmt.Sprintf("Message sent to %s:%s", channel, chatID),
+
 		Silent: true,
 	}
 }

pkg/tools/message_test.go

@@ -8,17 +8,23 @@ import (
 
 func TestMessageTool_Execute_Success(t *testing.T) {
 	tool := NewMessageTool()
+
 	tool.SetContext("test-channel", "test-chat-id")
 
 	var sentChannel, sentChatID, sentContent string
+
 	tool.SetSendCallback(func(channel, chatID, content string) error {
 		sentChannel = channel
+
 		sentChatID = chatID
+
 		sentContent = content
+
 		return nil
 	})
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "Hello, world!",
 	}
@@ -26,33 +32,41 @@ func TestMessageTool_Execute_Success(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Verify message was sent with correct parameters
+
 	if sentChannel != "test-channel" {
 		t.Errorf("Expected channel 'test-channel', got '%s'", sentChannel)
 	}
+
 	if sentChatID != "test-chat-id" {
 		t.Errorf("Expected chatID 'test-chat-id', got '%s'", sentChatID)
 	}
+
 	if sentContent != "Hello, world!" {
 		t.Errorf("Expected content 'Hello, world!', got '%s'", sentContent)
 	}
 
 	// Verify ToolResult meets US-011 criteria:
+
 	// - Send success returns SilentResult (Silent=true)
+
 	if !result.Silent {
 		t.Error("Expected Silent=true for successful send")
 	}
 
 	// - ForLLM contains send status description
+
 	if result.ForLLM != "Message sent to test-channel:test-chat-id" {
 		t.Errorf("Expected ForLLM 'Message sent to test-channel:test-chat-id', got '%s'", result.ForLLM)
 	}
 
 	// - ForUser is empty (user already received message directly)
+
 	if result.ForUser != "" {
 		t.Errorf("Expected ForUser to be empty, got '%s'", result.ForUser)
 	}
 
 	// - IsError should be false
+
 	if result.IsError {
 		t.Error("Expected IsError=false for successful send")
 	}
@@ -60,28 +74,37 @@ func TestMessageTool_Execute_Success(t *testing.T) {
 
 func TestMessageTool_Execute_WithCustomChannel(t *testing.T) {
 	tool := NewMessageTool()
+
 	tool.SetContext("default-channel", "default-chat-id")
 
 	var sentChannel, sentChatID string
+
 	tool.SetSendCallback(func(channel, chatID, content string) error {
 		sentChannel = channel
+
 		sentChatID = chatID
+
 		return nil
 	})
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "Test message",
+
 		"channel": "custom-channel",
+
 		"chat_id": "custom-chat-id",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Verify custom channel/chatID were used instead of defaults
+
 	if sentChannel != "custom-channel" {
 		t.Errorf("Expected channel 'custom-channel', got '%s'", sentChannel)
 	}
+
 	if sentChatID != "custom-chat-id" {
 		t.Errorf("Expected chatID 'custom-chat-id', got '%s'", sentChatID)
 	}
@@ -89,6 +112,7 @@ func TestMessageTool_Execute_WithCustomChannel(t *testing.T) {
 	if !result.Silent {
 		t.Error("Expected Silent=true")
 	}
+
 	if result.ForLLM != "Message sent to custom-channel:custom-chat-id" {
 		t.Errorf("Expected ForLLM 'Message sent to custom-channel:custom-chat-id', got '%s'", result.ForLLM)
 	}
@@ -96,14 +120,17 @@ func TestMessageTool_Execute_WithCustomChannel(t *testing.T) {
 
 func TestMessageTool_Execute_SendFailure(t *testing.T) {
 	tool := NewMessageTool()
+
 	tool.SetContext("test-channel", "test-chat-id")
 
 	sendErr := errors.New("network error")
+
 	tool.SetSendCallback(func(channel, chatID, content string) error {
 		return sendErr
 	})
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "Test message",
 	}
@@ -111,21 +138,27 @@ func TestMessageTool_Execute_SendFailure(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Verify ToolResult for send failure:
+
 	// - Send failure returns ErrorResult (IsError=true)
+
 	if !result.IsError {
 		t.Error("Expected IsError=true for failed send")
 	}
 
 	// - ForLLM contains error description
+
 	expectedErrMsg := "sending message: network error"
+
 	if result.ForLLM != expectedErrMsg {
 		t.Errorf("Expected ForLLM '%s', got '%s'", expectedErrMsg, result.ForLLM)
 	}
 
 	// - Err field should contain original error
+
 	if result.Err == nil {
 		t.Error("Expected Err to be set")
 	}
+
 	if result.Err != sendErr {
 		t.Errorf("Expected Err to be sendErr, got %v", result.Err)
 	}
@@ -133,17 +166,21 @@ func TestMessageTool_Execute_SendFailure(t *testing.T) {
 
 func TestMessageTool_Execute_MissingContent(t *testing.T) {
 	tool := NewMessageTool()
+
 	tool.SetContext("test-channel", "test-chat-id")
 
 	ctx := context.Background()
+
 	args := map[string]any{} // content missing
 
 	result := tool.Execute(ctx, args)
 
 	// Verify error result for missing content
+
 	if !result.IsError {
 		t.Error("Expected IsError=true for missing content")
 	}
+
 	if result.ForLLM != "content is required" {
 		t.Errorf("Expected ForLLM 'content is required', got '%s'", result.ForLLM)
 	}
@@ -151,6 +188,7 @@ func TestMessageTool_Execute_MissingContent(t *testing.T) {
 
 func TestMessageTool_Execute_NoTargetChannel(t *testing.T) {
 	tool := NewMessageTool()
+
 	// No SetContext called, so defaultChannel and defaultChatID are empty
 
 	tool.SetSendCallback(func(channel, chatID, content string) error {
@@ -158,6 +196,7 @@ func TestMessageTool_Execute_NoTargetChannel(t *testing.T) {
 	})
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "Test message",
 	}
@@ -165,9 +204,11 @@ func TestMessageTool_Execute_NoTargetChannel(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Verify error when no target channel specified
+
 	if !result.IsError {
 		t.Error("Expected IsError=true when no target channel")
 	}
+
 	if result.ForLLM != "No target channel/chat specified" {
 		t.Errorf("Expected ForLLM 'No target channel/chat specified', got '%s'", result.ForLLM)
 	}
@@ -175,10 +216,13 @@ func TestMessageTool_Execute_NoTargetChannel(t *testing.T) {
 
 func TestMessageTool_Execute_NotConfigured(t *testing.T) {
 	tool := NewMessageTool()
+
 	tool.SetContext("test-channel", "test-chat-id")
+
 	// No SetSendCallback called
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"content": "Test message",
 	}
@@ -186,9 +230,11 @@ func TestMessageTool_Execute_NotConfigured(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Verify error when send callback not configured
+
 	if !result.IsError {
 		t.Error("Expected IsError=true when send callback not configured")
 	}
+
 	if result.ForLLM != "Message sending not configured" {
 		t.Errorf("Expected ForLLM 'Message sending not configured', got '%s'", result.ForLLM)
 	}
@@ -196,6 +242,7 @@ func TestMessageTool_Execute_NotConfigured(t *testing.T) {
 
 func TestMessageTool_Name(t *testing.T) {
 	tool := NewMessageTool()
+
 	if tool.Name() != "message" {
 		t.Errorf("Expected name 'message', got '%s'", tool.Name())
 	}
@@ -203,7 +250,9 @@ func TestMessageTool_Name(t *testing.T) {
 
 func TestMessageTool_Description(t *testing.T) {
 	tool := NewMessageTool()
+
 	desc := tool.Description()
+
 	if desc == "" {
 		t.Error("Description should not be empty")
 	}
@@ -211,48 +260,63 @@ func TestMessageTool_Description(t *testing.T) {
 
 func TestMessageTool_Parameters(t *testing.T) {
 	tool := NewMessageTool()
+
 	params := tool.Parameters()
 
 	// Verify parameters structure
+
 	typ, ok := params["type"].(string)
+
 	if !ok || typ != "object" {
 		t.Error("Expected type 'object'")
 	}
 
 	props, ok := params["properties"].(map[string]any)
+
 	if !ok {
 		t.Fatal("Expected properties to be a map")
 	}
 
 	// Check required properties
+
 	required, ok := params["required"].([]string)
+
 	if !ok || len(required) != 1 || required[0] != "content" {
 		t.Error("Expected 'content' to be required")
 	}
 
 	// Check content property
+
 	contentProp, ok := props["content"].(map[string]any)
+
 	if !ok {
 		t.Error("Expected 'content' property")
 	}
+
 	if contentProp["type"] != "string" {
 		t.Error("Expected content type to be 'string'")
 	}
 
 	// Check channel property (optional)
+
 	channelProp, ok := props["channel"].(map[string]any)
+
 	if !ok {
 		t.Error("Expected 'channel' property")
 	}
+
 	if channelProp["type"] != "string" {
 		t.Error("Expected channel type to be 'string'")
 	}
 
 	// Check chat_id property (optional)
+
 	chatIDProp, ok := props["chat_id"].(map[string]any)
+
 	if !ok {
 		t.Error("Expected 'chat_id' property")
 	}
+
 	if chatIDProp["type"] != "string" {
 		t.Error("Expected chat_id type to be 'string'")
 	}

pkg/tools/registry.go

@@ -14,9 +14,12 @@ import (
 )
 
 // NormalizeToolName keeps only lowercase ASCII letters.
+
 // "read_file" → "readfile", "ReadFile" → "readfile", "read-file" → "readfile".
+
 func NormalizeToolName(s string) string {
 	var b strings.Builder
+
 	for _, r := range s {
 		if r >= 'A' && r <= 'Z' {
 			b.WriteRune(r + 32)
@@ -24,12 +27,14 @@ func NormalizeToolName(s string) string {
 			b.WriteRune(r)
 		}
 	}
+
 	return b.String()
 }
 
 type ToolRegistry struct {
 	tools map[string]Tool
-	mu    sync.RWMutex
+
+	mu sync.RWMutex
 }
 
 func NewToolRegistry() *ToolRegistry {
@@ -40,24 +45,33 @@ func NewToolRegistry() *ToolRegistry {
 
 func (r *ToolRegistry) Register(tool Tool) {
 	r.mu.Lock()
+
 	defer r.mu.Unlock()
+
 	r.tools[tool.Name()] = tool
 }
 
 func (r *ToolRegistry) Get(name string) (Tool, bool) {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
+
 	// Exact match first
+
 	if tool, ok := r.tools[name]; ok {
 		return tool, true
 	}
+
 	// Fuzzy fallback: normalize and compare (handles "readfile" → "read_file" etc.)
+
 	norm := NormalizeToolName(name)
+
 	for _, tool := range r.tools {
 		if NormalizeToolName(tool.Name()) == norm {
 			return tool, true
 		}
 	}
+
 	return nil, false
 }
 
@@ -66,70 +80,99 @@ func (r *ToolRegistry) Execute(ctx context.Context, name string, args map[string
 }
 
 // ExecuteWithContext executes a tool with channel/chatID context and optional async callback.
+
 // If the tool implements AsyncTool and a non-nil callback is provided,
+
 // the callback will be set on the tool before execution.
+
 func (r *ToolRegistry) ExecuteWithContext(
 	ctx context.Context,
+
 	name string,
+
 	args map[string]any,
+
 	channel, chatID string,
+
 	asyncCallback AsyncCallback,
 ) *ToolResult {
 	logger.InfoCF("tool", "Tool execution started",
+
 		map[string]any{
 			"tool": name,
+
 			"args": args,
 		})
 
 	tool, ok := r.Get(name)
+
 	if !ok {
 		available := strings.Join(r.List(), ", ")
+
 		logger.ErrorCF("tool", "Tool not found",
+
 			map[string]any{
 				"tool": name,
 			})
+
 		return ErrorResult(fmt.Sprintf(
+
 			"tool %q not found. Available tools: %s", name, available,
 		)).WithError(fmt.Errorf("tool not found"))
 	}
 
 	// If tool implements ContextualTool, set context
+
 	if contextualTool, ok := tool.(ContextualTool); ok && channel != "" && chatID != "" {
 		contextualTool.SetContext(channel, chatID)
 	}
 
 	// If tool implements AsyncTool and callback is provided, set callback
+
 	if asyncTool, ok := tool.(AsyncTool); ok && asyncCallback != nil {
 		asyncTool.SetCallback(asyncCallback)
+
 		logger.DebugCF("tool", "Async callback injected",
+
 			map[string]any{
 				"tool": name,
 			})
 	}
 
 	start := time.Now()
+
 	result := tool.Execute(ctx, args)
+
 	duration := time.Since(start)
 
 	// Log based on result type
+
 	if result.IsError {
 		logger.ErrorCF("tool", "Tool execution failed",
+
 			map[string]any{
-				"tool":     name,
+				"tool": name,
+
 				"duration": duration.Milliseconds(),
-				"error":    result.ForLLM,
+
+				"error": result.ForLLM,
 			})
 	} else if result.Async {
 		logger.InfoCF("tool", "Tool started (async)",
+
 			map[string]any{
-				"tool":     name,
+				"tool": name,
+
 				"duration": duration.Milliseconds(),
 			})
 	} else {
 		logger.InfoCF("tool", "Tool execution completed",
+
 			map[string]any{
-				"tool":          name,
-				"duration_ms":   duration.Milliseconds(),
+				"tool": name,
+
+				"duration_ms": duration.Milliseconds(),
+
 				"result_length": len(result.ForLLM),
 			})
 	}
@@ -138,53 +181,75 @@ func (r *ToolRegistry) ExecuteWithContext(
 }
 
 // sortedToolNames returns tool names in sorted order for deterministic iteration.
+
 // This is critical for KV cache stability: non-deterministic map iteration would
+
 // produce different system prompts and tool definitions on each call, invalidating
+
 // the LLM's prefix cache even when no tools have changed.
+
 func (r *ToolRegistry) sortedToolNames() []string {
 	names := make([]string, 0, len(r.tools))
+
 	for name := range r.tools {
 		names = append(names, name)
 	}
+
 	sort.Strings(names)
+
 	return names
 }
 
 func (r *ToolRegistry) GetDefinitions() []map[string]any {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
 
 	sorted := r.sortedToolNames()
+
 	definitions := make([]map[string]any, 0, len(sorted))
+
 	for _, name := range sorted {
 		definitions = append(definitions, ToolToSchema(r.tools[name]))
 	}
+
 	return definitions
 }
 
 // ToProviderDefs converts tool definitions to provider-compatible format.
+
 // This is the format expected by LLM provider APIs.
+
 func (r *ToolRegistry) ToProviderDefs() []providers.ToolDefinition {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
 
 	sorted := r.sortedToolNames()
+
 	definitions := make([]providers.ToolDefinition, 0, len(sorted))
+
 	for _, name := range sorted {
 		tool := r.tools[name]
+
 		schema := ToolToSchema(tool)
 
 		// Safely extract nested values with type checks
+
 		fn, ok := schema["function"].(map[string]any)
+
 		if !ok {
 			continue
 		}
 
 		name, _ := fn["name"].(string)
+
 		desc, _ := fn["description"].(string)
+
 		params, _ := fn["parameters"].(map[string]any)
 
 		paramsRaw := json.RawMessage(`{}`)
+
 		if len(params) > 0 {
 			if payload, err := json.Marshal(params); err == nil {
 				paramsRaw = json.RawMessage(payload)
@@ -193,38 +258,51 @@ func (r *ToolRegistry) ToProviderDefs() []providers.ToolDefinition {
 
 		definitions = append(definitions, providers.ToolDefinition{
 			Type: "function",
+
 			Function: providers.ToolFunctionDefinition{
-				Name:        name,
+				Name: name,
+
 				Description: desc,
-				Parameters:  paramsRaw,
+
+				Parameters: paramsRaw,
 			},
 		})
 	}
+
 	return definitions
 }
 
 // List returns a list of all registered tool names.
+
 func (r *ToolRegistry) List() []string {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
 
 	return r.sortedToolNames()
 }
 
 // Count returns the number of registered tools.
+
 func (r *ToolRegistry) Count() int {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
+
 	return len(r.tools)
 }
 
 // GetRuntimeStatus aggregates runtime status from all tools that implement StatusProvider.
+
 // Returns empty string if no tool has status to report.
+
 func (r *ToolRegistry) GetRuntimeStatus() string {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
 
 	var parts []string
+
 	for _, tool := range r.tools {
 		if sp, ok := tool.(StatusProvider); ok {
 			if s := sp.RuntimeStatus(); s != "" {
@@ -232,40 +310,53 @@ func (r *ToolRegistry) GetRuntimeStatus() string {
 			}
 		}
 	}
+
 	if len(parts) == 0 {
 		return ""
 	}
+
 	return strings.Join(parts, "\n\n")
 }
 
 // buildParamHint extracts parameter names from a JSON schema and returns
+
 // a hint string like "(task, label?, preset?)". Required params are bare,
+
 // optional params have a trailing "?".
+
 func buildParamHint(schema map[string]any) string {
 	props, _ := schema["properties"].(map[string]any)
+
 	if len(props) == 0 {
 		return ""
 	}
 
 	reqSlice, _ := schema["required"].([]string)
+
 	reqSet := make(map[string]bool, len(reqSlice))
+
 	for _, r := range reqSlice {
 		reqSet[r] = true
 	}
 
 	names := make([]string, 0, len(props))
+
 	for name := range props {
 		names = append(names, name)
 	}
+
 	sort.Strings(names)
 
 	parts := make([]string, 0, len(names))
+
 	// Required params first, then optional
+
 	for _, name := range names {
 		if reqSet[name] {
 			parts = append(parts, name)
 		}
 	}
+
 	for _, name := range names {
 		if !reqSet[name] {
 			parts = append(parts, name+"?")
@@ -276,17 +367,25 @@ func buildParamHint(schema map[string]any) string {
 }
 
 // GetSummaries returns human-readable summaries of all registered tools.
+
 // Returns a slice of "- `name`(params) - description" strings.
+
 func (r *ToolRegistry) GetSummaries() []string {
 	r.mu.RLock()
+
 	defer r.mu.RUnlock()
 
 	sorted := r.sortedToolNames()
+
 	summaries := make([]string, 0, len(sorted))
+
 	for _, name := range sorted {
 		tool := r.tools[name]
+
 		hint := buildParamHint(tool.Parameters())
+
 		summaries = append(summaries, fmt.Sprintf("- `%s`%s - %s", tool.Name(), hint, tool.Description()))
 	}
+
 	return summaries
 }

pkg/tools/registry_test.go

@@ -12,32 +12,42 @@ import (
 // --- mock types ---
 
 type mockRegistryTool struct {
-	name   string
-	desc   string
+	name string
+
+	desc string
+
 	params map[string]any
+
 	result *ToolResult
 }
 
-func (m *mockRegistryTool) Name() string               { return m.name }
-func (m *mockRegistryTool) Description() string        { return m.desc }
+func (m *mockRegistryTool) Name() string { return m.name }
+
+func (m *mockRegistryTool) Description() string { return m.desc }
+
 func (m *mockRegistryTool) Parameters() map[string]any { return m.params }
+
 func (m *mockRegistryTool) Execute(_ context.Context, _ map[string]any) *ToolResult {
 	return m.result
 }
 
 type mockCtxTool struct {
 	mockRegistryTool
+
 	channel string
-	chatID  string
+
+	chatID string
 }
 
 func (m *mockCtxTool) SetContext(channel, chatID string) {
 	m.channel = channel
+
 	m.chatID = chatID
 }
 
 type mockAsyncRegistryTool struct {
 	mockRegistryTool
+
 	cb AsyncCallback
 }
 
@@ -49,9 +59,12 @@ func (m *mockAsyncRegistryTool) SetCallback(cb AsyncCallback) {
 
 func newMockTool(name, desc string) *mockRegistryTool {
 	return &mockRegistryTool{
-		name:   name,
-		desc:   desc,
+		name: name,
+
+		desc: desc,
+
 		params: map[string]any{"type": "object"},
+
 		result: SilentResult("ok"),
 	}
 }
@@ -63,15 +76,23 @@ func TestNormalizeToolName(t *testing.T) {
 		input, want string
 	}{
 		{"read_file", "readfile"},
+
 		{"readfile", "readfile"},
+
 		{"ReadFile", "readfile"},
+
 		{"read-file", "readfile"},
+
 		{"edit_file", "editfile"},
+
 		{"web_search", "websearch"},
+
 		{"EXEC", "exec"},
 	}
+
 	for _, tt := range tests {
 		got := NormalizeToolName(tt.input)
+
 		if got != tt.want {
 			t.Errorf("NormalizeToolName(%q) = %q, want %q", tt.input, got, tt.want)
 		}
@@ -80,9 +101,11 @@ func TestNormalizeToolName(t *testing.T) {
 
 func TestNewToolRegistry(t *testing.T) {
 	r := NewToolRegistry()
+
 	if r.Count() != 0 {
 		t.Errorf("expected empty registry, got count %d", r.Count())
 	}
+
 	if len(r.List()) != 0 {
 		t.Errorf("expected empty list, got %v", r.List())
 	}
@@ -90,13 +113,17 @@ func TestNewToolRegistry(t *testing.T) {
 
 func TestToolRegistry_RegisterAndGet(t *testing.T) {
 	r := NewToolRegistry()
+
 	tool := newMockTool("echo", "echoes input")
+
 	r.Register(tool)
 
 	got, ok := r.Get("echo")
+
 	if !ok {
 		t.Fatal("expected to find registered tool")
 	}
+
 	if got.Name() != "echo" {
 		t.Errorf("expected name 'echo', got %q", got.Name())
 	}
@@ -104,7 +131,9 @@ func TestToolRegistry_RegisterAndGet(t *testing.T) {
 
 func TestToolRegistry_Get_NotFound(t *testing.T) {
 	r := NewToolRegistry()
+
 	_, ok := r.Get("nonexistent")
+
 	if ok {
 		t.Error("expected ok=false for unregistered tool")
 	}
@@ -112,28 +141,42 @@ func TestToolRegistry_Get_NotFound(t *testing.T) {
 
 func TestToolRegistry_Get_FuzzyMatch(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(newMockTool("read_file", "reads a file"))
+
 	r.Register(newMockTool("edit_file", "edits a file"))
+
 	r.Register(newMockTool("web_search", "searches the web"))
 
 	tests := []struct {
-		query    string
+		query string
+
 		wantName string
 	}{
 		{"readfile", "read_file"},
+
 		{"ReadFile", "read_file"},
+
 		{"read-file", "read_file"},
+
 		{"editfile", "edit_file"},
+
 		{"EditFile", "edit_file"},
+
 		{"websearch", "web_search"},
+
 		{"WebSearch", "web_search"},
 	}
+
 	for _, tt := range tests {
 		tool, ok := r.Get(tt.query)
+
 		if !ok {
 			t.Errorf("Get(%q) not found, want %q", tt.query, tt.wantName)
+
 			continue
 		}
+
 		if tool.Name() != tt.wantName {
 			t.Errorf("Get(%q).Name() = %q, want %q", tt.query, tool.Name(), tt.wantName)
 		}
@@ -142,13 +185,17 @@ func TestToolRegistry_Get_FuzzyMatch(t *testing.T) {
 
 func TestToolRegistry_RegisterOverwrite(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(newMockTool("dup", "first"))
+
 	r.Register(newMockTool("dup", "second"))
 
 	if r.Count() != 1 {
 		t.Errorf("expected count 1 after overwrite, got %d", r.Count())
 	}
+
 	tool, _ := r.Get("dup")
+
 	if tool.Description() != "second" {
 		t.Errorf("expected overwritten description 'second', got %q", tool.Description())
 	}
@@ -156,17 +203,23 @@ func TestToolRegistry_RegisterOverwrite(t *testing.T) {
 
 func TestToolRegistry_Execute_Success(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(&mockRegistryTool{
-		name:   "greet",
-		desc:   "says hello",
+		name: "greet",
+
+		desc: "says hello",
+
 		params: map[string]any{},
+
 		result: SilentResult("hello"),
 	})
 
 	result := r.Execute(context.Background(), "greet", nil)
+
 	if result.IsError {
 		t.Errorf("expected success, got error: %s", result.ForLLM)
 	}
+
 	if result.ForLLM != "hello" {
 		t.Errorf("expected ForLLM 'hello', got %q", result.ForLLM)
 	}
@@ -174,13 +227,17 @@ func TestToolRegistry_Execute_Success(t *testing.T) {
 
 func TestToolRegistry_Execute_NotFound(t *testing.T) {
 	r := NewToolRegistry()
+
 	result := r.Execute(context.Background(), "missing", nil)
+
 	if !result.IsError {
 		t.Error("expected error for missing tool")
 	}
+
 	if !strings.Contains(result.ForLLM, "not found") {
 		t.Errorf("expected 'not found' in error, got %q", result.ForLLM)
 	}
+
 	if result.Err == nil {
 		t.Error("expected Err to be set via WithError")
 	}
@@ -188,9 +245,11 @@ func TestToolRegistry_Execute_NotFound(t *testing.T) {
 
 func TestToolRegistry_ExecuteWithContext_ContextualTool(t *testing.T) {
 	r := NewToolRegistry()
+
 	ct := &mockCtxTool{
 		mockRegistryTool: *newMockTool("ctx_tool", "needs context"),
 	}
+
 	r.Register(ct)
 
 	r.ExecuteWithContext(context.Background(), "ctx_tool", nil, "telegram", "chat-42", nil)
@@ -198,6 +257,7 @@ func TestToolRegistry_ExecuteWithContext_ContextualTool(t *testing.T) {
 	if ct.channel != "telegram" {
 		t.Errorf("expected channel 'telegram', got %q", ct.channel)
 	}
+
 	if ct.chatID != "chat-42" {
 		t.Errorf("expected chatID 'chat-42', got %q", ct.chatID)
 	}
@@ -205,9 +265,11 @@ func TestToolRegistry_ExecuteWithContext_ContextualTool(t *testing.T) {
 
 func TestToolRegistry_ExecuteWithContext_SkipsEmptyContext(t *testing.T) {
 	r := NewToolRegistry()
+
 	ct := &mockCtxTool{
 		mockRegistryTool: *newMockTool("ctx_tool", "needs context"),
 	}
+
 	r.Register(ct)
 
 	r.ExecuteWithContext(context.Background(), "ctx_tool", nil, "", "", nil)
@@ -219,24 +281,31 @@ func TestToolRegistry_ExecuteWithContext_SkipsEmptyContext(t *testing.T) {
 
 func TestToolRegistry_ExecuteWithContext_AsyncCallback(t *testing.T) {
 	r := NewToolRegistry()
+
 	at := &mockAsyncRegistryTool{
 		mockRegistryTool: *newMockTool("async_tool", "async work"),
 	}
+
 	at.result = AsyncResult("started")
+
 	r.Register(at)
 
 	called := false
+
 	cb := func(_ context.Context, _ *ToolResult) { called = true }
 
 	result := r.ExecuteWithContext(context.Background(), "async_tool", nil, "", "", cb)
+
 	if at.cb == nil {
 		t.Error("expected SetCallback to have been called")
 	}
+
 	if !result.Async {
 		t.Error("expected async result")
 	}
 
 	at.cb(context.Background(), SilentResult("done"))
+
 	if !called {
 		t.Error("expected callback to be invoked")
 	}
@@ -244,22 +313,29 @@ func TestToolRegistry_ExecuteWithContext_AsyncCallback(t *testing.T) {
 
 func TestToolRegistry_GetDefinitions(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(newMockTool("alpha", "tool A"))
 
 	defs := r.GetDefinitions()
+
 	if len(defs) != 1 {
 		t.Fatalf("expected 1 definition, got %d", len(defs))
 	}
+
 	if defs[0]["type"] != "function" {
 		t.Errorf("expected type 'function', got %v", defs[0]["type"])
 	}
+
 	fn, ok := defs[0]["function"].(map[string]any)
+
 	if !ok {
 		t.Fatal("expected 'function' key to be a map")
 	}
+
 	if fn["name"] != "alpha" {
 		t.Errorf("expected name 'alpha', got %v", fn["name"])
 	}
+
 	if fn["description"] != "tool A" {
 		t.Errorf("expected description 'tool A', got %v", fn["description"])
 	}
@@ -267,34 +343,47 @@ func TestToolRegistry_GetDefinitions(t *testing.T) {
 
 func TestToolRegistry_ToProviderDefs(t *testing.T) {
 	r := NewToolRegistry()
+
 	params := map[string]any{"type": "object", "properties": map[string]any{}}
+
 	r.Register(&mockRegistryTool{
-		name:   "beta",
-		desc:   "tool B",
+		name: "beta",
+
+		desc: "tool B",
+
 		params: params,
+
 		result: SilentResult("ok"),
 	})
 
 	defs := r.ToProviderDefs()
+
 	if len(defs) != 1 {
 		t.Fatalf("expected 1 provider def, got %d", len(defs))
 	}
 
 	want := providers.ToolDefinition{
 		Type: "function",
+
 		Function: providers.ToolFunctionDefinition{
-			Name:        "beta",
+			Name: "beta",
+
 			Description: "tool B",
-			Parameters:  providers.MustMarshalParameters(params),
+
+			Parameters: providers.MustMarshalParameters(params),
 		},
 	}
+
 	got := defs[0]
+
 	if got.Type != want.Type {
 		t.Errorf("Type: want %q, got %q", want.Type, got.Type)
 	}
+
 	if got.Function.Name != want.Function.Name {
 		t.Errorf("Name: want %q, got %q", want.Function.Name, got.Function.Name)
 	}
+
 	if got.Function.Description != want.Function.Description {
 		t.Errorf("Description: want %q, got %q", want.Function.Description, got.Function.Description)
 	}
@@ -302,18 +391,23 @@ func TestToolRegistry_ToProviderDefs(t *testing.T) {
 
 func TestToolRegistry_List(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(newMockTool("x", ""))
+
 	r.Register(newMockTool("y", ""))
 
 	names := r.List()
+
 	if len(names) != 2 {
 		t.Fatalf("expected 2 names, got %d", len(names))
 	}
 
 	nameSet := map[string]bool{}
+
 	for _, n := range names {
 		nameSet[n] = true
 	}
+
 	if !nameSet["x"] || !nameSet["y"] {
 		t.Errorf("expected names {x, y}, got %v", names)
 	}
@@ -321,17 +415,21 @@ func TestToolRegistry_List(t *testing.T) {
 
 func TestToolRegistry_Count(t *testing.T) {
 	r := NewToolRegistry()
+
 	if r.Count() != 0 {
 		t.Errorf("expected 0, got %d", r.Count())
 	}
 
 	r.Register(newMockTool("a", ""))
+
 	r.Register(newMockTool("b", ""))
+
 	if r.Count() != 2 {
 		t.Errorf("expected 2, got %d", r.Count())
 	}
 
 	r.Register(newMockTool("a", "replaced"))
+
 	if r.Count() != 2 {
 		t.Errorf("expected 2 after overwrite, got %d", r.Count())
 	}
@@ -339,62 +437,91 @@ func TestToolRegistry_Count(t *testing.T) {
 
 func TestBuildParamHint(t *testing.T) {
 	tests := []struct {
-		name   string
+		name string
+
 		schema map[string]any
-		want   string
+
+		want string
 	}{
 		{
 			name: "required and optional",
+
 			schema: map[string]any{
 				"type": "object",
+
 				"properties": map[string]any{
-					"task":  map[string]any{"type": "string"},
+					"task": map[string]any{"type": "string"},
+
 					"label": map[string]any{"type": "string"},
 				},
+
 				"required": []string{"task"},
 			},
+
 			want: "(task, label?)",
 		},
+
 		{
 			name: "all required",
+
 			schema: map[string]any{
 				"type": "object",
+
 				"properties": map[string]any{
 					"command": map[string]any{"type": "string"},
 				},
+
 				"required": []string{"command"},
 			},
+
 			want: "(command)",
 		},
+
 		{
 			name: "no properties",
+
 			schema: map[string]any{
 				"type": "object",
 			},
+
 			want: "",
 		},
+
 		{
-			name:   "empty schema",
+			name: "empty schema",
+
 			schema: map[string]any{},
-			want:   "",
+
+			want: "",
 		},
+
 		{
-			name:   "nil schema",
+			name: "nil schema",
+
 			schema: nil,
-			want:   "",
+
+			want: "",
 		},
+
 		{
 			name: "multiple optional sorted",
+
 			schema: map[string]any{
 				"type": "object",
+
 				"properties": map[string]any{
-					"task":     map[string]any{"type": "string"},
-					"preset":   map[string]any{"type": "string"},
-					"label":    map[string]any{"type": "string"},
+					"task": map[string]any{"type": "string"},
+
+					"preset": map[string]any{"type": "string"},
+
+					"label": map[string]any{"type": "string"},
+
 					"agent_id": map[string]any{"type": "string"},
 				},
+
 				"required": []string{"task"},
 			},
+
 			want: "(task, agent_id?, label?, preset?)",
 		},
 	}
@@ -402,6 +529,7 @@ func TestBuildParamHint(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			got := buildParamHint(tt.schema)
+
 			if got != tt.want {
 				t.Errorf("buildParamHint() = %q, want %q", got, tt.want)
 			}
@@ -411,15 +539,19 @@ func TestBuildParamHint(t *testing.T) {
 
 func TestToolRegistry_GetSummaries(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(newMockTool("read_file", "Reads a file"))
 
 	summaries := r.GetSummaries()
+
 	if len(summaries) != 1 {
 		t.Fatalf("expected 1 summary, got %d", len(summaries))
 	}
+
 	if !strings.Contains(summaries[0], "`read_file`") {
 		t.Errorf("expected backtick-quoted name in summary, got %q", summaries[0])
 	}
+
 	if !strings.Contains(summaries[0], "Reads a file") {
 		t.Errorf("expected description in summary, got %q", summaries[0])
 	}
@@ -427,25 +559,35 @@ func TestToolRegistry_GetSummaries(t *testing.T) {
 
 func TestToolRegistry_GetSummaries_WithParamHint(t *testing.T) {
 	r := NewToolRegistry()
+
 	r.Register(&mockRegistryTool{
 		name: "spawn",
+
 		desc: "Spawn a subagent",
+
 		params: map[string]any{
 			"type": "object",
+
 			"properties": map[string]any{
-				"task":   map[string]any{"type": "string"},
+				"task": map[string]any{"type": "string"},
+
 				"preset": map[string]any{"type": "string"},
 			},
+
 			"required": []string{"task"},
 		},
+
 		result: SilentResult("ok"),
 	})
 
 	summaries := r.GetSummaries()
+
 	if len(summaries) != 1 {
 		t.Fatalf("expected 1 summary, got %d", len(summaries))
 	}
+
 	// Should contain param hint
+
 	if !strings.Contains(summaries[0], "(task, preset?)") {
 		t.Errorf("expected param hint in summary, got %q", summaries[0])
 	}
@@ -453,21 +595,27 @@ func TestToolRegistry_GetSummaries_WithParamHint(t *testing.T) {
 
 func TestToolToSchema(t *testing.T) {
 	tool := newMockTool("demo", "demo tool")
+
 	schema := ToolToSchema(tool)
 
 	if schema["type"] != "function" {
 		t.Errorf("expected type 'function', got %v", schema["type"])
 	}
+
 	fn, ok := schema["function"].(map[string]any)
+
 	if !ok {
 		t.Fatal("expected 'function' to be a map")
 	}
+
 	if fn["name"] != "demo" {
 		t.Errorf("expected name 'demo', got %v", fn["name"])
 	}
+
 	if fn["description"] != "demo tool" {
 		t.Errorf("expected description 'demo tool', got %v", fn["description"])
 	}
+
 	if fn["parameters"] == nil {
 		t.Error("expected parameters to be set")
 	}
@@ -475,17 +623,25 @@ func TestToolToSchema(t *testing.T) {
 
 func TestToolRegistry_ConcurrentAccess(t *testing.T) {
 	r := NewToolRegistry()
+
 	var wg sync.WaitGroup
 
 	for i := range 50 {
 		wg.Add(1)
+
 		go func(n int) {
 			defer wg.Done()
+
 			name := string(rune('A' + n%26))
+
 			r.Register(newMockTool(name, "concurrent"))
+
 			r.Get(name)
+
 			r.Count()
+
 			r.List()
+
 			r.GetDefinitions()
 		}(i)
 	}

pkg/tools/result_test.go

@@ -12,12 +12,15 @@ func TestNewToolResult(t *testing.T) {
 	if result.ForLLM != "test content" {
 		t.Errorf("Expected ForLLM 'test content', got '%s'", result.ForLLM)
 	}
+
 	if result.Silent {
 		t.Error("Expected Silent to be false")
 	}
+
 	if result.IsError {
 		t.Error("Expected IsError to be false")
 	}
+
 	if result.Async {
 		t.Error("Expected Async to be false")
 	}
@@ -29,12 +32,15 @@ func TestSilentResult(t *testing.T) {
 	if result.ForLLM != "silent operation" {
 		t.Errorf("Expected ForLLM 'silent operation', got '%s'", result.ForLLM)
 	}
+
 	if !result.Silent {
 		t.Error("Expected Silent to be true")
 	}
+
 	if result.IsError {
 		t.Error("Expected IsError to be false")
 	}
+
 	if result.Async {
 		t.Error("Expected Async to be false")
 	}
@@ -46,12 +52,15 @@ func TestAsyncResult(t *testing.T) {
 	if result.ForLLM != "async task started" {
 		t.Errorf("Expected ForLLM 'async task started', got '%s'", result.ForLLM)
 	}
+
 	if result.Silent {
 		t.Error("Expected Silent to be false")
 	}
+
 	if result.IsError {
 		t.Error("Expected IsError to be false")
 	}
+
 	if !result.Async {
 		t.Error("Expected Async to be true")
 	}
@@ -63,12 +72,15 @@ func TestErrorResult(t *testing.T) {
 	if result.ForLLM != "operation failed" {
 		t.Errorf("Expected ForLLM 'operation failed', got '%s'", result.ForLLM)
 	}
+
 	if result.Silent {
 		t.Error("Expected Silent to be false")
 	}
+
 	if !result.IsError {
 		t.Error("Expected IsError to be true")
 	}
+
 	if result.Async {
 		t.Error("Expected Async to be false")
 	}
@@ -76,20 +88,25 @@ func TestErrorResult(t *testing.T) {
 
 func TestUserResult(t *testing.T) {
 	content := "user visible message"
+
 	result := UserResult(content)
 
 	if result.ForLLM != content {
 		t.Errorf("Expected ForLLM '%s', got '%s'", content, result.ForLLM)
 	}
+
 	if result.ForUser != content {
 		t.Errorf("Expected ForUser '%s', got '%s'", content, result.ForUser)
 	}
+
 	if result.Silent {
 		t.Error("Expected Silent to be false")
 	}
+
 	if result.IsError {
 		t.Error("Expected IsError to be false")
 	}
+
 	if result.Async {
 		t.Error("Expected Async to be false")
 	}
@@ -97,27 +114,37 @@ func TestUserResult(t *testing.T) {
 
 func TestToolResultJSONSerialization(t *testing.T) {
 	tests := []struct {
-		name   string
+		name string
+
 		result *ToolResult
 	}{
 		{
-			name:   "basic result",
+			name: "basic result",
+
 			result: NewToolResult("basic content"),
 		},
+
 		{
-			name:   "silent result",
+			name: "silent result",
+
 			result: SilentResult("silent content"),
 		},
+
 		{
-			name:   "async result",
+			name: "async result",
+
 			result: AsyncResult("async content"),
 		},
+
 		{
-			name:   "error result",
+			name: "error result",
+
 			result: ErrorResult("error content"),
 		},
+
 		{
-			name:   "user result",
+			name: "user result",
+
 			result: UserResult("user content"),
 		},
 	}
@@ -125,30 +152,38 @@ func TestToolResultJSONSerialization(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			// Marshal to JSON
+
 			data, err := json.Marshal(tt.result)
 			if err != nil {
 				t.Fatalf("Failed to marshal: %v", err)
 			}
 
 			// Unmarshal back
+
 			var decoded ToolResult
+
 			if err := json.Unmarshal(data, &decoded); err != nil {
 				t.Fatalf("Failed to unmarshal: %v", err)
 			}
 
 			// Verify fields match (Err should be excluded)
+
 			if decoded.ForLLM != tt.result.ForLLM {
 				t.Errorf("ForLLM mismatch: got '%s', want '%s'", decoded.ForLLM, tt.result.ForLLM)
 			}
+
 			if decoded.ForUser != tt.result.ForUser {
 				t.Errorf("ForUser mismatch: got '%s', want '%s'", decoded.ForUser, tt.result.ForUser)
 			}
+
 			if decoded.Silent != tt.result.Silent {
 				t.Errorf("Silent mismatch: got %v, want %v", decoded.Silent, tt.result.Silent)
 			}
+
 			if decoded.IsError != tt.result.IsError {
 				t.Errorf("IsError mismatch: got %v, want %v", decoded.IsError, tt.result.IsError)
 			}
+
 			if decoded.Async != tt.result.Async {
 				t.Errorf("Async mismatch: got %v, want %v", decoded.Async, tt.result.Async)
 			}
@@ -158,22 +193,27 @@ func TestToolResultJSONSerialization(t *testing.T) {
 
 func TestToolResultWithErrors(t *testing.T) {
 	err := errors.New("underlying error")
+
 	result := ErrorResult("error message").WithError(err)
 
 	if result.Err == nil {
 		t.Error("Expected Err to be set")
 	}
+
 	if result.Err.Error() != "underlying error" {
 		t.Errorf("Expected Err message 'underlying error', got '%s'", result.Err.Error())
 	}
 
 	// Verify Err is not serialized
+
 	data, marshalErr := json.Marshal(result)
+
 	if marshalErr != nil {
 		t.Fatalf("Failed to marshal: %v", marshalErr)
 	}
 
 	var decoded ToolResult
+
 	if unmarshalErr := json.Unmarshal(data, &decoded); unmarshalErr != nil {
 		t.Fatalf("Failed to unmarshal: %v", unmarshalErr)
 	}
@@ -192,37 +232,47 @@ func TestToolResultJSONStructure(t *testing.T) {
 	}
 
 	// Verify JSON structure
+
 	var parsed map[string]any
+
 	if err := json.Unmarshal(data, &parsed); err != nil {
 		t.Fatalf("Failed to parse JSON: %v", err)
 	}
 
 	// Check expected keys exist
+
 	if _, ok := parsed["for_llm"]; !ok {
 		t.Error("Expected 'for_llm' key in JSON")
 	}
+
 	if _, ok := parsed["for_user"]; !ok {
 		t.Error("Expected 'for_user' key in JSON")
 	}
+
 	if _, ok := parsed["silent"]; !ok {
 		t.Error("Expected 'silent' key in JSON")
 	}
+
 	if _, ok := parsed["is_error"]; !ok {
 		t.Error("Expected 'is_error' key in JSON")
 	}
+
 	if _, ok := parsed["async"]; !ok {
 		t.Error("Expected 'async' key in JSON")
 	}
 
 	// Check that 'err' is NOT present (it should have json:"-" tag)
+
 	if _, ok := parsed["err"]; ok {
 		t.Error("Expected 'err' key to be excluded from JSON")
 	}
 
 	// Verify values
+
 	if parsed["for_llm"] != "test content" {
 		t.Errorf("Expected for_llm 'test content', got %v", parsed["for_llm"])
 	}
+
 	if parsed["silent"] != false {
 		t.Errorf("Expected silent false, got %v", parsed["silent"])
 	}

pkg/tools/session_recorder.go

@@ -16,4 +16,10 @@ type SessionRecorder interface {
 
 	// RecordReport injects a TurnReport into the conductor session.
 	RecordReport(conductorSessionKey, subagentSessionKey, senderID, content string) error
+
+	// RecordQuestion injects a TurnQuestion into the conductor session (subagent escalation).
+	RecordQuestion(conductorKey, subagentKey, taskID, question string) error
+
+	// RecordPlanSubmit injects a TurnPlanSubmit into the conductor session (plan review request).
+	RecordPlanSubmit(conductorKey, subagentKey, taskID, planText string) error
 }

pkg/tools/shell_process_unix.go

@@ -11,75 +11,129 @@ import (
 )
 
 func prepareCommandForTermination(cmd *exec.Cmd) {
+
 	if cmd == nil {
+
 		return
+
 	}
+
 	cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
+
 }
 
 func terminateProcessTree(cmd *exec.Cmd) error {
+
 	if cmd == nil || cmd.Process == nil {
+
 		return nil
+
 	}
 
 	pid := cmd.Process.Pid
+
 	if pid <= 0 {
+
 		return nil
+
 	}
 
 	// Kill the entire process group spawned by the shell command.
+
 	_ = syscall.Kill(-pid, syscall.SIGKILL)
+
 	// Some shells/background jobs may still leave descendants around
+
 	// briefly; aggressively walk /proc and kill child processes too.
+
 	killDescendants(pid)
+
 	// Fallback kill on the shell process itself.
+
 	_ = cmd.Process.Kill()
+
 	return nil
+
 }
 
 func killDescendants(ppid int) {
+
 	if ppid <= 0 {
+
 		return
+
 	}
 
 	entries, err := os.ReadDir("/proc")
+
 	if err != nil {
+
 		return
+
 	}
 
 	for _, e := range entries {
+
 		if !e.IsDir() {
+
 			continue
+
 		}
+
 		childPID, err := strconv.Atoi(e.Name())
+
 		if err != nil || childPID <= 0 || childPID == ppid {
+
 			continue
+
 		}
 
 		statPath := "/proc/" + e.Name() + "/stat"
+
 		data, err := os.ReadFile(statPath)
+
 		if err != nil {
+
 			continue
+
 		}
 
 		// /proc/<pid>/stat: pid (comm) state ppid ...
+
 		raw := string(data)
+
 		end := strings.LastIndex(raw, ")")
+
 		if end == -1 || end+2 >= len(raw) {
+
 			continue
+
 		}
+
 		fields := strings.Fields(raw[end+2:])
+
 		if len(fields) < 2 {
+
 			continue
+
 		}
+
 		parent, err := strconv.Atoi(fields[1])
+
 		if err != nil || parent != ppid {
+
 			continue
+
 		}
 
 		// Recurse first, then kill child process/group.
+
 		killDescendants(childPID)
+
 		_ = syscall.Kill(-childPID, syscall.SIGKILL)
+
 		_ = syscall.Kill(childPID, syscall.SIGKILL)
+
 	}
+
 }

pkg/tools/shell_process_windows.go

@@ -17,11 +17,14 @@ func terminateProcessTree(cmd *exec.Cmd) error {
 	}
 
 	pid := cmd.Process.Pid
+
 	if pid <= 0 {
 		return nil
 	}
 
 	_ = exec.Command("taskkill", "/T", "/F", "/PID", strconv.Itoa(pid)).Run()
+
 	_ = cmd.Process.Kill()
+
 	return nil
 }

pkg/tools/shell_timeout_unix_test.go

@@ -14,73 +14,122 @@ import (
 )
 
 func processRunning(pid int) bool {
+
 	if pid <= 0 {
+
 		return false
+
 	}
+
 	// kill(0) can return success for zombie processes too, so inspect /proc
+
 	// state and treat zombies as not-running for timeout cleanup assertions.
+
 	err := syscall.Kill(pid, 0)
+
 	if err != nil && err != syscall.EPERM {
+
 		return false
+
 	}
 
 	data, readErr := os.ReadFile("/proc/" + strconv.Itoa(pid) + "/stat")
+
 	if readErr != nil {
+
 		return false
+
 	}
+
 	raw := string(data)
+
 	end := strings.LastIndex(raw, ")")
+
 	if end == -1 || end+2 >= len(raw) {
+
 		return true // best effort fallback
+
 	}
+
 	fields := strings.Fields(raw[end+2:])
+
 	if len(fields) == 0 {
+
 		return true // best effort fallback
+
 	}
 
 	state := fields[0]
+
 	return state != "Z"
+
 }
 
 func TestShellTool_TimeoutKillsChildProcess(t *testing.T) {
+
 	tool, err := NewExecTool(t.TempDir(), false)
+
 	if err != nil {
+
 		t.Errorf("unable to configure exec tool: %s", err)
+
 	}
 
 	tool.SetTimeout(500 * time.Millisecond)
 
 	args := map[string]any{
+
 		// Spawn a child process that would outlive the shell unless process-group kill is used.
+
 		"command": "sleep 60 & echo $! > child.pid; wait",
 	}
 
 	result := tool.Execute(context.Background(), args)
+
 	if !result.IsError {
+
 		t.Fatalf("expected timeout error, got success: %s", result.ForLLM)
+
 	}
+
 	if !strings.Contains(result.ForLLM, "timed out") {
+
 		t.Fatalf("expected timeout message, got: %s", result.ForLLM)
+
 	}
 
 	childPIDPath := filepath.Join(tool.workingDir, "child.pid")
+
 	data, err := os.ReadFile(childPIDPath)
+
 	if err != nil {
+
 		t.Fatalf("failed to read child pid file: %v", err)
+
 	}
 
 	childPID, err := strconv.Atoi(strings.TrimSpace(string(data)))
+
 	if err != nil {
+
 		t.Fatalf("failed to parse child pid: %v", err)
+
 	}
 
 	deadline := time.Now().Add(2 * time.Second)
+
 	for time.Now().Before(deadline) {
+
 		if !processRunning(childPID) {
+
 			return
+
 		}
+
 		time.Sleep(50 * time.Millisecond)
+
 	}
 
 	t.Fatalf("child process %d is still running after timeout", childPID)
+
 }

pkg/tools/skills_install.go

@@ -16,22 +16,32 @@ import (
 )
 
 // InstallSkillTool allows the LLM agent to install skills from registries.
+
 // It shares the same RegistryManager that FindSkillsTool uses,
+
 // so all registries configured in config are available for installation.
+
 type InstallSkillTool struct {
 	registryMgr *skills.RegistryManager
-	workspace   string
-	mu          sync.Mutex
+
+	workspace string
+
+	mu sync.Mutex
 }
 
 // NewInstallSkillTool creates a new InstallSkillTool.
+
 // registryMgr is the shared registry manager (same instance as FindSkillsTool).
+
 // workspace is the root workspace directory; skills install to {workspace}/skills/{slug}/.
+
 func NewInstallSkillTool(registryMgr *skills.RegistryManager, workspace string) *InstallSkillTool {
 	return &InstallSkillTool{
 		registryMgr: registryMgr,
-		workspace:   workspace,
-		mu:          sync.Mutex{},
+
+		workspace: workspace,
+
+		mu: sync.Mutex{},
 	}
 }
 
@@ -46,151 +56,210 @@ func (t *InstallSkillTool) Description() string {
 func (t *InstallSkillTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"slug": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The unique slug of the skill to install (e.g., 'github', 'docker-compose')",
 			},
+
 			"version": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Specific version to install (optional, defaults to latest)",
 			},
+
 			"registry": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Registry to install from (required, e.g., 'clawhub')",
 			},
+
 			"force": map[string]any{
-				"type":        "boolean",
+				"type": "boolean",
+
 				"description": "Force reinstall if skill already exists (default false)",
 			},
 		},
+
 		"required": []string{"slug", "registry"},
 	}
 }
 
 func (t *InstallSkillTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	// Install lock to prevent concurrent directory operations.
+
 	// Ideally this should be done at a `slug` level, currently, its at a `workspace` level.
+
 	t.mu.Lock()
+
 	defer t.mu.Unlock()
 
 	// Validate slug
+
 	slug, _ := args["slug"].(string)
+
 	if err := utils.ValidateSkillIdentifier(slug); err != nil {
 		return ErrorResult(fmt.Sprintf("invalid slug %q: error: %s", slug, err.Error()))
 	}
 
 	// Validate registry
+
 	registryName, _ := args["registry"].(string)
+
 	if err := utils.ValidateSkillIdentifier(registryName); err != nil {
 		return ErrorResult(fmt.Sprintf("invalid registry %q: error: %s", registryName, err.Error()))
 	}
 
 	version, _ := args["version"].(string)
+
 	force, _ := args["force"].(bool)
 
 	// Check if already installed.
+
 	skillsDir := filepath.Join(t.workspace, "skills")
+
 	targetDir := filepath.Join(skillsDir, slug)
 
 	if !force {
 		if _, err := os.Stat(targetDir); err == nil {
 			return ErrorResult(
+
 				fmt.Sprintf("skill %q already installed at %s. Use force=true to reinstall.", slug, targetDir),
 			)
 		}
 	} else {
 		// Force: remove existing if present.
+
 		os.RemoveAll(targetDir)
 	}
 
 	// Resolve which registry to use.
+
 	registry := t.registryMgr.GetRegistry(registryName)
+
 	if registry == nil {
 		return ErrorResult(fmt.Sprintf("registry %q not found", registryName))
 	}
 
 	// Ensure skills directory exists.
+
 	if err := os.MkdirAll(skillsDir, 0o755); err != nil {
 		return ErrorResult(fmt.Sprintf("failed to create skills directory: %v", err))
 	}
 
 	// Download and install (handles metadata, version resolution, extraction).
+
 	result, err := registry.DownloadAndInstall(ctx, slug, version, targetDir)
 	if err != nil {
 		// Clean up partial install.
+
 		rmErr := os.RemoveAll(targetDir)
+
 		if rmErr != nil {
 			logger.ErrorCF("tool", "Failed to remove partial install",
+
 				map[string]any{
-					"tool":       "install_skill",
+					"tool": "install_skill",
+
 					"target_dir": targetDir,
-					"error":      rmErr.Error(),
+
+					"error": rmErr.Error(),
 				})
 		}
+
 		return ErrorResult(fmt.Sprintf("failed to install %q: %v", slug, err))
 	}
 
 	// Moderation: block malware.
+
 	if result.IsMalwareBlocked {
 		rmErr := os.RemoveAll(targetDir)
+
 		if rmErr != nil {
 			logger.ErrorCF("tool", "Failed to remove partial install",
+
 				map[string]any{
-					"tool":       "install_skill",
+					"tool": "install_skill",
+
 					"target_dir": targetDir,
-					"error":      rmErr.Error(),
+
+					"error": rmErr.Error(),
 				})
 		}
+
 		return ErrorResult(fmt.Sprintf("skill %q is flagged as malicious and cannot be installed", slug))
 	}
 
 	// Write origin metadata.
+
 	if err := writeOriginMeta(targetDir, registry.Name(), slug, result.Version); err != nil {
 		logger.ErrorCF("tool", "Failed to write origin metadata",
+
 			map[string]any{
-				"tool":     "install_skill",
-				"error":    err.Error(),
-				"target":   targetDir,
+				"tool": "install_skill",
+
+				"error": err.Error(),
+
+				"target": targetDir,
+
 				"registry": registry.Name(),
-				"slug":     slug,
-				"version":  result.Version,
+
+				"slug": slug,
+
+				"version": result.Version,
 			})
+
 		_ = err
 	}
 
 	// Build result with moderation warning if suspicious.
+
 	var output string
+
 	if result.IsSuspicious {
 		output = fmt.Sprintf("⚠️ Warning: skill %q is flagged as suspicious (may contain risky patterns).\n\n", slug)
 	}
+
 	output += fmt.Sprintf("Successfully installed skill %q v%s from %s registry.\nLocation: %s\n",
+
 		slug, result.Version, registry.Name(), targetDir)
 
 	if result.Summary != "" {
 		output += fmt.Sprintf("Description: %s\n", result.Summary)
 	}
+
 	output += "\nThe skill is now available and can be loaded in the current session."
 
 	return SilentResult(output)
 }
 
 // originMeta tracks which registry a skill was installed from.
+
 type originMeta struct {
-	Version          int    `json:"version"`
-	Registry         string `json:"registry"`
-	Slug             string `json:"slug"`
+	Version int `json:"version"`
+
+	Registry string `json:"registry"`
+
+	Slug string `json:"slug"`
+
 	InstalledVersion string `json:"installed_version"`
-	InstalledAt      int64  `json:"installed_at"`
+
+	InstalledAt int64 `json:"installed_at"`
 }
 
 func writeOriginMeta(targetDir, registryName, slug, version string) error {
 	meta := originMeta{
-		Version:          1,
-		Registry:         registryName,
-		Slug:             slug,
+		Version: 1,
+
+		Registry: registryName,
+
+		Slug: slug,
+
 		InstalledVersion: version,
-		InstalledAt:      time.Now().UnixMilli(),
+
+		InstalledAt: time.Now().UnixMilli(),
 	}
 
 	data, err := json.MarshalIndent(meta, "", "  ")
@@ -199,5 +268,6 @@ func writeOriginMeta(targetDir, registryName, slug, version string) error {
 	}
 
 	// Use unified atomic write utility with explicit sync for flash storage reliability.
+
 	return fileutil.WriteFileAtomic(filepath.Join(targetDir, ".skill-origin.json"), data, 0o600)
 }

pkg/tools/skills_install_test.go

@@ -14,22 +14,29 @@ import (
 
 func TestInstallSkillToolName(t *testing.T) {
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), t.TempDir())
+
 	assert.Equal(t, "install_skill", tool.Name())
 }
 
 func TestInstallSkillToolMissingSlug(t *testing.T) {
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), t.TempDir())
+
 	result := tool.Execute(context.Background(), map[string]any{})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "identifier is required and must be a non-empty string")
 }
 
 func TestInstallSkillToolEmptySlug(t *testing.T) {
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), t.TempDir())
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"slug": "   ",
 	})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "identifier is required and must be a non-empty string")
 }
 
@@ -38,7 +45,9 @@ func TestInstallSkillToolUnsafeSlug(t *testing.T) {
 
 	cases := []string{
 		"../etc/passwd",
+
 		"path/traversal",
+
 		"path\\traversal",
 	}
 
@@ -46,59 +55,85 @@ func TestInstallSkillToolUnsafeSlug(t *testing.T) {
 		result := tool.Execute(context.Background(), map[string]any{
 			"slug": slug,
 		})
+
 		assert.True(t, result.IsError, "slug %q should be rejected", slug)
+
 		assert.Contains(t, result.ForLLM, "invalid slug")
 	}
 }
 
 func TestInstallSkillToolAlreadyExists(t *testing.T) {
 	workspace := t.TempDir()
+
 	skillDir := filepath.Join(workspace, "skills", "existing-skill")
+
 	require.NoError(t, os.MkdirAll(skillDir, 0o755))
 
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), workspace)
+
 	result := tool.Execute(context.Background(), map[string]any{
-		"slug":     "existing-skill",
+		"slug": "existing-skill",
+
 		"registry": "clawhub",
 	})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "already installed")
 }
 
 func TestInstallSkillToolRegistryNotFound(t *testing.T) {
 	workspace := t.TempDir()
+
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), workspace)
+
 	result := tool.Execute(context.Background(), map[string]any{
-		"slug":     "some-skill",
+		"slug": "some-skill",
+
 		"registry": "nonexistent",
 	})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "registry")
+
 	assert.Contains(t, result.ForLLM, "not found")
 }
 
 func TestInstallSkillToolParameters(t *testing.T) {
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), t.TempDir())
+
 	params := tool.Parameters()
 
 	props, ok := params["properties"].(map[string]any)
+
 	assert.True(t, ok)
+
 	assert.Contains(t, props, "slug")
+
 	assert.Contains(t, props, "version")
+
 	assert.Contains(t, props, "registry")
+
 	assert.Contains(t, props, "force")
 
 	required, ok := params["required"].([]string)
+
 	assert.True(t, ok)
+
 	assert.Contains(t, required, "slug")
+
 	assert.Contains(t, required, "registry")
 }
 
 func TestInstallSkillToolMissingRegistry(t *testing.T) {
 	tool := NewInstallSkillTool(skills.NewRegistryManager(), t.TempDir())
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"slug": "some-skill",
 	})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "invalid registry")
 }

pkg/tools/skills_search.go

@@ -9,18 +9,24 @@ import (
 )
 
 // FindSkillsTool allows the LLM agent to search for installable skills from registries.
+
 type FindSkillsTool struct {
 	registryMgr *skills.RegistryManager
-	cache       *skills.SearchCache
+
+	cache *skills.SearchCache
 }
 
 // NewFindSkillsTool creates a new FindSkillsTool.
+
 // registryMgr is the shared registry manager (built from config in createToolRegistry).
+
 // cache is the search cache for deduplicating similar queries.
+
 func NewFindSkillsTool(registryMgr *skills.RegistryManager, cache *skills.SearchCache) *FindSkillsTool {
 	return &FindSkillsTool{
 		registryMgr: registryMgr,
-		cache:       cache,
+
+		cache: cache,
 	}
 }
 
@@ -35,38 +41,50 @@ func (t *FindSkillsTool) Description() string {
 func (t *FindSkillsTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"query": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Search query describing the desired skill capability (e.g., 'github integration', 'database management')",
 			},
+
 			"limit": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Maximum number of results to return (1-20, default 5)",
-				"minimum":     1.0,
-				"maximum":     20.0,
+
+				"minimum": 1.0,
+
+				"maximum": 20.0,
 			},
 		},
+
 		"required": []string{"query"},
 	}
 }
 
 func (t *FindSkillsTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	query, ok := args["query"].(string)
+
 	query = strings.ToLower(strings.TrimSpace(query))
+
 	if !ok || query == "" {
 		return ErrorResult("query is required and must be a non-empty string")
 	}
 
 	limit := 5
+
 	if l, ok := args["limit"].(float64); ok {
 		li := int(l)
+
 		if li >= 1 && li <= 20 {
 			limit = li
 		}
 	}
 
 	// Check cache first.
+
 	if t.cache != nil {
 		if cached, hit := t.cache.Get(query); hit {
 			return SilentResult(formatSearchResults(query, cached, true))
@@ -74,12 +92,14 @@ func (t *FindSkillsTool) Execute(ctx context.Context, args map[string]any) *Tool
 	}
 
 	// Search all registries.
+
 	results, err := t.registryMgr.SearchAll(ctx, query, limit)
 	if err != nil {
 		return ErrorResult(fmt.Sprintf("skill search failed: %v", err))
 	}
 
 	// Cache the results.
+
 	if t.cache != nil && len(results) > 0 {
 		t.cache.Put(query, results)
 	}
@@ -93,27 +113,36 @@ func formatSearchResults(query string, results []skills.SearchResult, cached boo
 	}
 
 	var sb strings.Builder
+
 	source := ""
+
 	if cached {
 		source = " (cached)"
 	}
+
 	sb.WriteString(fmt.Sprintf("Found %d skills for %q%s:\n\n", len(results), query, source))
 
 	for i, r := range results {
 		sb.WriteString(fmt.Sprintf("%d. **%s**", i+1, r.Slug))
+
 		if r.Version != "" {
 			sb.WriteString(fmt.Sprintf(" v%s", r.Version))
 		}
+
 		sb.WriteString(fmt.Sprintf("  (score: %.3f, registry: %s)\n", r.Score, r.RegistryName))
+
 		if r.DisplayName != "" && r.DisplayName != r.Slug {
 			sb.WriteString(fmt.Sprintf("   Name: %s\n", r.DisplayName))
 		}
+
 		if r.Summary != "" {
 			sb.WriteString(fmt.Sprintf("   %s\n", r.Summary))
 		}
+
 		sb.WriteString("\n")
 	}
 
 	sb.WriteString("Use install_skill with the slug to install a skill.")
+
 	return sb.String()
 }

pkg/tools/skills_search_test.go

@@ -11,80 +11,110 @@ import (
 
 func TestFindSkillsToolName(t *testing.T) {
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), nil)
+
 	assert.Equal(t, "find_skills", tool.Name())
 }
 
 func TestFindSkillsToolMissingQuery(t *testing.T) {
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), nil)
+
 	result := tool.Execute(context.Background(), map[string]any{})
+
 	assert.True(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "query is required")
 }
 
 func TestFindSkillsToolEmptyQuery(t *testing.T) {
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), nil)
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"query": "   ",
 	})
+
 	assert.True(t, result.IsError)
 }
 
 func TestFindSkillsToolCacheHit(t *testing.T) {
 	cache := skills.NewSearchCache(10, 5*60*1000*1000*1000) // 5 min
+
 	cache.Put("github", []skills.SearchResult{
 		{Slug: "github", Score: 0.9, RegistryName: "clawhub"},
 	})
 
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), cache)
+
 	result := tool.Execute(context.Background(), map[string]any{
 		"query": "github",
 	})
 
 	assert.False(t, result.IsError)
+
 	assert.Contains(t, result.ForLLM, "github")
+
 	assert.Contains(t, result.ForLLM, "cached")
 }
 
 func TestFindSkillsToolParameters(t *testing.T) {
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), nil)
+
 	params := tool.Parameters()
 
 	props, ok := params["properties"].(map[string]any)
+
 	assert.True(t, ok)
+
 	assert.Contains(t, props, "query")
+
 	assert.Contains(t, props, "limit")
 
 	required, ok := params["required"].([]string)
+
 	assert.True(t, ok)
+
 	assert.Contains(t, required, "query")
 }
 
 func TestFindSkillsToolDescription(t *testing.T) {
 	tool := NewFindSkillsTool(skills.NewRegistryManager(), nil)
+
 	assert.NotEmpty(t, tool.Description())
+
 	assert.Contains(t, tool.Description(), "skill")
 }
 
 func TestFormatSearchResultsEmpty(t *testing.T) {
 	result := formatSearchResults("test query", nil, false)
+
 	assert.Contains(t, result, "No skills found")
 }
 
 func TestFormatSearchResultsWithData(t *testing.T) {
 	results := []skills.SearchResult{
 		{
-			Slug:         "github",
-			Score:        0.95,
-			DisplayName:  "GitHub",
-			Summary:      "GitHub API integration",
-			Version:      "1.0.0",
+			Slug: "github",
+
+			Score: 0.95,
+
+			DisplayName: "GitHub",
+
+			Summary: "GitHub API integration",
+
+			Version: "1.0.0",
+
 			RegistryName: "clawhub",
 		},
 	}
+
 	output := formatSearchResults("github", results, false)
+
 	assert.Contains(t, output, "github")
+
 	assert.Contains(t, output, "v1.0.0")
+
 	assert.Contains(t, output, "0.950")
+
 	assert.Contains(t, output, "clawhub")
+
 	assert.Contains(t, output, "install_skill")
 }

pkg/tools/spawn.go

@@ -7,22 +7,29 @@ import (
 )
 
 type SpawnTool struct {
-	manager        *SubagentManager
-	originChannel  string
-	originChatID   string
+	manager *SubagentManager
+
+	originChannel string
+
+	originChatID string
+
 	allowlistCheck func(targetAgentID string) bool
-	callback       AsyncCallback // For async completion notification
+
+	callback AsyncCallback // For async completion notification
 }
 
 func NewSpawnTool(manager *SubagentManager) *SpawnTool {
 	return &SpawnTool{
-		manager:       manager,
+		manager: manager,
+
 		originChannel: "cli",
-		originChatID:  "direct",
+
+		originChatID: "direct",
 	}
 }
 
 // SetCallback implements AsyncTool interface for async completion notification
+
 func (t *SpawnTool) SetCallback(cb AsyncCallback) {
 	t.callback = cb
 }
@@ -38,31 +45,42 @@ func (t *SpawnTool) Description() string {
 func (t *SpawnTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"task": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "The task for subagent to complete",
 			},
+
 			"label": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Optional short label for the task (for display)",
 			},
+
 			"agent_id": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Optional target agent ID to delegate the task to",
 			},
+
 			"preset": map[string]any{
-				"type":        "string",
-				"enum":        []string{"scout", "analyst", "coder", "worker", "coordinator"},
+				"type": "string",
+
+				"enum": []string{"scout", "analyst", "coder", "worker", "coordinator"},
+
 				"description": "Optional capability tier: scout (explore), analyst (analyze), coder (code), worker (build), coordinator (orchestrate)",
 			},
 		},
+
 		"required": []string{"task"},
 	}
 }
 
 func (t *SpawnTool) SetContext(channel, chatID string) {
 	t.originChannel = channel
+
 	t.originChatID = chatID
 }
 
@@ -72,20 +90,28 @@ func (t *SpawnTool) SetAllowlistChecker(check func(targetAgentID string) bool) {
 
 func (t *SpawnTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	task, ok := args["task"].(string)
+
 	if !ok || strings.TrimSpace(task) == "" {
 		return ErrorResult(
+
 			`Required parameter "task" (string) is missing. ` +
+
 				`Example: {"task": "describe what you need done", "preset": "scout"}`,
 		)
 	}
 
 	label, _ := args["label"].(string)
+
 	agentID, _ := args["agent_id"].(string)
+
 	preset, _ := args["preset"].(string)
 
 	// Check allowlist if targeting a specific agent ID.
+
 	// Presets (scout, analyst, etc.) are NOT agent IDs — they are validated
+
 	// separately by IsValidPreset() in the subagent manager.
+
 	if agentID != "" && t.allowlistCheck != nil {
 		if !t.allowlistCheck(agentID) {
 			return ErrorResult(fmt.Sprintf("agent %q is not in the allowed agents list", agentID))
@@ -93,9 +119,12 @@ func (t *SpawnTool) Execute(ctx context.Context, args map[string]any) *ToolResul
 	}
 
 	// Validate preset name if provided
+
 	if preset != "" && !IsValidPreset(Preset(preset)) {
 		return ErrorResult(fmt.Sprintf(
+
 			"preset %q is not valid. Available presets: scout, analyst, coder, worker, coordinator",
+
 			preset,
 		))
 	}
@@ -105,11 +134,13 @@ func (t *SpawnTool) Execute(ctx context.Context, args map[string]any) *ToolResul
 	}
 
 	// Pass callback to manager for async completion notification
+
 	result, err := t.manager.Spawn(ctx, task, label, agentID, t.originChannel, t.originChatID, preset, t.callback)
 	if err != nil {
 		return ErrorResult(fmt.Sprintf("failed to spawn subagent: %v", err))
 	}
 
 	// Return AsyncResult since the task runs in background
+
 	return AsyncResult(result)
 }

pkg/tools/spawn_test.go

@@ -8,31 +8,41 @@ import (
 
 func TestSpawnTool_Execute_EmptyTask(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, nil, WebSearchToolOptions{})
+
 	tool := NewSpawnTool(manager)
 
 	ctx := context.Background()
 
 	tests := []struct {
 		name string
+
 		args map[string]any
 	}{
 		{"empty string", map[string]any{"task": ""}},
+
 		{"whitespace only", map[string]any{"task": "   "}},
+
 		{"tabs and newlines", map[string]any{"task": "\t\n  "}},
+
 		{"missing task key", map[string]any{"label": "test"}},
+
 		{"wrong type", map[string]any{"task": 123}},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			result := tool.Execute(ctx, tt.args)
+
 			if result == nil {
 				t.Fatal("Result should not be nil")
 			}
+
 			if !result.IsError {
 				t.Error("Expected error for invalid task parameter")
 			}
+
 			if !strings.Contains(result.ForLLM, `"task"`) {
 				t.Errorf("Error message should mention '\"task\"', got: %s", result.ForLLM)
 			}
@@ -42,22 +52,29 @@ func TestSpawnTool_Execute_EmptyTask(t *testing.T) {
 
 func TestSpawnTool_Execute_ValidTask(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, nil, WebSearchToolOptions{})
+
 	tool := NewSpawnTool(manager)
 
 	ctx := context.Background()
+
 	args := map[string]any{
-		"task":  "Write a haiku about coding",
+		"task": "Write a haiku about coding",
+
 		"label": "haiku-task",
 	}
 
 	result := tool.Execute(ctx, args)
+
 	if result == nil {
 		t.Fatal("Result should not be nil")
 	}
+
 	if result.IsError {
 		t.Errorf("Expected success for valid task, got error: %s", result.ForLLM)
 	}
+
 	if !result.Async {
 		t.Error("SpawnTool should return async result")
 	}
@@ -67,12 +84,15 @@ func TestSpawnTool_Execute_NilManager(t *testing.T) {
 	tool := NewSpawnTool(nil)
 
 	ctx := context.Background()
+
 	args := map[string]any{"task": "test task"}
 
 	result := tool.Execute(ctx, args)
+
 	if !result.IsError {
 		t.Error("Expected error for nil manager")
 	}
+
 	if !strings.Contains(result.ForLLM, "spawn tool is not available") {
 		t.Errorf("Error message should mention spawn tool not available, got: %s", result.ForLLM)
 	}

pkg/tools/spi.go

@@ -10,6 +10,7 @@ import (
 )
 
 // SPITool provides SPI bus interaction for high-speed peripheral communication.
+
 type SPITool struct{}
 
 func NewSPITool() *SPITool {
@@ -27,42 +28,61 @@ func (t *SPITool) Description() string {
 func (t *SPITool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"action": map[string]any{
-				"type":        "string",
-				"enum":        []string{"list", "transfer", "read"},
+				"type": "string",
+
+				"enum": []string{"list", "transfer", "read"},
+
 				"description": "Action to perform: list (find available SPI devices), transfer (full-duplex send/receive), read (receive bytes by sending zeros)",
 			},
+
 			"device": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "SPI device identifier (e.g. \"2.0\" for /dev/spidev2.0). Required for transfer/read.",
 			},
+
 			"speed": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "SPI clock speed in Hz. Default: 1000000 (1 MHz).",
 			},
+
 			"mode": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "SPI mode (0-3). Default: 0. Mode sets CPOL and CPHA: 0=0,0 1=0,1 2=1,0 3=1,1.",
 			},
+
 			"bits": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Bits per word. Default: 8.",
 			},
+
 			"data": map[string]any{
-				"type":        "array",
-				"items":       map[string]any{"type": "integer"},
+				"type": "array",
+
+				"items": map[string]any{"type": "integer"},
+
 				"description": "Bytes to send (0-255 each). Required for transfer action.",
 			},
+
 			"length": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Number of bytes to read (1-4096). Required for read action.",
 			},
+
 			"confirm": map[string]any{
-				"type":        "boolean",
+				"type": "boolean",
+
 				"description": "Must be true for transfer operations. Safety guard to prevent accidental writes.",
 			},
 		},
+
 		"required": []string{"action"},
 	}
 }
@@ -73,23 +93,32 @@ func (t *SPITool) Execute(ctx context.Context, args map[string]any) *ToolResult
 	}
 
 	action, ok := args["action"].(string)
+
 	if !ok {
 		return ErrorResult("action is required")
 	}
 
 	switch action {
 	case "list":
+
 		return t.list()
+
 	case "transfer":
+
 		return t.transfer(args)
+
 	case "read":
+
 		return t.readDevice(args)
+
 	default:
+
 		return ErrorResult(fmt.Sprintf("unknown action: %s (valid: list, transfer, read)", action))
 	}
 }
 
 // list finds available SPI devices by globbing /dev/spidev*
+
 func (t *SPITool) list() *ToolResult {
 	matches, err := filepath.Glob("/dev/spidev*")
 	if err != nil {
@@ -103,12 +132,15 @@ func (t *SPITool) list() *ToolResult {
 	}
 
 	type devInfo struct {
-		Path   string `json:"path"`
+		Path string `json:"path"`
+
 		Device string `json:"device"`
 	}
 
 	devices := make([]devInfo, 0, len(matches))
+
 	re := regexp.MustCompile(`/dev/spidev(\d+\.\d+)`)
+
 	for _, m := range matches {
 		if sub := re.FindStringSubmatch(m); sub != nil {
 			devices = append(devices, devInfo{Path: m, Device: sub[1]})
@@ -116,45 +148,58 @@ func (t *SPITool) list() *ToolResult {
 	}
 
 	result, _ := json.MarshalIndent(devices, "", "  ")
+
 	return SilentResult(fmt.Sprintf("Found %d SPI device(s):\n%s", len(devices), string(result)))
 }
 
 // Helper function for SPI operations (used by platform-specific implementations)
 
 // parseSPIArgs extracts and validates common SPI parameters
+
 //
+
 //nolint:unused // Used by spi_linux.go
+
 func parseSPIArgs(args map[string]any) (device string, speed uint32, mode uint8, bits uint8, errMsg string) {
 	dev, ok := args["device"].(string)
+
 	if !ok || dev == "" {
 		return "", 0, 0, 0, "device is required (e.g. \"2.0\" for /dev/spidev2.0)"
 	}
+
 	matched, _ := regexp.MatchString(`^\d+\.\d+$`, dev)
+
 	if !matched {
 		return "", 0, 0, 0, "invalid device identifier: must be in format \"X.Y\" (e.g. \"2.0\")"
 	}
 
 	speed = 1000000 // default 1 MHz
+
 	if s, ok := args["speed"].(float64); ok {
 		if s < 1 || s > 125000000 {
 			return "", 0, 0, 0, "speed must be between 1 Hz and 125 MHz"
 		}
+
 		speed = uint32(s)
 	}
 
 	mode = 0
+
 	if m, ok := args["mode"].(float64); ok {
 		if int(m) < 0 || int(m) > 3 {
 			return "", 0, 0, 0, "mode must be 0-3"
 		}
+
 		mode = uint8(m)
 	}
 
 	bits = 8
+
 	if b, ok := args["bits"].(float64); ok {
 		if int(b) < 1 || int(b) > 32 {
 			return "", 0, 0, 0, "bits must be between 1 and 32"
 		}
+
 		bits = uint8(b)
 	}
 

pkg/tools/spi_linux.go

@@ -9,190 +9,321 @@ import (
 )
 
 // SPI ioctl constants from Linux kernel headers.
+
 // Calculated from _IOW('k', nr, size) macro:
+
 //
+
 //	direction(1)<<30 | size<<16 | type(0x6B)<<8 | nr
+
 const (
-	spiIocWrMode        = 0x40016B01 // _IOW('k', 1, __u8)
+	spiIocWrMode = 0x40016B01 // _IOW('k', 1, __u8)
+
 	spiIocWrBitsPerWord = 0x40016B03 // _IOW('k', 3, __u8)
-	spiIocWrMaxSpeedHz  = 0x40046B04 // _IOW('k', 4, __u32)
-	spiIocMessage1      = 0x40206B00 // _IOW('k', 0, struct spi_ioc_transfer) — 32 bytes
+
+	spiIocWrMaxSpeedHz = 0x40046B04 // _IOW('k', 4, __u32)
+
+	spiIocMessage1 = 0x40206B00 // _IOW('k', 0, struct spi_ioc_transfer) — 32 bytes
+
 )
 
 // spiTransfer matches Linux kernel struct spi_ioc_transfer (32 bytes on all architectures).
+
 type spiTransfer struct {
-	txBuf       uint64
-	rxBuf       uint64
-	length      uint32
-	speedHz     uint32
-	delayUsecs  uint16
+	txBuf uint64
+
+	rxBuf uint64
+
+	length uint32
+
+	speedHz uint32
+
+	delayUsecs uint16
+
 	bitsPerWord uint8
-	csChange    uint8
-	txNbits     uint8
-	rxNbits     uint8
-	wordDelay   uint8
-	pad         uint8
+
+	csChange uint8
+
+	txNbits uint8
+
+	rxNbits uint8
+
+	wordDelay uint8
+
+	pad uint8
 }
 
 // configureSPI opens an SPI device and sets mode, bits per word, and speed
+
 func configureSPI(devPath string, mode uint8, bits uint8, speed uint32) (int, *ToolResult) {
+
 	fd, err := syscall.Open(devPath, syscall.O_RDWR, 0)
+
 	if err != nil {
+
 		return -1, ErrorResult(fmt.Sprintf("failed to open %s: %v (check permissions and spidev module)", devPath, err))
+
 	}
 
 	// Set SPI mode
+
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), spiIocWrMode, uintptr(unsafe.Pointer(&mode)))
+
 	if errno != 0 {
+
 		syscall.Close(fd)
+
 		return -1, ErrorResult(fmt.Sprintf("failed to set SPI mode %d: %v", mode, errno))
+
 	}
 
 	// Set bits per word
+
 	_, _, errno = syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), spiIocWrBitsPerWord, uintptr(unsafe.Pointer(&bits)))
+
 	if errno != 0 {
+
 		syscall.Close(fd)
+
 		return -1, ErrorResult(fmt.Sprintf("failed to set bits per word %d: %v", bits, errno))
+
 	}
 
 	// Set max speed
+
 	_, _, errno = syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), spiIocWrMaxSpeedHz, uintptr(unsafe.Pointer(&speed)))
+
 	if errno != 0 {
+
 		syscall.Close(fd)
+
 		return -1, ErrorResult(fmt.Sprintf("failed to set SPI speed %d Hz: %v", speed, errno))
+
 	}
 
 	return fd, nil
+
 }
 
 // transfer performs a full-duplex SPI transfer
+
 func (t *SPITool) transfer(args map[string]any) *ToolResult {
+
 	confirm, _ := args["confirm"].(bool)
+
 	if !confirm {
+
 		return ErrorResult(
+
 			"transfer operations require confirm: true. Please confirm with the user before sending data to SPI devices.",
 		)
+
 	}
 
 	dev, speed, mode, bits, errMsg := parseSPIArgs(args)
+
 	if errMsg != "" {
+
 		return ErrorResult(errMsg)
+
 	}
 
 	dataRaw, ok := args["data"].([]any)
+
 	if !ok || len(dataRaw) == 0 {
+
 		return ErrorResult("data is required for transfer (array of byte values 0-255)")
+
 	}
+
 	if len(dataRaw) > 4096 {
+
 		return ErrorResult("data too long: maximum 4096 bytes per SPI transfer")
+
 	}
 
 	txBuf := make([]byte, len(dataRaw))
+
 	for i, v := range dataRaw {
+
 		f, ok := v.(float64)
+
 		if !ok {
+
 			return ErrorResult(fmt.Sprintf("data[%d] is not a valid byte value", i))
+
 		}
+
 		b := int(f)
+
 		if b < 0 || b > 255 {
+
 			return ErrorResult(fmt.Sprintf("data[%d] = %d is out of byte range (0-255)", i, b))
+
 		}
+
 		txBuf[i] = byte(b)
+
 	}
 
 	devPath := fmt.Sprintf("/dev/spidev%s", dev)
+
 	fd, errResult := configureSPI(devPath, mode, bits, speed)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
+
 	defer syscall.Close(fd)
 
 	rxBuf := make([]byte, len(txBuf))
 
 	xfer := spiTransfer{
-		txBuf:       uint64(uintptr(unsafe.Pointer(&txBuf[0]))),
-		rxBuf:       uint64(uintptr(unsafe.Pointer(&rxBuf[0]))),
-		length:      uint32(len(txBuf)),
-		speedHz:     speed,
+
+		txBuf: uint64(uintptr(unsafe.Pointer(&txBuf[0]))),
+
+		rxBuf: uint64(uintptr(unsafe.Pointer(&rxBuf[0]))),
+
+		length: uint32(len(txBuf)),
+
+		speedHz: speed,
+
 		bitsPerWord: bits,
 	}
 
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), spiIocMessage1, uintptr(unsafe.Pointer(&xfer)))
+
 	runtime.KeepAlive(txBuf)
+
 	runtime.KeepAlive(rxBuf)
+
 	if errno != 0 {
+
 		return ErrorResult(fmt.Sprintf("SPI transfer failed: %v", errno))
+
 	}
 
 	// Format received bytes
+
 	hexBytes := make([]string, len(rxBuf))
+
 	intBytes := make([]int, len(rxBuf))
+
 	for i, b := range rxBuf {
+
 		hexBytes[i] = fmt.Sprintf("0x%02x", b)
+
 		intBytes[i] = int(b)
+
 	}
 
 	result, _ := json.MarshalIndent(map[string]any{
-		"device":   devPath,
-		"sent":     len(txBuf),
+
+		"device": devPath,
+
+		"sent": len(txBuf),
+
 		"received": intBytes,
-		"hex":      hexBytes,
+
+		"hex": hexBytes,
 	}, "", "  ")
+
 	return SilentResult(string(result))
+
 }
 
 // readDevice reads bytes from SPI by sending zeros (read-only, no confirm needed)
+
 func (t *SPITool) readDevice(args map[string]any) *ToolResult {
+
 	dev, speed, mode, bits, errMsg := parseSPIArgs(args)
+
 	if errMsg != "" {
+
 		return ErrorResult(errMsg)
+
 	}
 
 	length := 0
+
 	if l, ok := args["length"].(float64); ok {
+
 		length = int(l)
+
 	}
+
 	if length < 1 || length > 4096 {
+
 		return ErrorResult("length is required for read (1-4096)")
+
 	}
 
 	devPath := fmt.Sprintf("/dev/spidev%s", dev)
+
 	fd, errResult := configureSPI(devPath, mode, bits, speed)
+
 	if errResult != nil {
+
 		return errResult
+
 	}
+
 	defer syscall.Close(fd)
 
 	txBuf := make([]byte, length) // zeros
+
 	rxBuf := make([]byte, length)
 
 	xfer := spiTransfer{
-		txBuf:       uint64(uintptr(unsafe.Pointer(&txBuf[0]))),
-		rxBuf:       uint64(uintptr(unsafe.Pointer(&rxBuf[0]))),
-		length:      uint32(length),
-		speedHz:     speed,
+
+		txBuf: uint64(uintptr(unsafe.Pointer(&txBuf[0]))),
+
+		rxBuf: uint64(uintptr(unsafe.Pointer(&rxBuf[0]))),
+
+		length: uint32(length),
+
+		speedHz: speed,
+
 		bitsPerWord: bits,
 	}
 
 	_, _, errno := syscall.Syscall(syscall.SYS_IOCTL, uintptr(fd), spiIocMessage1, uintptr(unsafe.Pointer(&xfer)))
+
 	runtime.KeepAlive(txBuf)
+
 	runtime.KeepAlive(rxBuf)
+
 	if errno != 0 {
+
 		return ErrorResult(fmt.Sprintf("SPI read failed: %v", errno))
+
 	}
 
 	hexBytes := make([]string, len(rxBuf))
+
 	intBytes := make([]int, len(rxBuf))
+
 	for i, b := range rxBuf {
+
 		hexBytes[i] = fmt.Sprintf("0x%02x", b)
+
 		intBytes[i] = int(b)
+
 	}
 
 	result, _ := json.MarshalIndent(map[string]any{
+
 		"device": devPath,
-		"bytes":  intBytes,
-		"hex":    hexBytes,
+
+		"bytes": intBytes,
+
+		"hex": hexBytes,
+
 		"length": len(rxBuf),
 	}, "", "  ")
+
 	return SilentResult(string(result))
+
 }

pkg/tools/subagent_container_test.go

@@ -0,0 +1,125 @@
+package tools
+
+import (
+	"testing"
+)
+
+func TestIsDeliberatePreset(t *testing.T) {
+	tests := []struct {
+		preset Preset
+		want   bool
+	}{
+		{PresetScout, false},
+		{PresetAnalyst, false},
+		{PresetCoder, true},
+		{PresetWorker, true},
+		{PresetCoordinator, true},
+		{"unknown", false},
+	}
+	for _, tt := range tests {
+		got := isDeliberatePreset(tt.preset)
+		if got != tt.want {
+			t.Errorf("isDeliberatePreset(%q) = %v, want %v", tt.preset, got, tt.want)
+		}
+	}
+}
+
+func TestContainerMessageChannels(t *testing.T) {
+	// Simulate channel creation for a deliberate preset task.
+	task := &SubagentTask{
+		ID:    "subagent-1",
+		inCh:  make(chan string, 1),
+		outCh: make(chan ContainerMessage, 4),
+	}
+
+	// Subagent sends a question.
+	task.outCh <- ContainerMessage{
+		Type:    "question",
+		Content: "Which DB schema?",
+		TaskID:  task.ID,
+	}
+
+	// Drain pending messages.
+	var msgs []ContainerMessage
+	for {
+		select {
+		case msg := <-task.outCh:
+			msgs = append(msgs, msg)
+		default:
+			goto done
+		}
+	}
+done:
+	if len(msgs) != 1 {
+		t.Fatalf("msgs count = %d, want 1", len(msgs))
+	}
+	if msgs[0].Type != "question" {
+		t.Errorf("Type = %q, want %q", msgs[0].Type, "question")
+	}
+	if msgs[0].Content != "Which DB schema?" {
+		t.Errorf("Content = %q, want %q", msgs[0].Content, "Which DB schema?")
+	}
+
+	// Conductor answers.
+	task.inCh <- "Use PostgreSQL"
+	answer := <-task.inCh
+	if answer != "Use PostgreSQL" {
+		t.Errorf("answer = %q, want %q", answer, "Use PostgreSQL")
+	}
+}
+
+func TestPendingQuestionsAndAnswerQuestion(t *testing.T) {
+	mgr := &SubagentManager{
+		tasks: map[string]*SubagentTask{
+			"subagent-1": {
+				ID:    "subagent-1",
+				outCh: make(chan ContainerMessage, 4),
+				inCh:  make(chan string, 1),
+			},
+			"subagent-2": {
+				ID: "subagent-2",
+				// No channels — exploratory preset.
+			},
+		},
+	}
+
+	// Send question from subagent-1.
+	mgr.tasks["subagent-1"].outCh <- ContainerMessage{
+		Type:    "question",
+		Content: "What port?",
+		TaskID:  "subagent-1",
+	}
+
+	msgs := mgr.PendingQuestions()
+	if len(msgs) != 1 {
+		t.Fatalf("pending count = %d, want 1", len(msgs))
+	}
+	if msgs[0].TaskID != "subagent-1" {
+		t.Errorf("TaskID = %q, want %q", msgs[0].TaskID, "subagent-1")
+	}
+
+	// Second call should return empty (already drained).
+	msgs2 := mgr.PendingQuestions()
+	if len(msgs2) != 0 {
+		t.Errorf("second pending count = %d, want 0", len(msgs2))
+	}
+
+	// Answer the question.
+	if err := mgr.AnswerQuestion("subagent-1", "8080"); err != nil {
+		t.Fatalf("AnswerQuestion: %v", err)
+	}
+	answer := <-mgr.tasks["subagent-1"].inCh
+	if answer != "8080" {
+		t.Errorf("answer = %q, want %q", answer, "8080")
+	}
+
+	// Answer non-existent task.
+	if err := mgr.AnswerQuestion("subagent-99", "x"); err == nil {
+		t.Error("expected error for non-existent task")
+	}
+
+	// Answer task without channels.
+	if err := mgr.AnswerQuestion("subagent-2", "x"); err == nil {
+		t.Error("expected error for task without escalation channel")
+	}
+}

pkg/tools/subagent_env_test.go

@@ -0,0 +1,192 @@
+package tools
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestExtractPlanContext(t *testing.T) {
+	tmpDir := t.TempDir()
+
+	memDir := filepath.Join(tmpDir, "memory")
+
+	if err := os.MkdirAll(memDir, 0o755); err != nil {
+		t.Fatal(err)
+	}
+
+	memContent := `# Active Plan
+
+> Task: Implement authentication
+
+> Status: executing
+
+> Phase: 1
+
+
+
+## Context
+
+The project uses JWT tokens for auth.
+
+Database is PostgreSQL.
+
+
+
+## Phase 1: Setup
+
+- [x] Add middleware
+
+- [ ] Add JWT validation
+
+
+
+## Commands
+
+build: go build ./...
+
+test: go test ./...
+
+
+
+## Orchestration
+
+### Delegated
+
+- auth-scout: investigate patterns
+
+### Findings
+
+- Found existing middleware in pkg/auth
+
+`
+
+	if err := os.WriteFile(filepath.Join(memDir, "MEMORY.md"), []byte(memContent), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	ctx := extractPlanContext(tmpDir)
+
+	if ctx == "" {
+		t.Fatal("extractPlanContext returned empty")
+	}
+
+	// Should contain the task line.
+
+	if !strings.Contains(ctx, "> Task: Implement authentication") {
+		t.Error("missing task line")
+	}
+
+	// Should contain Context section.
+
+	if !strings.Contains(ctx, "JWT tokens") {
+		t.Error("missing Context section content")
+	}
+
+	// Should contain Commands section.
+
+	if !strings.Contains(ctx, "go build") {
+		t.Error("missing Commands section content")
+	}
+
+	// Should contain Orchestration section.
+
+	if !strings.Contains(ctx, "auth-scout") {
+		t.Error("missing Orchestration section content")
+	}
+}
+
+func TestExtractPlanContext_NoFile(t *testing.T) {
+	ctx := extractPlanContext(t.TempDir())
+
+	if ctx != "" {
+		t.Errorf("expected empty for missing MEMORY.md, got %q", ctx)
+	}
+}
+
+func TestExtractSection(t *testing.T) {
+	content := `## Context
+
+Some context here.
+
+
+
+## Commands
+
+build: go build
+
+
+
+## Other
+
+stuff`
+
+	section := extractSection(content, "## Context")
+
+	if !strings.Contains(section, "Some context here.") {
+		t.Errorf("Context section = %q, missing content", section)
+	}
+
+	if strings.Contains(section, "## Commands") {
+		t.Errorf("Context section leaked into next section")
+	}
+
+	section = extractSection(content, "## Commands")
+
+	if !strings.Contains(section, "go build") {
+		t.Errorf("Commands section = %q, missing content", section)
+	}
+
+	section = extractSection(content, "## Nonexistent")
+
+	if section != "" {
+		t.Errorf("expected empty for nonexistent section, got %q", section)
+	}
+}
+
+func TestBuildSubagentSystemPrompt(t *testing.T) {
+	// With no workspace/MEMORY.md, should return base prompt unchanged.
+
+	base := "You are a subagent."
+
+	got := buildSubagentSystemPrompt(base, t.TempDir())
+
+	if got != base {
+		t.Errorf("expected base prompt unchanged, got %q", got)
+	}
+
+	// With MEMORY.md, should append environment context.
+
+	tmpDir := t.TempDir()
+
+	memDir := filepath.Join(tmpDir, "memory")
+
+	os.MkdirAll(memDir, 0o755)
+
+	os.WriteFile(filepath.Join(memDir, "MEMORY.md"), []byte(`# Plan
+
+> Task: Test task
+
+
+
+## Context
+
+Test context info.
+
+`), 0o644)
+
+	got = buildSubagentSystemPrompt(base, tmpDir)
+
+	if !strings.Contains(got, base) {
+		t.Error("result should contain base prompt")
+	}
+
+	if !strings.Contains(got, "Environment Context") {
+		t.Error("result should contain Environment Context header")
+	}
+
+	if !strings.Contains(got, "Test context info") {
+		t.Error("result should contain MEMORY.md context")
+	}
+}

pkg/tools/subagent_plan_test.go

@@ -0,0 +1,98 @@
+package tools
+
+import (
+	"testing"
+)
+
+func TestSubagentPlanStateString(t *testing.T) {
+	tests := []struct {
+		state SubagentPlanState
+		want  string
+	}{
+		{PlanNone, "none"},
+		{PlanClarifying, "clarifying"},
+		{PlanReview, "review"},
+		{PlanExecuting, "executing"},
+		{PlanCompleted, "completed"},
+	}
+	for _, tt := range tests {
+		got := tt.state.String()
+		if got != tt.want {
+			t.Errorf("SubagentPlanState(%d).String() = %q, want %q", tt.state, got, tt.want)
+		}
+	}
+}
+
+func TestFormatPlanSteps(t *testing.T) {
+	steps := []string{"Read config", "Add middleware", "Write tests"}
+	got := formatPlanSteps(steps)
+	want := "1. Read config\n2. Add middleware\n3. Write tests\n"
+	if got != want {
+		t.Errorf("formatPlanSteps = %q, want %q", got, want)
+	}
+}
+
+func TestClarifyingSystemPrompt(t *testing.T) {
+	prompt := clarifyingSystemPrompt()
+	if prompt == "" {
+		t.Error("clarifyingSystemPrompt returned empty string")
+	}
+	// Should mention ask_conductor and submit_plan.
+	for _, keyword := range []string{"ask_conductor", "submit_plan", "CLARIFYING"} {
+		if !containsString(prompt, keyword) {
+			t.Errorf("clarifyingSystemPrompt missing keyword %q", keyword)
+		}
+	}
+}
+
+func TestExecutingSystemPrompt(t *testing.T) {
+	prompt := executingSystemPrompt()
+	if prompt == "" {
+		t.Error("executingSystemPrompt returned empty string")
+	}
+	if !containsString(prompt, "EXECUTING") {
+		t.Error("executingSystemPrompt missing keyword EXECUTING")
+	}
+}
+
+func TestExploratorySystemPrompt(t *testing.T) {
+	scoutPrompt := exploratorySystemPrompt(PresetScout)
+	if scoutPrompt == "" {
+		t.Error("exploratorySystemPrompt(scout) returned empty")
+	}
+	defaultPrompt := exploratorySystemPrompt("unknown")
+	if defaultPrompt == "" {
+		t.Error("exploratorySystemPrompt(unknown) returned empty")
+	}
+	if scoutPrompt == defaultPrompt {
+		t.Error("scout and unknown prompts should differ")
+	}
+}
+
+func TestDeliberateTaskChannelsCreated(t *testing.T) {
+	// Verify that channels and initial state are correct for deliberate presets.
+	task := &SubagentTask{
+		inCh:  make(chan string, 1),
+		outCh: make(chan ContainerMessage, 4),
+	}
+	if task.inCh == nil || task.outCh == nil {
+		t.Fatal("expected channels to be non-nil for deliberate task")
+	}
+	if task.PlanState != PlanNone {
+		t.Errorf("initial PlanState = %v, want PlanNone", task.PlanState)
+	}
+}
+
+// containsString checks if s contains substr.
+func containsString(s, substr string) bool {
+	return len(s) >= len(substr) && (s == substr || len(s) > 0 && containsSubstr(s, substr))
+}
+
+func containsSubstr(s, sub string) bool {
+	for i := 0; i <= len(s)-len(sub); i++ {
+		if s[i:i+len(sub)] == sub {
+			return true
+		}
+	}
+	return false
+}

pkg/tools/subagent_reporter_test.go

@@ -11,8 +11,11 @@ import (
 )
 
 // blockingProvider blocks inside Chat until the context is canceled.
+
 // The ready channel is closed the moment Chat is entered, so callers can
+
 // synchronize before canceling the context.
+
 type blockingProvider struct {
 	ready chan struct{}
 }
@@ -23,42 +26,63 @@ func newBlockingProvider() *blockingProvider {
 
 func (p *blockingProvider) Chat(
 	ctx context.Context,
+
 	_ []providers.Message,
+
 	_ []providers.ToolDefinition,
+
 	_ string,
+
 	_ map[string]any,
 ) (*providers.LLMResponse, error) {
 	close(p.ready) // signal: we are now blocking
+
 	<-ctx.Done()
+
 	return nil, ctx.Err()
 }
 
 func (p *blockingProvider) GetDefaultModel() string { return "test" }
 
 // TestSubagentManager_Spawn_EmitsLifecycleEvents verifies that Spawn() fires
+
 // the correct sequence of orchestration events through a real Broadcaster:
+
 //
+
 //	agent_spawn → conversation(conductor→sub) → agent_state(waiting) →
+
 //	conversation(sub→conductor) → agent_gc(completed)
+
 //
+
 // It also verifies that the snapshot is empty after ReportGC and that the
+
 // completion callback is invoked.
+
 func TestSubagentManager_Spawn_EmitsLifecycleEvents(t *testing.T) {
 	b := orch.NewBroadcaster()
+
 	sub := b.Subscribe()
+
 	defer b.Unsubscribe(sub)
 
 	provider := &MockLLMProvider{}
+
 	mgr := NewSubagentManager(provider, "test-model", "/tmp/test", nil, b, WebSearchToolOptions{})
 
 	var callbackCalled int32
+
 	cb := AsyncCallback(func(_ context.Context, _ *ToolResult) {
 		atomic.StoreInt32(&callbackCalled, 1)
 	})
 
 	_, err := mgr.Spawn(
+
 		context.Background(),
+
 		"say hello", "hello-task", "", "cli", "direct", "",
+
 		cb,
 	)
 	if err != nil {
@@ -66,94 +90,131 @@ func TestSubagentManager_Spawn_EmitsLifecycleEvents(t *testing.T) {
 	}
 
 	// Collect events until agent_gc or timeout.
+
 	var events []orch.Event
+
 	deadline := time.After(3 * time.Second)
+
 loop:
+
 	for {
 		select {
 		case ev := <-sub.Ch:
+
 			events = append(events, ev)
+
 			if ev.Type == "agent_gc" {
 				break loop
 			}
+
 		case <-deadline:
+
 			t.Fatalf("timed out waiting for agent_gc; events so far: %+v", events)
 		}
 	}
 
 	// 1. First event must be agent_spawn with the correct label.
+
 	if len(events) == 0 || events[0].Type != "agent_spawn" {
 		t.Fatalf("first event must be agent_spawn, got: %+v", events)
 	}
+
 	if events[0].Label != "hello-task" {
 		t.Errorf("agent_spawn label = %q, want %q", events[0].Label, "hello-task")
 	}
+
 	spawnedID := events[0].ID
 
 	// 2. There must be a conversation from conductor → subagent.
+
 	var hasConvToSub bool
+
 	for _, ev := range events {
 		if ev.Type == "conversation" && ev.From == "conductor" && ev.To == spawnedID {
 			hasConvToSub = true
+
 			break
 		}
 	}
+
 	if !hasConvToSub {
 		t.Errorf("missing conversation(conductor → %s); events: %+v", spawnedID, events)
 	}
 
 	// 3. There must be at least one agent_state(waiting) for the subagent.
+
 	var hasWaiting bool
+
 	for _, ev := range events {
 		if ev.Type == "agent_state" && ev.ID == spawnedID && ev.State == "waiting" {
 			hasWaiting = true
+
 			break
 		}
 	}
+
 	if !hasWaiting {
 		t.Errorf("missing agent_state(waiting) for %s; events: %+v", spawnedID, events)
 	}
 
 	// 4. Last event must be agent_gc with reason "completed".
+
 	last := events[len(events)-1]
+
 	if last.Type != "agent_gc" || last.ID != spawnedID || last.Reason != "completed" {
 		t.Errorf("last event must be agent_gc(completed), got: %+v", last)
 	}
 
 	// 5. Snapshot must be empty after GC (agent removed from live map).
+
 	if snap := b.Snapshot(); len(snap) != 0 {
 		t.Errorf("snapshot must be empty after agent_gc, got: %v", snap)
 	}
 
 	// 6. Callback must be called. The callback fires in the same goroutine
+
 	// as ReportGC (after the deferred unlock), so we poll briefly.
+
 	for i := 0; i < 100; i++ {
 		if atomic.LoadInt32(&callbackCalled) == 1 {
 			break
 		}
+
 		time.Sleep(10 * time.Millisecond)
 	}
+
 	if atomic.LoadInt32(&callbackCalled) != 1 {
 		t.Error("completion callback was not called after agent_gc")
 	}
 }
 
 // TestSubagentManager_Spawn_SnapshotLiveDuringExecution verifies that the
+
 // Broadcaster snapshot contains the agent between agent_spawn and agent_gc.
+
 // Because Publish() updates the agent map before dispatching to subscribers,
+
 // the snapshot is guaranteed to be non-empty as soon as agent_spawn is
+
 // received on the channel.
+
 func TestSubagentManager_Spawn_SnapshotLiveDuringExecution(t *testing.T) {
 	b := orch.NewBroadcaster()
+
 	sub := b.Subscribe()
+
 	defer b.Unsubscribe(sub)
 
 	provider := &MockLLMProvider{}
+
 	mgr := NewSubagentManager(provider, "test-model", "/tmp/test", nil, b, WebSearchToolOptions{})
 
 	_, err := mgr.Spawn(
+
 		context.Background(),
+
 		"any task", "live-test", "", "cli", "direct", "",
+
 		nil,
 	)
 	if err != nil {
@@ -161,39 +222,59 @@ func TestSubagentManager_Spawn_SnapshotLiveDuringExecution(t *testing.T) {
 	}
 
 	// Wait for agent_spawn, then immediately check snapshot.
+
 	deadline := time.After(2 * time.Second)
+
 	for {
 		select {
 		case ev := <-sub.Ch:
+
 			if ev.Type == "agent_spawn" {
 				snap := b.Snapshot()
+
 				if len(snap) == 0 {
 					t.Error("snapshot must contain the spawned agent after agent_spawn event")
 				}
+
 				return // test complete; background goroutine drains safely
 			}
+
 		case <-deadline:
+
 			t.Fatal("timed out waiting for agent_spawn event")
 		}
 	}
 }
 
 // TestSubagentManager_Spawn_CancelledDuringExecution verifies that when the
+
 // context is canceled while a subagent's LLM call is in progress, the
+
 // Broadcaster receives agent_gc with reason="canceled" and the agent is
+
 // removed from the snapshot.
+
 //
+
 // Synchronization:
+
 //  1. blockingProvider.ready is closed when Chat() is entered (goroutine is
+
 //     now blocked inside the LLM call).
+
 //  2. Only then is the context canceled, so there is no race between spawn
+
 //     and cancellation.
+
 func TestSubagentManager_Spawn_CancelledDuringExecution(t *testing.T) {
 	b := orch.NewBroadcaster()
+
 	sub := b.Subscribe()
+
 	defer b.Unsubscribe(sub)
 
 	bp := newBlockingProvider()
+
 	mgr := NewSubagentManager(bp, "test-model", "/tmp/test", nil, b, WebSearchToolOptions{})
 
 	_, err := mgr.Spawn(context.Background(), "long task", "cancel-me", "", "cli", "direct", "", nil)
@@ -202,44 +283,61 @@ func TestSubagentManager_Spawn_CancelledDuringExecution(t *testing.T) {
 	}
 
 	// Wait until the subagent goroutine is inside Chat (blocking on ctx).
+
 	select {
 	case <-bp.ready:
+
 	case <-time.After(3 * time.Second):
+
 		t.Fatal("timed out waiting for blockingProvider to enter Chat")
 	}
 
 	// Cancel via CancelTask — the spawned goroutine's detached context is canceled.
+
 	mgr.CancelTask("subagent-1")
 
 	// Collect events until agent_gc.
+
 	var events []orch.Event
+
 	deadline := time.After(3 * time.Second)
+
 loop:
+
 	for {
 		select {
 		case ev := <-sub.Ch:
+
 			events = append(events, ev)
+
 			if ev.Type == "agent_gc" {
 				break loop
 			}
+
 		case <-deadline:
+
 			t.Fatalf("timed out waiting for agent_gc; events so far: %+v", events)
 		}
 	}
 
 	// Locate agent_gc and verify reason = "canceled".
+
 	var gcEv orch.Event
+
 	for _, ev := range events {
 		if ev.Type == "agent_gc" {
 			gcEv = ev
+
 			break
 		}
 	}
+
 	if gcEv.Reason != "canceled" {
 		t.Errorf("agent_gc reason = %q, want %q; events: %+v", gcEv.Reason, "canceled", events)
 	}
 
 	// Snapshot must be empty after the GC event.
+
 	if snap := b.Snapshot(); len(snap) != 0 {
 		t.Errorf("snapshot must be empty after agent_gc(canceled), got: %v", snap)
 	}

pkg/tools/subagent_tool_test.go

@@ -12,19 +12,26 @@ import (
 )
 
 // MockLLMProvider is a test implementation of LLMProvider
+
 type MockLLMProvider struct {
 	lastOptions map[string]any
 }
 
 func (m *MockLLMProvider) Chat(
 	ctx context.Context,
+
 	messages []providers.Message,
+
 	tools []providers.ToolDefinition,
+
 	model string,
+
 	options map[string]any,
 ) (*providers.LLMResponse, error) {
 	m.lastOptions = options
+
 	// Find the last user message to generate a response
+
 	for i := len(messages) - 1; i >= 0; i-- {
 		if messages[i].Role == "user" {
 			return &providers.LLMResponse{
@@ -32,6 +39,7 @@ func (m *MockLLMProvider) Chat(
 			}, nil
 		}
 	}
+
 	return &providers.LLMResponse{Content: "No task provided"}, nil
 }
 
@@ -49,13 +57,19 @@ func (m *MockLLMProvider) GetContextWindow() int {
 
 func TestSubagentManager_SetLLMOptions_AppliesToRunToolLoop(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	manager.SetLLMOptions(2048, 0.6)
+
 	tool := NewSubagentTool(manager)
+
 	tool.SetContext("cli", "direct")
 
 	ctx := context.Background()
+
 	args := map[string]any{"task": "Do something"}
+
 	result := tool.Execute(ctx, args)
 
 	if result == nil || result.IsError {
@@ -65,18 +79,23 @@ func TestSubagentManager_SetLLMOptions_AppliesToRunToolLoop(t *testing.T) {
 	if provider.lastOptions == nil {
 		t.Fatal("Expected LLM options to be passed, got nil")
 	}
+
 	if provider.lastOptions["max_tokens"] != 2048 {
 		t.Fatalf("max_tokens = %v, want %d", provider.lastOptions["max_tokens"], 2048)
 	}
+
 	if provider.lastOptions["temperature"] != 0.6 {
 		t.Fatalf("temperature = %v, want %v", provider.lastOptions["temperature"], 0.6)
 	}
 }
 
 // TestSubagentTool_Name verifies tool name
+
 func TestSubagentTool_Name(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	if tool.Name() != "subagent" {
@@ -85,150 +104,198 @@ func TestSubagentTool_Name(t *testing.T) {
 }
 
 // TestSubagentTool_Description verifies tool description
+
 func TestSubagentTool_Description(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	desc := tool.Description()
+
 	if desc == "" {
 		t.Error("Description should not be empty")
 	}
+
 	if !strings.Contains(desc, "BLOCK") {
 		t.Errorf("Description should mention 'BLOCK', got: %s", desc)
 	}
+
 	if !strings.Contains(desc, "spawn") {
 		t.Errorf("Description should contrast with spawn, got: %s", desc)
 	}
 }
 
 // TestSubagentTool_Parameters verifies tool parameters schema
+
 func TestSubagentTool_Parameters(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	params := tool.Parameters()
+
 	if params == nil {
 		t.Error("Parameters should not be nil")
 	}
 
 	// Check type
+
 	if params["type"] != "object" {
 		t.Errorf("Expected type 'object', got: %v", params["type"])
 	}
 
 	// Check properties
+
 	props, ok := params["properties"].(map[string]any)
+
 	if !ok {
 		t.Fatal("Properties should be a map")
 	}
 
 	// Verify task parameter
+
 	task, ok := props["task"].(map[string]any)
+
 	if !ok {
 		t.Fatal("Task parameter should exist")
 	}
+
 	if task["type"] != "string" {
 		t.Errorf("Task type should be 'string', got: %v", task["type"])
 	}
 
 	// Verify label parameter
+
 	label, ok := props["label"].(map[string]any)
+
 	if !ok {
 		t.Fatal("Label parameter should exist")
 	}
+
 	if label["type"] != "string" {
 		t.Errorf("Label type should be 'string', got: %v", label["type"])
 	}
 
 	// Check required fields
+
 	required, ok := params["required"].([]string)
+
 	if !ok {
 		t.Fatal("Required should be a string array")
 	}
+
 	if len(required) != 1 || required[0] != "task" {
 		t.Errorf("Required should be ['task'], got: %v", required)
 	}
 }
 
 // TestSubagentTool_SetContext verifies context setting
+
 func TestSubagentTool_SetContext(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	tool.SetContext("test-channel", "test-chat")
 
 	// Verify context is set (we can't directly access private fields,
+
 	// but we can verify it doesn't crash)
+
 	// The actual context usage is tested in Execute tests
 }
 
 // TestSubagentTool_Execute_Success tests successful execution
+
 func TestSubagentTool_Execute_Success(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	msgBus := bus.NewMessageBus()
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", msgBus, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
+
 	tool.SetContext("telegram", "chat-123")
 
 	ctx := context.Background()
+
 	args := map[string]any{
-		"task":  "Write a haiku about coding",
+		"task": "Write a haiku about coding",
+
 		"label": "haiku-task",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// Verify basic ToolResult structure
+
 	if result == nil {
 		t.Fatal("Result should not be nil")
 	}
 
 	// Verify no error
+
 	if result.IsError {
 		t.Errorf("Expected success, got error: %s", result.ForLLM)
 	}
 
 	// Verify not async
+
 	if result.Async {
 		t.Error("SubagentTool should be synchronous, not async")
 	}
 
 	// Verify not silent
+
 	if result.Silent {
 		t.Error("SubagentTool should not be silent")
 	}
 
 	// Verify ForUser contains brief summary (not empty)
+
 	if result.ForUser == "" {
 		t.Error("ForUser should contain result summary")
 	}
+
 	if !strings.Contains(result.ForUser, "Task completed") {
 		t.Errorf("ForUser should contain task completion, got: %s", result.ForUser)
 	}
 
 	// Verify ForLLM contains full details
+
 	if result.ForLLM == "" {
 		t.Error("ForLLM should contain full details")
 	}
+
 	if !strings.Contains(result.ForLLM, "haiku-task") {
 		t.Errorf("ForLLM should contain label 'haiku-task', got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "Task completed:") {
 		t.Errorf("ForLLM should contain task result, got: %s", result.ForLLM)
 	}
 }
 
 // TestSubagentTool_Execute_NoLabel tests execution without label
+
 func TestSubagentTool_Execute_NoLabel(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	msgBus := bus.NewMessageBus()
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", msgBus, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"task": "Test task without label",
 	}
@@ -240,18 +307,23 @@ func TestSubagentTool_Execute_NoLabel(t *testing.T) {
 	}
 
 	// ForLLM should show (unnamed) for missing label
+
 	if !strings.Contains(result.ForLLM, "(unnamed)") {
 		t.Errorf("ForLLM should show '(unnamed)' for missing label, got: %s", result.ForLLM)
 	}
 }
 
 // TestSubagentTool_Execute_MissingTask tests error handling for missing task
+
 func TestSubagentTool_Execute_MissingTask(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", nil, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"label": "test",
 	}
@@ -259,29 +331,35 @@ func TestSubagentTool_Execute_MissingTask(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error
+
 	if !result.IsError {
 		t.Error("Expected error for missing task parameter")
 	}
 
 	// ForLLM should contain helpful error with example
+
 	if !strings.Contains(result.ForLLM, `"task"`) {
 		t.Errorf("Error message should mention '\"task\"', got: %s", result.ForLLM)
 	}
+
 	if !strings.Contains(result.ForLLM, "Example") {
 		t.Errorf("Error message should include usage example, got: %s", result.ForLLM)
 	}
 
 	// Err should be set
+
 	if result.Err == nil {
 		t.Error("Err should be set for validation failure")
 	}
 }
 
 // TestSubagentTool_Execute_NilManager tests error handling for nil manager
+
 func TestSubagentTool_Execute_NilManager(t *testing.T) {
 	tool := NewSubagentTool(nil)
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"task": "test task",
 	}
@@ -289,6 +367,7 @@ func TestSubagentTool_Execute_NilManager(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error
+
 	if !result.IsError {
 		t.Error("Expected error for nil manager")
 	}
@@ -299,18 +378,26 @@ func TestSubagentTool_Execute_NilManager(t *testing.T) {
 }
 
 // TestSubagentTool_Execute_ContextPassing verifies context is properly used
+
 func TestSubagentTool_Execute_ContextPassing(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	msgBus := bus.NewMessageBus()
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", msgBus, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	// Set context
+
 	channel := "test-channel"
+
 	chatID := "test-chat"
+
 	tool.SetContext(channel, chatID)
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"task": "Test context passing",
 	}
@@ -318,40 +405,53 @@ func TestSubagentTool_Execute_ContextPassing(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should succeed
+
 	if result.IsError {
 		t.Errorf("Expected success with context, got error: %s", result.ForLLM)
 	}
 
 	// The context is used internally; we can't directly test it
+
 	// but execution success indicates context was handled properly
 }
 
 // TestSubagentTool_ForUserTruncation verifies long content is truncated for user
+
 func TestSubagentTool_ForUserTruncation(t *testing.T) {
 	// Create a mock provider that returns very long content
+
 	provider := &MockLLMProvider{}
+
 	msgBus := bus.NewMessageBus()
+
 	manager := NewSubagentManager(provider, "test-model", "/tmp/test", msgBus, orch.Noop, WebSearchToolOptions{})
+
 	tool := NewSubagentTool(manager)
 
 	ctx := context.Background()
 
 	// Create a task that will generate long response
+
 	longTask := strings.Repeat("This is a very long task description. ", 100)
+
 	args := map[string]any{
-		"task":  longTask,
+		"task": longTask,
+
 		"label": "long-test",
 	}
 
 	result := tool.Execute(ctx, args)
 
 	// ForUser should be truncated to 500 chars + "..."
+
 	maxUserLen := 500
+
 	if len(result.ForUser) > maxUserLen+3 { // +3 for "..."
 		t.Errorf("ForUser should be truncated to ~%d chars, got: %d", maxUserLen, len(result.ForUser))
 	}
 
 	// ForLLM should have full content
+
 	if !strings.Contains(result.ForLLM, longTask[:50]) {
 		t.Error("ForLLM should contain reference to original task")
 	}
@@ -359,21 +459,29 @@ func TestSubagentTool_ForUserTruncation(t *testing.T) {
 
 func TestFormatToolStats(t *testing.T) {
 	tests := []struct {
-		name  string
+		name string
+
 		stats map[string]int
-		want  string
+
+		want string
 	}{
 		{"empty", map[string]int{}, ""},
+
 		{"single", map[string]int{"exec": 3}, "exec:3"},
+
 		{
 			"multiple sorted",
+
 			map[string]int{"read_file": 5, "exec": 3, "write_file": 1},
+
 			"exec:3,read_file:5,write_file:1",
 		},
 	}
+
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			got := formatToolStats(tt.stats)
+
 			if got != tt.want {
 				t.Errorf("formatToolStats(%v) = %q, want %q", tt.stats, got, tt.want)
 			}
@@ -382,15 +490,22 @@ func TestFormatToolStats(t *testing.T) {
 }
 
 // TestSubagentManager_Spawn_SetsMetadata verifies that the bus message from a
+
 // completed spawn includes execution statistics in Metadata.
+
 func TestSubagentManager_Spawn_SetsMetadata(t *testing.T) {
 	provider := &MockLLMProvider{}
+
 	msgBus := bus.NewMessageBus()
+
 	mgr := NewSubagentManager(provider, "test-model", "/tmp/test", msgBus, orch.Noop, WebSearchToolOptions{})
 
 	_, err := mgr.Spawn(
+
 		context.Background(),
+
 		"say hello", "meta-test", "", "cli", "direct", "",
+
 		nil,
 	)
 	if err != nil {
@@ -398,9 +513,13 @@ func TestSubagentManager_Spawn_SetsMetadata(t *testing.T) {
 	}
 
 	// Consume the inbound message from the bus
+
 	ctx, cancel := context.WithTimeout(context.Background(), 3*time.Second)
+
 	defer cancel()
+
 	received, ok := msgBus.ConsumeInbound(ctx)
+
 	if !ok {
 		t.Fatal("timed out waiting for bus message")
 	}
@@ -408,16 +527,21 @@ func TestSubagentManager_Spawn_SetsMetadata(t *testing.T) {
 	if received.Channel != "system" {
 		t.Fatalf("expected channel 'system', got %q", received.Channel)
 	}
+
 	if received.Metadata == nil {
 		t.Fatal("Metadata should not be nil")
 	}
+
 	if received.Metadata["iterations"] != "1" {
 		t.Errorf("iterations = %q, want %q", received.Metadata["iterations"], "1")
 	}
+
 	if received.Metadata["tool_calls"] != "0" {
 		t.Errorf("tool_calls = %q, want %q", received.Metadata["tool_calls"], "0")
 	}
+
 	// duration_ms should be a non-negative number
+
 	if received.Metadata["duration_ms"] == "" {
 		t.Error("duration_ms should be present")
 	}

pkg/tools/submit_plan.go

@@ -0,0 +1,186 @@
+package tools
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"strings"
+)
+
+// SubmitPlanTool allows a subagent to submit a plan for conductor review.
+
+// The subagent blocks until the conductor approves or rejects.
+
+type SubmitPlanTool struct {
+	taskID string
+
+	conductorKey string
+
+	subagentKey string
+
+	outCh chan<- ContainerMessage
+
+	inCh <-chan string
+
+	recorder SessionRecorder
+
+	setPlan func(goal string, steps []string) // callback to record plan on task
+}
+
+func NewSubmitPlanTool(
+	taskID, conductorKey, subagentKey string,
+
+	outCh chan<- ContainerMessage,
+
+	inCh <-chan string,
+
+	recorder SessionRecorder,
+) *SubmitPlanTool {
+	return &SubmitPlanTool{
+		taskID: taskID,
+
+		conductorKey: conductorKey,
+
+		subagentKey: subagentKey,
+
+		outCh: outCh,
+
+		inCh: inCh,
+
+		recorder: recorder,
+	}
+}
+
+// SetPlanCallback sets the function called when a plan is approved to record
+
+// the goal and steps on the parent SubagentTask.
+
+func (t *SubmitPlanTool) SetPlanCallback(fn func(goal string, steps []string)) {
+	t.setPlan = fn
+}
+
+func (t *SubmitPlanTool) Name() string { return "submit_plan" }
+
+func (t *SubmitPlanTool) Description() string {
+	return "Submit your execution plan for conductor review. Blocks until the conductor approves or rejects. On rejection, revise and resubmit."
+}
+
+func (t *SubmitPlanTool) Parameters() map[string]any {
+	return map[string]any{
+		"type": "object",
+
+		"properties": map[string]any{
+			"goal": map[string]any{
+				"type": "string",
+
+				"description": "The goal of the plan",
+			},
+
+			"steps": map[string]any{
+				"type": "array",
+
+				"description": "Ordered list of steps to execute",
+
+				"items": map[string]any{"type": "string"},
+			},
+		},
+
+		"required": []string{"goal", "steps"},
+	}
+}
+
+func (t *SubmitPlanTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
+	goal, _ := args["goal"].(string)
+
+	if goal == "" {
+		return ErrorResult("required parameter \"goal\" (string) is missing")
+	}
+
+	stepsRaw, _ := args["steps"]
+
+	var steps []string
+
+	switch v := stepsRaw.(type) {
+	case []any:
+
+		steps = make([]string, 0, len(v))
+
+		for _, s := range v {
+			if str, ok := s.(string); ok {
+				steps = append(steps, str)
+			}
+		}
+
+	case []string:
+
+		steps = v
+	}
+
+	if len(steps) == 0 {
+		return ErrorResult("required parameter \"steps\" (array of strings) is missing or empty")
+	}
+
+	// Build plan text for recording and display.
+
+	var b strings.Builder
+
+	b.WriteString("Goal: ")
+
+	b.WriteString(goal)
+
+	b.WriteByte('\n')
+
+	for i, step := range steps {
+		fmt.Fprintf(&b, "%d. %s\n", i+1, step)
+	}
+
+	planText := b.String()
+
+	// Record in session DAG.
+
+	if t.recorder != nil {
+		_ = t.recorder.RecordPlanSubmit(t.conductorKey, t.subagentKey, t.taskID, planText)
+	}
+
+	// Encode plan as JSON for the conductor.
+
+	planJSON, _ := json.Marshal(map[string]any{"goal": goal, "steps": steps})
+
+	// Send plan_review to conductor.
+
+	select {
+	case t.outCh <- ContainerMessage{Type: "plan_review", Content: string(planJSON), TaskID: t.taskID}:
+
+	case <-ctx.Done():
+
+		return ErrorResult(fmt.Sprintf("context canceled while submitting plan: %v", ctx.Err()))
+	}
+
+	// Wait for conductor's decision.
+
+	select {
+	case decision := <-t.inCh:
+
+		if strings.HasPrefix(decision, "approved") {
+			if t.setPlan != nil {
+				t.setPlan(goal, steps)
+			}
+
+			return &ToolResult{
+				ForLLM: "Plan approved by conductor. Proceed with execution.",
+
+				ForUser: "Plan approved",
+			}
+		}
+
+		return &ToolResult{
+			ForLLM: fmt.Sprintf("Plan rejected by conductor: %s\nRevise your plan and resubmit.", decision),
+
+			ForUser: fmt.Sprintf("Plan rejected: %s", decision),
+		}
+
+	case <-ctx.Done():
+
+		return ErrorResult(fmt.Sprintf("context canceled while waiting for review: %v", ctx.Err()))
+	}
+}

pkg/tools/submit_plan_test.go

@@ -0,0 +1,229 @@
+package tools
+
+import (
+	"context"
+	"testing"
+	"time"
+)
+
+func TestSubmitPlanTool_Execute_Approved(t *testing.T) {
+	outCh := make(chan ContainerMessage, 4)
+
+	inCh := make(chan string, 1)
+
+	tool := NewSubmitPlanTool("subagent-1", "conductor:main", "subagent:subagent-1", outCh, inCh, nil)
+
+	var gotGoal string
+
+	var gotSteps []string
+
+	tool.SetPlanCallback(func(goal string, steps []string) {
+		gotGoal = goal
+
+		gotSteps = steps
+	})
+
+	if tool.Name() != "submit_plan" {
+		t.Errorf("Name() = %q, want %q", tool.Name(), "submit_plan")
+	}
+
+	// Simulate conductor approving in background.
+
+	go func() {
+		msg := <-outCh
+
+		if msg.Type != "plan_review" {
+			t.Errorf("msg.Type = %q, want %q", msg.Type, "plan_review")
+		}
+
+		inCh <- "approved"
+	}()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+
+	defer cancel()
+
+	result := tool.Execute(ctx, map[string]any{
+		"goal": "Add auth",
+
+		"steps": []any{"Add middleware", "Add JWT"},
+	})
+
+	if result.IsError {
+		t.Fatalf("unexpected error: %s", result.ForLLM)
+	}
+
+	if gotGoal != "Add auth" {
+		t.Errorf("setPlan goal = %q, want %q", gotGoal, "Add auth")
+	}
+
+	if len(gotSteps) != 2 {
+		t.Errorf("setPlan steps count = %d, want 2", len(gotSteps))
+	}
+}
+
+func TestSubmitPlanTool_Execute_Rejected(t *testing.T) {
+	outCh := make(chan ContainerMessage, 4)
+
+	inCh := make(chan string, 1)
+
+	tool := NewSubmitPlanTool("subagent-1", "conductor:main", "subagent:subagent-1", outCh, inCh, nil)
+
+	var planSet bool
+
+	tool.SetPlanCallback(func(goal string, steps []string) {
+		planSet = true
+	})
+
+	go func() {
+		<-outCh
+
+		inCh <- "rejected: needs more detail on step 2"
+	}()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+
+	defer cancel()
+
+	result := tool.Execute(ctx, map[string]any{
+		"goal": "Add auth",
+
+		"steps": []any{"Add middleware"},
+	})
+
+	if result.IsError {
+		t.Fatalf("unexpected error: %s", result.ForLLM)
+	}
+
+	if planSet {
+		t.Error("setPlan should NOT be called on rejection")
+	}
+
+	if result.ForLLM == "" {
+		t.Error("ForLLM should contain rejection message")
+	}
+}
+
+func TestSubmitPlanTool_MissingParams(t *testing.T) {
+	tool := NewSubmitPlanTool("subagent-1", "", "", nil, nil, nil)
+
+	result := tool.Execute(context.Background(), map[string]any{})
+
+	if !result.IsError {
+		t.Error("expected error for missing goal")
+	}
+
+	result = tool.Execute(context.Background(), map[string]any{"goal": "test"})
+
+	if !result.IsError {
+		t.Error("expected error for missing steps")
+	}
+}
+
+func TestSubmitPlanTool_ContextCanceled(t *testing.T) {
+	outCh := make(chan ContainerMessage) // unbuffered
+
+	inCh := make(chan string)
+
+	tool := NewSubmitPlanTool("subagent-1", "", "", outCh, inCh, nil)
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	cancel()
+
+	result := tool.Execute(ctx, map[string]any{
+		"goal": "test",
+
+		"steps": []any{"step1"},
+	})
+
+	if !result.IsError {
+		t.Error("expected error on canceled context")
+	}
+}
+
+func TestAnswerSubagentTool_Execute(t *testing.T) {
+	mgr := &SubagentManager{
+		tasks: map[string]*SubagentTask{
+			"subagent-1": {
+				ID: "subagent-1",
+
+				inCh: make(chan string, 1),
+			},
+		},
+	}
+
+	tool := NewAnswerSubagentTool(mgr)
+
+	if tool.Name() != "answer_subagent" {
+		t.Errorf("Name() = %q, want %q", tool.Name(), "answer_subagent")
+	}
+
+	result := tool.Execute(context.Background(), map[string]any{
+		"task_id": "subagent-1",
+
+		"answer": "Use port 8080",
+	})
+
+	if result.IsError {
+		t.Fatalf("unexpected error: %s", result.ForLLM)
+	}
+
+	// Verify the answer was sent.
+
+	answer := <-mgr.tasks["subagent-1"].inCh
+
+	if answer != "Use port 8080" {
+		t.Errorf("answer = %q, want %q", answer, "Use port 8080")
+	}
+}
+
+func TestAnswerSubagentTool_MissingParams(t *testing.T) {
+	tool := NewAnswerSubagentTool(nil)
+
+	result := tool.Execute(context.Background(), map[string]any{})
+
+	if !result.IsError {
+		t.Error("expected error for missing task_id")
+	}
+
+	result = tool.Execute(context.Background(), map[string]any{"task_id": "x"})
+
+	if !result.IsError {
+		t.Error("expected error for missing answer")
+	}
+}
+
+func TestReviewSubagentPlanTool_Execute(t *testing.T) {
+	mgr := &SubagentManager{
+		tasks: map[string]*SubagentTask{
+			"subagent-1": {
+				ID: "subagent-1",
+
+				inCh: make(chan string, 1),
+			},
+		},
+	}
+
+	tool := NewReviewSubagentPlanTool(mgr)
+
+	if tool.Name() != "review_subagent_plan" {
+		t.Errorf("Name() = %q, want %q", tool.Name(), "review_subagent_plan")
+	}
+
+	result := tool.Execute(context.Background(), map[string]any{
+		"task_id": "subagent-1",
+
+		"decision": "approved",
+	})
+
+	if result.IsError {
+		t.Fatalf("unexpected error: %s", result.ForLLM)
+	}
+
+	decision := <-mgr.tasks["subagent-1"].inCh
+
+	if decision != "approved" {
+		t.Errorf("decision = %q, want %q", decision, "approved")
+	}
+}

pkg/tools/toolloop.go

@@ -1,7 +1,11 @@
 // PicoClaw - Ultra-lightweight personal AI agent
+
 // Inspired by and based on nanobot: https://github.com/HKUDS/nanobot
+
 // License: MIT
+
 //
+
 // Copyright (c) 2026 PicoClaw contributors
 
 package tools
@@ -18,140 +22,207 @@ import (
 )
 
 // ToolLoopConfig configures the tool execution loop.
+
 type ToolLoopConfig struct {
-	Provider      providers.LLMProvider
-	Model         string
-	Tools         *ToolRegistry
+	Provider providers.LLMProvider
+
+	Model string
+
+	Tools *ToolRegistry
+
 	MaxIterations int
-	LLMOptions    map[string]any
+
+	LLMOptions map[string]any
+
 	// Reporter and AgentID replace the old OnStateChange func.
+
 	// Reporter is called with ReportStateChange("waiting","") before each LLM
+
 	// call and ReportStateChange("toolcall", toolName) when each tool starts.
+
 	// Pass nil or orch.Noop to disable. nil is treated as orch.Noop internally.
+
 	Reporter orch.AgentReporter
-	AgentID  string
+
+	AgentID string
 }
 
 // ToolLoopResult contains the result of running the tool loop.
+
 type ToolLoopResult struct {
-	Content    string
+	Content string
+
 	Iterations int
-	ToolCalls  int            // total tool call count across all iterations
-	ToolStats  map[string]int // tool name → call count
+
+	ToolCalls int // total tool call count across all iterations
+
+	ToolStats map[string]int // tool name → call count
 }
 
 // RunToolLoop executes the LLM + tool call iteration loop.
+
 // This is the core agent logic that can be reused by both main agent and subagents.
+
 func RunToolLoop(
 	ctx context.Context,
+
 	config ToolLoopConfig,
+
 	messages []providers.Message,
+
 	channel, chatID string,
 ) (*ToolLoopResult, error) {
 	reporter := config.Reporter
+
 	if reporter == nil {
 		reporter = orch.Noop
 	}
 
 	iteration := 0
+
 	totalToolCalls := 0
+
 	toolStats := map[string]int{}
+
 	var finalContent string
 
 	for iteration < config.MaxIterations {
 		iteration++
 
 		logger.DebugCF("toolloop", "LLM iteration",
+
 			map[string]any{
 				"iteration": iteration,
-				"max":       config.MaxIterations,
+
+				"max": config.MaxIterations,
 			})
 
 		// 1. Build tool definitions
+
 		var providerToolDefs []providers.ToolDefinition
+
 		if config.Tools != nil {
 			providerToolDefs = config.Tools.ToProviderDefs()
 		}
 
 		// 2. Set default LLM options
+
 		llmOpts := config.LLMOptions
+
 		if llmOpts == nil {
 			llmOpts = map[string]any{}
 		}
+
 		// 3. Call LLM  (hook: waiting for response)
+
 		reporter.ReportStateChange(config.AgentID, orch.AgentStateWaiting, "")
+
 		response, err := config.Provider.Chat(ctx, messages, providerToolDefs, config.Model, llmOpts)
 		if err != nil {
 			logger.ErrorCF("toolloop", "LLM call failed",
+
 				map[string]any{
 					"iteration": iteration,
-					"error":     err.Error(),
+
+					"error": err.Error(),
 				})
+
 			return nil, fmt.Errorf("LLM call failed: %w", err)
 		}
 
 		// 4. If no tool calls, we're done
+
 		if len(response.ToolCalls) == 0 {
 			finalContent = response.Content
+
 			logger.InfoCF("toolloop", "LLM response without tool calls (direct answer)",
+
 				map[string]any{
-					"iteration":     iteration,
+					"iteration": iteration,
+
 					"content_chars": len(finalContent),
 				})
+
 			break
 		}
 
 		normalizedToolCalls := make([]providers.ToolCall, 0, len(response.ToolCalls))
+
 		for _, tc := range response.ToolCalls {
 			normalizedToolCalls = append(normalizedToolCalls, providers.NormalizeToolCall(tc))
 		}
 
 		// 5. Log tool calls
+
 		toolNames := make([]string, 0, len(normalizedToolCalls))
+
 		for _, tc := range normalizedToolCalls {
 			toolNames = append(toolNames, tc.Name)
 		}
+
 		logger.InfoCF("toolloop", "LLM requested tool calls",
+
 			map[string]any{
-				"tools":     toolNames,
-				"count":     len(normalizedToolCalls),
+				"tools": toolNames,
+
+				"count": len(normalizedToolCalls),
+
 				"iteration": iteration,
 			})
 
 		// 6. Build assistant message with tool calls
+
 		assistantMsg := providers.Message{
-			Role:    "assistant",
+			Role: "assistant",
+
 			Content: response.Content,
 		}
+
 		for _, tc := range normalizedToolCalls {
 			assistantMsg.ToolCalls = append(assistantMsg.ToolCalls, providers.ToolCall{
-				ID:        tc.ID,
-				Type:      "function",
-				Name:      tc.Name,
+				ID: tc.ID,
+
+				Type: "function",
+
+				Name: tc.Name,
+
 				Arguments: tc.Arguments,
+
 				Function: &providers.FunctionCall{
-					Name:      tc.Name,
+					Name: tc.Name,
+
 					Arguments: tc.Arguments,
 				},
 			})
 		}
+
 		messages = append(messages, assistantMsg)
 
 		// 7. Execute tool calls  (hook: toolcall per tool)
+
 		for _, tc := range normalizedToolCalls {
 			argsJSON, _ := json.Marshal(tc.Arguments)
+
 			argsPreview := utils.Truncate(string(argsJSON), 200)
+
 			logger.InfoCF("toolloop", fmt.Sprintf("Tool call: %s(%s)", tc.Name, argsPreview),
+
 				map[string]any{
-					"tool":      tc.Name,
+					"tool": tc.Name,
+
 					"iteration": iteration,
 				})
+
 			reporter.ReportStateChange(config.AgentID, orch.AgentStateToolCall, tc.Name)
+
 			totalToolCalls++
+
 			toolStats[tc.Name]++
 
 			// Execute tool (no async callback for subagents - they run independently)
+
 			var toolResult *ToolResult
+
 			if config.Tools != nil {
 				toolResult = config.Tools.ExecuteWithContext(ctx, tc.Name, tc.Arguments, channel, chatID, nil)
 			} else {
@@ -159,25 +230,34 @@ func RunToolLoop(
 			}
 
 			// Determine content for LLM
+
 			contentForLLM := toolResult.ForLLM
+
 			if contentForLLM == "" && toolResult.Err != nil {
 				contentForLLM = toolResult.Err.Error()
 			}
 
 			// Add tool result message
+
 			toolResultMsg := providers.Message{
-				Role:       "tool",
-				Content:    contentForLLM,
+				Role: "tool",
+
+				Content: contentForLLM,
+
 				ToolCallID: tc.ID,
 			}
+
 			messages = append(messages, toolResultMsg)
 		}
 	}
 
 	return &ToolLoopResult{
-		Content:    finalContent,
+		Content: finalContent,
+
 		Iterations: iteration,
-		ToolCalls:  totalToolCalls,
-		ToolStats:  toolStats,
+
+		ToolCalls: totalToolCalls,
+
+		ToolStats: toolStats,
 	}, nil
 }

pkg/tools/toolloop_reporter_test.go

@@ -10,53 +10,78 @@ import (
 )
 
 // reporterSpy records every ReportStateChange call in order.
+
 // Spawn/Conversation/GC are not needed for toolloop tests.
+
 type reporterSpy struct {
-	mu    sync.Mutex
+	mu sync.Mutex
+
 	calls []spyCall
 }
 
 type spyCall struct {
 	state orch.AgentState
-	tool  string
+
+	tool string
 }
 
-func (r *reporterSpy) ReportSpawn(id, label, task string)       {}
+func (r *reporterSpy) ReportSpawn(id, label, task string) {}
+
 func (r *reporterSpy) ReportConversation(from, to, text string) {}
-func (r *reporterSpy) ReportGC(id, reason string)               {}
+
+func (r *reporterSpy) ReportGC(id, reason string) {}
+
 func (r *reporterSpy) ReportStateChange(id string, state orch.AgentState, tool string) {
 	r.mu.Lock()
+
 	r.calls = append(r.calls, spyCall{state, tool})
+
 	r.mu.Unlock()
 }
 
 func (r *reporterSpy) snapshot() []spyCall {
 	r.mu.Lock()
+
 	defer r.mu.Unlock()
+
 	out := make([]spyCall, len(r.calls))
+
 	copy(out, r.calls)
+
 	return out
 }
 
 // sequenceMockProvider returns a tool call on the first Chat() call and a
+
 // plain text response on all subsequent calls. Used to exercise the
+
 // waiting → toolcall → waiting event sequence in RunToolLoop.
+
 type sequenceMockProvider struct {
-	mu        sync.Mutex
+	mu sync.Mutex
+
 	callCount int
 }
 
 func (m *sequenceMockProvider) Chat(
 	_ context.Context,
+
 	_ []providers.Message,
+
 	_ []providers.ToolDefinition,
+
 	_ string,
+
 	_ map[string]any,
 ) (*providers.LLMResponse, error) {
 	m.mu.Lock()
+
 	m.callCount++
+
 	n := m.callCount
+
 	m.mu.Unlock()
+
 	if n == 1 {
 		return &providers.LLMResponse{
 			ToolCalls: []providers.ToolCall{
@@ -64,17 +89,24 @@ func (m *sequenceMockProvider) Chat(
 			},
 		}, nil
 	}
+
 	return &providers.LLMResponse{Content: "done"}, nil
 }
+
 func (m *sequenceMockProvider) GetDefaultModel() string { return "test" }
-func (m *sequenceMockProvider) SupportsTools() bool     { return true }
-func (m *sequenceMockProvider) GetContextWindow() int   { return 4096 }
+
+func (m *sequenceMockProvider) SupportsTools() bool { return true }
+
+func (m *sequenceMockProvider) GetContextWindow() int { return 4096 }
 
 // echoTool is a minimal Tool stub registered as "echo_tool".
+
 type echoTool struct{}
 
-func (t *echoTool) Name() string        { return "echo_tool" }
+func (t *echoTool) Name() string { return "echo_tool" }
+
 func (t *echoTool) Description() string { return "echo" }
+
 func (t *echoTool) Parameters() map[string]any {
 	return map[string]any{"type": "object", "properties": map[string]any{}}
 }
@@ -84,13 +116,19 @@ func (t *echoTool) Execute(_ context.Context, _ map[string]any) *ToolResult {
 }
 
 // TestToolLoop_NilReporter_FallsBackToNoop ensures that passing nil as
+
 // Reporter does not panic — the loop must substitute orch.Noop internally.
+
 func TestToolLoop_NilReporter_FallsBackToNoop(t *testing.T) {
 	_, err := RunToolLoop(context.Background(), ToolLoopConfig{
-		Provider:      &MockLLMProvider{},
-		Model:         "test",
+		Provider: &MockLLMProvider{},
+
+		Model: "test",
+
 		MaxIterations: 1,
-		Reporter:      nil, // must not panic
+
+		Reporter: nil, // must not panic
+
 	}, []providers.Message{{Role: "user", Content: "hi"}}, "cli", "direct")
 	if err != nil {
 		t.Fatalf("unexpected error with nil reporter: %v", err)
@@ -98,120 +136,171 @@ func TestToolLoop_NilReporter_FallsBackToNoop(t *testing.T) {
 }
 
 // TestToolLoop_Reporter_WaitingBeforeLLM verifies that ReportStateChange is
+
 // called with state="waiting" before the first LLM call. The mock provider
+
 // returns a direct text answer (no tool calls), so exactly one waiting event
+
 // is expected.
+
 func TestToolLoop_Reporter_WaitingBeforeLLM(t *testing.T) {
 	rep := &reporterSpy{}
+
 	_, err := RunToolLoop(context.Background(), ToolLoopConfig{
-		Provider:      &MockLLMProvider{},
-		Model:         "test",
+		Provider: &MockLLMProvider{},
+
+		Model: "test",
+
 		MaxIterations: 1,
-		Reporter:      rep,
-		AgentID:       "sess-1",
+
+		Reporter: rep,
+
+		AgentID: "sess-1",
 	}, []providers.Message{{Role: "user", Content: "hi"}}, "cli", "direct")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
+
 	calls := rep.snapshot()
+
 	if len(calls) == 0 {
 		t.Fatal("expected at least one ReportStateChange call")
 	}
+
 	if calls[0].state != orch.AgentStateWaiting {
 		t.Fatalf("first call must be state=waiting, got %+v", calls[0])
 	}
 }
 
 // TestToolLoop_Reporter_ToolcallOrderedAfterWaiting verifies the canonical
+
 // two-iteration sequence:
+
 //
+
 //	waiting  (before 1st LLM call)
+
 //	toolcall(echo_tool)  (before tool execution)
+
 //	waiting  (before 2nd LLM call)
+
 //
+
 // The sequenceMockProvider returns a tool call on iteration 1 and a text
+
 // response on iteration 2, driving exactly this path.
+
 func TestToolLoop_Reporter_ToolcallOrderedAfterWaiting(t *testing.T) {
 	rep := &reporterSpy{}
+
 	reg := NewToolRegistry()
+
 	reg.Register(&echoTool{})
 
 	_, err := RunToolLoop(context.Background(), ToolLoopConfig{
-		Provider:      &sequenceMockProvider{},
-		Model:         "test",
-		Tools:         reg,
+		Provider: &sequenceMockProvider{},
+
+		Model: "test",
+
+		Tools: reg,
+
 		MaxIterations: 5,
-		Reporter:      rep,
-		AgentID:       "sess-1",
+
+		Reporter: rep,
+
+		AgentID: "sess-1",
 	}, []providers.Message{{Role: "user", Content: "do it"}}, "cli", "direct")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
 
 	calls := rep.snapshot()
+
 	if len(calls) < 3 {
 		t.Fatalf("expected at least 3 calls, got %d: %+v", len(calls), calls)
 	}
+
 	if calls[0].state != orch.AgentStateWaiting {
 		t.Fatalf("calls[0] must be waiting, got %+v", calls[0])
 	}
+
 	if calls[1].state != orch.AgentStateToolCall || calls[1].tool != "echo_tool" {
 		t.Fatalf("calls[1] must be toolcall(echo_tool), got %+v", calls[1])
 	}
+
 	if calls[2].state != orch.AgentStateWaiting {
 		t.Fatalf("calls[2] must be waiting (2nd LLM iteration), got %+v", calls[2])
 	}
 }
 
 // TestToolLoop_ToolCallStats verifies that ToolLoopResult.ToolCalls and
+
 // ToolStats are populated correctly after a tool call iteration.
+
 func TestToolLoop_ToolCallStats(t *testing.T) {
 	reg := NewToolRegistry()
+
 	reg.Register(&echoTool{})
 
 	result, err := RunToolLoop(context.Background(), ToolLoopConfig{
-		Provider:      &sequenceMockProvider{},
-		Model:         "test",
-		Tools:         reg,
+		Provider: &sequenceMockProvider{},
+
+		Model: "test",
+
+		Tools: reg,
+
 		MaxIterations: 5,
 	}, []providers.Message{{Role: "user", Content: "do it"}}, "cli", "direct")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
+
 	if result.ToolCalls != 1 {
 		t.Errorf("ToolCalls = %d, want 1", result.ToolCalls)
 	}
+
 	if result.ToolStats["echo_tool"] != 1 {
 		t.Errorf("ToolStats[echo_tool] = %d, want 1", result.ToolStats["echo_tool"])
 	}
+
 	if result.Iterations != 2 {
 		t.Errorf("Iterations = %d, want 2", result.Iterations)
 	}
 }
 
 // TestToolLoop_NoToolCalls_ZeroStats verifies that a direct answer (no tool
+
 // calls) produces zero ToolCalls and an empty ToolStats map.
+
 func TestToolLoop_NoToolCalls_ZeroStats(t *testing.T) {
 	result, err := RunToolLoop(context.Background(), ToolLoopConfig{
-		Provider:      &MockLLMProvider{},
-		Model:         "test",
+		Provider: &MockLLMProvider{},
+
+		Model: "test",
+
 		MaxIterations: 1,
 	}, []providers.Message{{Role: "user", Content: "hi"}}, "cli", "direct")
 	if err != nil {
 		t.Fatalf("unexpected error: %v", err)
 	}
+
 	if result.ToolCalls != 0 {
 		t.Errorf("ToolCalls = %d, want 0", result.ToolCalls)
 	}
+
 	if len(result.ToolStats) != 0 {
 		t.Errorf("ToolStats = %v, want empty", result.ToolStats)
 	}
 }
 
 // TestToolLoop_Reporter_NoopImplementsInterface is a compile-time check that
+
 // orch.Noop satisfies the orch.AgentReporter interface accepted by
+
 // ToolLoopConfig.Reporter. If Noop ever stops implementing the interface the
+
 // build will fail here before any test runs.
+
 func TestToolLoop_Reporter_NoopImplementsInterface(t *testing.T) {
 	var _ orch.AgentReporter = orch.Noop
 }

pkg/tools/web.go

@@ -17,35 +17,51 @@ const (
 	userAgent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
 
 	// HTTP client timeouts for web tool providers.
-	searchTimeout     = 10 * time.Second // Brave, Tavily, DuckDuckGo
+
+	searchTimeout = 10 * time.Second // Brave, Tavily, DuckDuckGo
+
 	perplexityTimeout = 30 * time.Second // Perplexity (LLM-based, slower)
-	fetchTimeout      = 60 * time.Second // WebFetchTool
+
+	fetchTimeout = 60 * time.Second // WebFetchTool
 
 	defaultMaxChars = 50000
-	maxRedirects    = 5
+
+	maxRedirects = 5
 )
 
 // Pre-compiled regexes for HTML text extraction
+
 var (
-	reScript     = regexp.MustCompile(`<script[\s\S]*?</script>`)
-	reStyle      = regexp.MustCompile(`<style[\s\S]*?</style>`)
-	reTags       = regexp.MustCompile(`<[^>]+>`)
+	reScript = regexp.MustCompile(`<script[\s\S]*?</script>`)
+
+	reStyle = regexp.MustCompile(`<style[\s\S]*?</style>`)
+
+	reTags = regexp.MustCompile(`<[^>]+>`)
+
 	reWhitespace = regexp.MustCompile(`[^\S\n]+`)
+
 	reBlankLines = regexp.MustCompile(`\n{3,}`)
 
 	// DuckDuckGo result extraction
-	reDDGLink    = regexp.MustCompile(`<a[^>]*class="[^"]*result__a[^"]*"[^>]*href="([^"]+)"[^>]*>([\s\S]*?)</a>`)
+
+	reDDGLink = regexp.MustCompile(`<a[^>]*class="[^"]*result__a[^"]*"[^>]*href="([^"]+)"[^>]*>([\s\S]*?)</a>`)
+
 	reDDGSnippet = regexp.MustCompile(`<a class="result__snippet[^"]*".*?>([\s\S]*?)</a>`)
 )
 
 // createHTTPClient creates an HTTP client with optional proxy support
+
 func createHTTPClient(proxyURL string, timeout time.Duration) (*http.Client, error) {
 	client := &http.Client{
 		Timeout: timeout,
+
 		Transport: &http.Transport{
-			MaxIdleConns:        10,
-			IdleConnTimeout:     30 * time.Second,
-			DisableCompression:  false,
+			MaxIdleConns: 10,
+
+			IdleConnTimeout: 30 * time.Second,
+
+			DisableCompression: false,
+
 			TLSHandshakeTimeout: 15 * time.Second,
 		},
 	}
@@ -55,18 +71,26 @@ func createHTTPClient(proxyURL string, timeout time.Duration) (*http.Client, err
 		if err != nil {
 			return nil, fmt.Errorf("invalid proxy URL: %w", err)
 		}
+
 		scheme := strings.ToLower(proxy.Scheme)
+
 		switch scheme {
 		case "http", "https", "socks5", "socks5h":
+
 		default:
+
 			return nil, fmt.Errorf(
+
 				"unsupported proxy scheme %q (supported: http, https, socks5, socks5h)",
+
 				proxy.Scheme,
 			)
 		}
+
 		if proxy.Host == "" {
 			return nil, fmt.Errorf("invalid proxy URL: missing host")
 		}
+
 		client.Transport.(*http.Transport).Proxy = http.ProxyURL(proxy)
 	} else {
 		client.Transport.(*http.Transport).Proxy = http.ProxyFromEnvironment
@@ -80,8 +104,10 @@ type SearchProvider interface {
 }
 
 type searchResultItem struct {
-	Title   string
-	URL     string
+	Title string
+
+	URL string
+
 	Snippet string
 }
 
@@ -91,32 +117,41 @@ func formatWebSearchResults(query, provider string, results []searchResultItem,
 	}
 
 	header := fmt.Sprintf("Results for: %s", query)
+
 	if provider != "" {
 		header += " (via " + provider + ")"
 	}
 
 	var sb strings.Builder
+
 	sb.WriteString(header)
+
 	for i, item := range results {
 		if i >= count {
 			break
 		}
+
 		fmt.Fprintf(&sb, "\n%d. %s\n   %s", i+1, item.Title, item.URL)
+
 		if item.Snippet != "" {
 			fmt.Fprintf(&sb, "\n   %s", item.Snippet)
 		}
 	}
+
 	return sb.String()
 }
 
 type BraveSearchProvider struct {
 	apiKey string
-	proxy  string
+
+	proxy string
+
 	client *http.Client
 }
 
 func (p *BraveSearchProvider) Search(ctx context.Context, query string, count int) (string, error) {
 	searchURL := fmt.Sprintf("https://api.search.brave.com/res/v1/web/search?q=%s&count=%d",
+
 		url.QueryEscape(query), count)
 
 	req, err := http.NewRequestWithContext(ctx, "GET", searchURL, nil)
@@ -125,12 +160,14 @@ func (p *BraveSearchProvider) Search(ctx context.Context, query string, count in
 	}
 
 	req.Header.Set("Accept", "application/json")
+
 	req.Header.Set("X-Subscription-Token", p.apiKey)
 
 	resp, err := p.client.Do(req)
 	if err != nil {
 		return "", fmt.Errorf("request failed: %w", err)
 	}
+
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
@@ -141,8 +178,10 @@ func (p *BraveSearchProvider) Search(ctx context.Context, query string, count in
 	var searchResp struct {
 		Web struct {
 			Results []struct {
-				Title       string `json:"title"`
-				URL         string `json:"url"`
+				Title string `json:"title"`
+
+				URL string `json:"url"`
+
 				Description string `json:"description"`
 			} `json:"results"`
 		} `json:"web"`
@@ -150,16 +189,22 @@ func (p *BraveSearchProvider) Search(ctx context.Context, query string, count in
 
 	if err := json.Unmarshal(body, &searchResp); err != nil {
 		// Log error body for debugging
+
 		fmt.Printf("Brave API Error Body: %s\n", string(body))
+
 		return "", fmt.Errorf("failed to parse response: %w", err)
 	}
 
 	results := searchResp.Web.Results
+
 	items := make([]searchResultItem, 0, len(results))
+
 	for _, item := range results {
 		items = append(items, searchResultItem{
-			Title:   item.Title,
-			URL:     item.URL,
+			Title: item.Title,
+
+			URL: item.URL,
+
 			Snippet: item.Description,
 		})
 	}
@@ -168,26 +213,36 @@ func (p *BraveSearchProvider) Search(ctx context.Context, query string, count in
 }
 
 type TavilySearchProvider struct {
-	apiKey  string
+	apiKey string
+
 	baseURL string
-	proxy   string
-	client  *http.Client
+
+	proxy string
+
+	client *http.Client
 }
 
 func (p *TavilySearchProvider) Search(ctx context.Context, query string, count int) (string, error) {
 	searchURL := p.baseURL
+
 	if searchURL == "" {
 		searchURL = "https://api.tavily.com/search"
 	}
 
 	payload := map[string]any{
-		"api_key":             p.apiKey,
-		"query":               query,
-		"search_depth":        "advanced",
-		"include_answer":      false,
-		"include_images":      false,
+		"api_key": p.apiKey,
+
+		"query": query,
+
+		"search_depth": "advanced",
+
+		"include_answer": false,
+
+		"include_images": false,
+
 		"include_raw_content": false,
-		"max_results":         count,
+
+		"max_results": count,
 	}
 
 	bodyBytes, err := json.Marshal(payload)
@@ -201,12 +256,14 @@ func (p *TavilySearchProvider) Search(ctx context.Context, query string, count i
 	}
 
 	req.Header.Set("Content-Type", "application/json")
+
 	req.Header.Set("User-Agent", userAgent)
 
 	resp, err := p.client.Do(req)
 	if err != nil {
 		return "", fmt.Errorf("request failed: %w", err)
 	}
+
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
@@ -220,8 +277,10 @@ func (p *TavilySearchProvider) Search(ctx context.Context, query string, count i
 
 	var searchResp struct {
 		Results []struct {
-			Title   string `json:"title"`
-			URL     string `json:"url"`
+			Title string `json:"title"`
+
+			URL string `json:"url"`
+
 			Content string `json:"content"`
 		} `json:"results"`
 	}
@@ -231,11 +290,15 @@ func (p *TavilySearchProvider) Search(ctx context.Context, query string, count i
 	}
 
 	results := searchResp.Results
+
 	items := make([]searchResultItem, 0, len(results))
+
 	for _, item := range results {
 		items = append(items, searchResultItem{
-			Title:   item.Title,
-			URL:     item.URL,
+			Title: item.Title,
+
+			URL: item.URL,
+
 			Snippet: item.Content,
 		})
 	}
@@ -244,7 +307,8 @@ func (p *TavilySearchProvider) Search(ctx context.Context, query string, count i
 }
 
 type DuckDuckGoSearchProvider struct {
-	proxy  string
+	proxy string
+
 	client *http.Client
 }
 
@@ -262,6 +326,7 @@ func (p *DuckDuckGoSearchProvider) Search(ctx context.Context, query string, cou
 	if err != nil {
 		return "", fmt.Errorf("request failed: %w", err)
 	}
+
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
@@ -274,11 +339,15 @@ func (p *DuckDuckGoSearchProvider) Search(ctx context.Context, query string, cou
 
 func (p *DuckDuckGoSearchProvider) extractResults(html string, count int, query string) (string, error) {
 	// Simple regex based extraction for DDG HTML
+
 	// Strategy: Find all result containers or key anchors directly
 
 	// Try finding the result links directly first, as they are the most critical
+
 	// Pattern: <a class="result__a" href="...">Title</a>
+
 	// The previous regex was a bit strict. Let's make it more flexible for attributes order/content
+
 	matches := reDDGLink.FindAllStringSubmatch(html, count+5)
 
 	if len(matches) == 0 {
@@ -288,17 +357,22 @@ func (p *DuckDuckGoSearchProvider) extractResults(html string, count int, query
 	snippetMatches := reDDGSnippet.FindAllStringSubmatch(html, count+5)
 
 	maxItems := min(len(matches), count)
+
 	items := make([]searchResultItem, 0, maxItems)
 
 	for i := range maxItems {
 		urlStr := matches[i][1]
+
 		title := stripTags(matches[i][2])
+
 		title = strings.TrimSpace(title)
 
 		// URL decoding if needed
+
 		if strings.Contains(urlStr, "uddg=") {
 			if u, err := url.QueryUnescape(urlStr); err == nil {
 				_, after, ok := strings.Cut(u, "uddg=")
+
 				if ok {
 					urlStr = after
 				}
@@ -306,15 +380,20 @@ func (p *DuckDuckGoSearchProvider) extractResults(html string, count int, query
 		}
 
 		snippet := ""
+
 		// Attempt to attach snippet if available and index aligns
+
 		if i < len(snippetMatches) {
 			snippet = stripTags(snippetMatches[i][1])
+
 			snippet = strings.TrimSpace(snippet)
 		}
 
 		items = append(items, searchResultItem{
-			Title:   title,
-			URL:     urlStr,
+			Title: title,
+
+			URL: urlStr,
+
 			Snippet: snippet,
 		})
 	}
@@ -328,7 +407,9 @@ func stripTags(content string) string {
 
 type PerplexitySearchProvider struct {
 	apiKey string
-	proxy  string
+
+	proxy string
+
 	client *http.Client
 }
 
@@ -337,16 +418,21 @@ func (p *PerplexitySearchProvider) Search(ctx context.Context, query string, cou
 
 	payload := map[string]any{
 		"model": "sonar",
+
 		"messages": []map[string]string{
 			{
-				"role":    "system",
+				"role": "system",
+
 				"content": "You are a search assistant. Provide concise search results with titles, URLs, and brief descriptions in the following format:\n1. Title\n   URL\n   Description\n\nDo not add extra commentary.",
 			},
+
 			{
-				"role":    "user",
+				"role": "user",
+
 				"content": fmt.Sprintf("Search for: %s. Provide up to %d relevant results.", query, count),
 			},
 		},
+
 		"max_tokens": 1000,
 	}
 
@@ -361,13 +447,16 @@ func (p *PerplexitySearchProvider) Search(ctx context.Context, query string, cou
 	}
 
 	req.Header.Set("Content-Type", "application/json")
+
 	req.Header.Set("Authorization", "Bearer "+p.apiKey)
+
 	req.Header.Set("User-Agent", userAgent)
 
 	resp, err := p.client.Do(req)
 	if err != nil {
 		return "", fmt.Errorf("request failed: %w", err)
 	}
+
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
@@ -399,45 +488,66 @@ func (p *PerplexitySearchProvider) Search(ctx context.Context, query string, cou
 }
 
 type WebSearchTool struct {
-	provider     SearchProvider
+	provider SearchProvider
+
 	providerName string
-	maxResults   int
+
+	maxResults int
 }
 
 // ProviderName returns the name of the active search provider (e.g. "brave", "perplexity").
+
 func (t *WebSearchTool) ProviderName() string {
 	return t.providerName
 }
 
 type WebSearchToolOptions struct {
-	BraveAPIKey          string
-	BraveMaxResults      int
-	BraveEnabled         bool
-	TavilyAPIKey         string
-	TavilyBaseURL        string
-	TavilyMaxResults     int
-	TavilyEnabled        bool
+	BraveAPIKey string
+
+	BraveMaxResults int
+
+	BraveEnabled bool
+
+	TavilyAPIKey string
+
+	TavilyBaseURL string
+
+	TavilyMaxResults int
+
+	TavilyEnabled bool
+
 	DuckDuckGoMaxResults int
-	DuckDuckGoEnabled    bool
-	PerplexityAPIKey     string
+
+	DuckDuckGoEnabled bool
+
+	PerplexityAPIKey string
+
 	PerplexityMaxResults int
-	PerplexityEnabled    bool
-	Proxy                string
+
+	PerplexityEnabled bool
+
+	Proxy string
 }
 
 func NewWebSearchTool(opts WebSearchToolOptions) (*WebSearchTool, error) {
 	var provider SearchProvider
+
 	var providerName string
+
 	maxResults := 5
 
 	// Priority: Perplexity > Brave > Tavily > DuckDuckGo
+
 	if opts.PerplexityEnabled && opts.PerplexityAPIKey != "" {
 		client, err := createHTTPClient(opts.Proxy, perplexityTimeout)
 		if err != nil {
 			return nil, fmt.Errorf("failed to create HTTP client for Perplexity: %w", err)
 		}
+
 		provider = &PerplexitySearchProvider{apiKey: opts.PerplexityAPIKey, proxy: opts.Proxy, client: client}
+
 		providerName = "perplexity"
+
 		if opts.PerplexityMaxResults > 0 {
 			maxResults = opts.PerplexityMaxResults
 		}
@@ -446,8 +556,11 @@ func NewWebSearchTool(opts WebSearchToolOptions) (*WebSearchTool, error) {
 		if err != nil {
 			return nil, fmt.Errorf("failed to create HTTP client for Brave: %w", err)
 		}
+
 		provider = &BraveSearchProvider{apiKey: opts.BraveAPIKey, proxy: opts.Proxy, client: client}
+
 		providerName = "brave"
+
 		if opts.BraveMaxResults > 0 {
 			maxResults = opts.BraveMaxResults
 		}
@@ -456,13 +569,19 @@ func NewWebSearchTool(opts WebSearchToolOptions) (*WebSearchTool, error) {
 		if err != nil {
 			return nil, fmt.Errorf("failed to create HTTP client for Tavily: %w", err)
 		}
+
 		provider = &TavilySearchProvider{
-			apiKey:  opts.TavilyAPIKey,
+			apiKey: opts.TavilyAPIKey,
+
 			baseURL: opts.TavilyBaseURL,
-			proxy:   opts.Proxy,
-			client:  client,
+
+			proxy: opts.Proxy,
+
+			client: client,
 		}
+
 		providerName = "tavily"
+
 		if opts.TavilyMaxResults > 0 {
 			maxResults = opts.TavilyMaxResults
 		}
@@ -471,8 +590,11 @@ func NewWebSearchTool(opts WebSearchToolOptions) (*WebSearchTool, error) {
 		if err != nil {
 			return nil, fmt.Errorf("failed to create HTTP client for DuckDuckGo: %w", err)
 		}
+
 		provider = &DuckDuckGoSearchProvider{proxy: opts.Proxy, client: client}
+
 		providerName = "duckduckgo"
+
 		if opts.DuckDuckGoMaxResults > 0 {
 			maxResults = opts.DuckDuckGoMaxResults
 		}
@@ -481,9 +603,11 @@ func NewWebSearchTool(opts WebSearchToolOptions) (*WebSearchTool, error) {
 	}
 
 	return &WebSearchTool{
-		provider:     provider,
+		provider: provider,
+
 		providerName: providerName,
-		maxResults:   maxResults,
+
+		maxResults: maxResults,
 	}, nil
 }
 
@@ -498,29 +622,38 @@ func (t *WebSearchTool) Description() string {
 func (t *WebSearchTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"query": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "Search query",
 			},
+
 			"count": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Number of results (1-10)",
-				"minimum":     1.0,
-				"maximum":     10.0,
+
+				"minimum": 1.0,
+
+				"maximum": 10.0,
 			},
 		},
+
 		"required": []string{"query"},
 	}
 }
 
 func (t *WebSearchTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	query, ok := args["query"].(string)
+
 	if !ok {
 		return ErrorResult("query is required")
 	}
 
 	count := t.maxResults
+
 	if c, ok := args["count"].(float64); ok {
 		if int(c) > 0 && int(c) <= 10 {
 			count = int(c)
@@ -533,20 +666,25 @@ func (t *WebSearchTool) Execute(ctx context.Context, args map[string]any) *ToolR
 	}
 
 	return &ToolResult{
-		ForLLM:  result,
+		ForLLM: result,
+
 		ForUser: result,
 	}
 }
 
 type WebFetchTool struct {
 	maxChars int
-	proxy    string
-	client   *http.Client
+
+	proxy string
+
+	client *http.Client
 }
 
 func NewWebFetchTool(maxChars int) *WebFetchTool {
 	// createHTTPClient cannot fail with an empty proxy string.
+
 	tool, _ := NewWebFetchToolWithProxy(maxChars, "")
+
 	return tool
 }
 
@@ -554,20 +692,26 @@ func NewWebFetchToolWithProxy(maxChars int, proxy string) (*WebFetchTool, error)
 	if maxChars <= 0 {
 		maxChars = defaultMaxChars
 	}
+
 	client, err := createHTTPClient(proxy, fetchTimeout)
 	if err != nil {
 		return nil, fmt.Errorf("failed to create HTTP client for web fetch: %w", err)
 	}
+
 	client.CheckRedirect = func(req *http.Request, via []*http.Request) error {
 		if len(via) >= maxRedirects {
 			return fmt.Errorf("stopped after %d redirects", maxRedirects)
 		}
+
 		return nil
 	}
+
 	return &WebFetchTool{
 		maxChars: maxChars,
-		proxy:    proxy,
-		client:   client,
+
+		proxy: proxy,
+
+		client: client,
 	}, nil
 }
 
@@ -582,23 +726,30 @@ func (t *WebFetchTool) Description() string {
 func (t *WebFetchTool) Parameters() map[string]any {
 	return map[string]any{
 		"type": "object",
+
 		"properties": map[string]any{
 			"url": map[string]any{
-				"type":        "string",
+				"type": "string",
+
 				"description": "URL to fetch",
 			},
+
 			"maxChars": map[string]any{
-				"type":        "integer",
+				"type": "integer",
+
 				"description": "Maximum characters to extract",
-				"minimum":     100.0,
+
+				"minimum": 100.0,
 			},
 		},
+
 		"required": []string{"url"},
 	}
 }
 
 func (t *WebFetchTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
 	urlStr, ok := args["url"].(string)
+
 	if !ok {
 		return ErrorResult("url is required")
 	}
@@ -617,6 +768,7 @@ func (t *WebFetchTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	}
 
 	maxChars := t.maxChars
+
 	if mc, ok := args["maxChars"].(float64); ok {
 		if int(mc) > 100 {
 			maxChars = int(mc)
@@ -634,6 +786,7 @@ func (t *WebFetchTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	if err != nil {
 		return ErrorResult(fmt.Sprintf("request failed: %v", err))
 	}
+
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
@@ -646,71 +799,98 @@ func (t *WebFetchTool) Execute(ctx context.Context, args map[string]any) *ToolRe
 	var text, extractor string
 
 	bodyStr := string(body)
+
 	if strings.Contains(contentType, "application/json") {
 		var jsonData any
+
 		if err := json.Unmarshal(body, &jsonData); err == nil {
 			formatted, _ := json.MarshalIndent(jsonData, "", "  ")
+
 			text = string(formatted)
+
 			extractor = "json"
 		} else {
 			text = bodyStr
+
 			extractor = "raw"
 		}
 	} else if strings.Contains(contentType, "text/html") || len(body) > 0 &&
+
 		(strings.HasPrefix(bodyStr, "<!DOCTYPE") || strings.HasPrefix(strings.ToLower(bodyStr), "<html")) {
 		text = t.extractText(bodyStr)
+
 		extractor = "text"
 	} else {
 		text = bodyStr
+
 		extractor = "raw"
 	}
 
 	truncated := len(text) > maxChars
+
 	if truncated {
 		text = text[:maxChars]
 	}
 
 	result := map[string]any{
-		"url":       urlStr,
-		"status":    resp.StatusCode,
+		"url": urlStr,
+
+		"status": resp.StatusCode,
+
 		"extractor": extractor,
+
 		"truncated": truncated,
-		"length":    len(text),
-		"text":      text,
+
+		"length": len(text),
+
+		"text": text,
 	}
 
 	resultJSON, _ := json.MarshalIndent(result, "", "  ")
 
 	return &ToolResult{
 		ForLLM: fmt.Sprintf(
+
 			"Fetched %d bytes from %s (extractor: %s, truncated: %v)",
+
 			len(text),
+
 			urlStr,
+
 			extractor,
+
 			truncated,
 		),
+
 		ForUser: string(resultJSON),
 	}
 }
 
 func (t *WebFetchTool) extractText(htmlContent string) string {
 	result := reScript.ReplaceAllLiteralString(htmlContent, "")
+
 	result = reStyle.ReplaceAllLiteralString(result, "")
+
 	result = reTags.ReplaceAllLiteralString(result, "")
 
 	result = strings.TrimSpace(result)
 
 	result = reWhitespace.ReplaceAllString(result, " ")
+
 	result = reBlankLines.ReplaceAllString(result, "\n\n")
 
 	lines := strings.Split(result, "\n")
+
 	var sb strings.Builder
+
 	for _, line := range lines {
 		line = strings.TrimSpace(line)
+
 		if line != "" {
 			if sb.Len() > 0 {
 				sb.WriteByte('\n')
 			}
+
 			sb.WriteString(line)
 		}
 	}

pkg/tools/web_test.go

@@ -11,16 +11,22 @@ import (
 )
 
 // TestWebTool_WebFetch_Success verifies successful URL fetching
+
 func TestWebTool_WebFetch_Success(t *testing.T) {
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "text/html")
+
 		w.WriteHeader(http.StatusOK)
+
 		w.Write([]byte("<html><body><h1>Test Page</h1><p>Content here</p></body></html>"))
 	}))
+
 	defer server.Close()
 
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": server.URL,
 	}
@@ -28,35 +34,45 @@ func TestWebTool_WebFetch_Success(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForUser should contain the fetched content
+
 	if !strings.Contains(result.ForUser, "Test Page") {
 		t.Errorf("Expected ForUser to contain 'Test Page', got: %s", result.ForUser)
 	}
 
 	// ForLLM should contain summary
+
 	if !strings.Contains(result.ForLLM, "bytes") && !strings.Contains(result.ForLLM, "extractor") {
 		t.Errorf("Expected ForLLM to contain summary, got: %s", result.ForLLM)
 	}
 }
 
 // TestWebTool_WebFetch_JSON verifies JSON content handling
+
 func TestWebTool_WebFetch_JSON(t *testing.T) {
 	testData := map[string]string{"key": "value", "number": "123"}
+
 	expectedJSON, _ := json.MarshalIndent(testData, "", "  ")
 
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "application/json")
+
 		w.WriteHeader(http.StatusOK)
+
 		w.Write(expectedJSON)
 	}))
+
 	defer server.Close()
 
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": server.URL,
 	}
@@ -64,20 +80,25 @@ func TestWebTool_WebFetch_JSON(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForUser should contain formatted JSON
+
 	if !strings.Contains(result.ForUser, "key") && !strings.Contains(result.ForUser, "value") {
 		t.Errorf("Expected ForUser to contain JSON data, got: %s", result.ForUser)
 	}
 }
 
 // TestWebTool_WebFetch_InvalidURL verifies error handling for invalid URL
+
 func TestWebTool_WebFetch_InvalidURL(t *testing.T) {
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": "not-a-valid-url",
 	}
@@ -85,20 +106,25 @@ func TestWebTool_WebFetch_InvalidURL(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error for invalid URL")
 	}
 
 	// Should contain error message (either "invalid URL" or scheme error)
+
 	if !strings.Contains(result.ForLLM, "URL") && !strings.Contains(result.ForUser, "URL") {
 		t.Errorf("Expected error message for invalid URL, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestWebTool_WebFetch_UnsupportedScheme verifies error handling for non-http URLs
+
 func TestWebTool_WebFetch_UnsupportedScheme(t *testing.T) {
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": "ftp://example.com/file.txt",
 	}
@@ -106,48 +132,61 @@ func TestWebTool_WebFetch_UnsupportedScheme(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error for unsupported URL scheme")
 	}
 
 	// Should mention only http/https allowed
+
 	if !strings.Contains(result.ForLLM, "http/https") && !strings.Contains(result.ForUser, "http/https") {
 		t.Errorf("Expected scheme error message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestWebTool_WebFetch_MissingURL verifies error handling for missing URL
+
 func TestWebTool_WebFetch_MissingURL(t *testing.T) {
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when URL is missing")
 	}
 
 	// Should mention URL is required
+
 	if !strings.Contains(result.ForLLM, "url is required") && !strings.Contains(result.ForUser, "url is required") {
 		t.Errorf("Expected 'url is required' message, got ForLLM: %s", result.ForLLM)
 	}
 }
 
 // TestWebTool_WebFetch_Truncation verifies content truncation
+
 func TestWebTool_WebFetch_Truncation(t *testing.T) {
 	longContent := strings.Repeat("x", 20000)
 
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "text/plain")
+
 		w.WriteHeader(http.StatusOK)
+
 		w.Write([]byte(longContent))
 	}))
+
 	defer server.Close()
 
 	tool := NewWebFetchTool(1000) // Limit to 1000 chars
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": server.URL,
 	}
@@ -155,13 +194,17 @@ func TestWebTool_WebFetch_Truncation(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForUser should contain truncated content (not the full 20000 chars)
+
 	resultMap := make(map[string]any)
+
 	json.Unmarshal([]byte(result.ForUser), &resultMap)
+
 	if text, ok := resultMap["text"].(string); ok {
 		if len(text) > 1100 { // Allow some margin
 			t.Errorf("Expected content to be truncated to ~1000 chars, got: %d", len(text))
@@ -169,63 +212,79 @@ func TestWebTool_WebFetch_Truncation(t *testing.T) {
 	}
 
 	// Should be marked as truncated
+
 	if truncated, ok := resultMap["truncated"].(bool); !ok || !truncated {
 		t.Errorf("Expected 'truncated' to be true in result")
 	}
 }
 
 // TestWebTool_WebSearch_NoApiKey verifies that no tool is created when API key is missing
+
 func TestWebTool_WebSearch_NoApiKey(t *testing.T) {
 	tool, err := NewWebSearchTool(WebSearchToolOptions{BraveEnabled: true, BraveAPIKey: ""})
 	if err != nil {
 		t.Fatalf("Unexpected error: %v", err)
 	}
+
 	if tool != nil {
 		t.Errorf("Expected nil tool when Brave API key is empty")
 	}
 
 	// Also nil when nothing is enabled
+
 	tool, err = NewWebSearchTool(WebSearchToolOptions{})
 	if err != nil {
 		t.Fatalf("Unexpected error: %v", err)
 	}
+
 	if tool != nil {
 		t.Errorf("Expected nil tool when no provider is enabled")
 	}
 }
 
 // TestWebTool_WebSearch_MissingQuery verifies error handling for missing query
+
 func TestWebTool_WebSearch_MissingQuery(t *testing.T) {
 	tool, err := NewWebSearchTool(WebSearchToolOptions{BraveEnabled: true, BraveAPIKey: "test-key", BraveMaxResults: 5})
 	if err != nil {
 		t.Fatalf("Unexpected error: %v", err)
 	}
+
 	ctx := context.Background()
+
 	args := map[string]any{}
 
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error when query is missing")
 	}
 }
 
 // TestWebTool_WebFetch_HTMLExtraction verifies HTML text extraction
+
 func TestWebTool_WebFetch_HTMLExtraction(t *testing.T) {
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "text/html")
+
 		w.WriteHeader(http.StatusOK)
+
 		w.Write(
+
 			[]byte(
 				`<html><body><script>alert('test');</script><style>body{color:red;}</style><h1>Title</h1><p>Content</p></body></html>`,
 			),
 		)
 	}))
+
 	defer server.Close()
 
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": server.URL,
 	}
@@ -233,80 +292,105 @@ func TestWebTool_WebFetch_HTMLExtraction(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForUser should contain extracted text (without script/style tags)
+
 	if !strings.Contains(result.ForUser, "Title") && !strings.Contains(result.ForUser, "Content") {
 		t.Errorf("Expected ForUser to contain extracted text, got: %s", result.ForUser)
 	}
 
 	// Should NOT contain script or style tags
+
 	if strings.Contains(result.ForUser, "<script>") || strings.Contains(result.ForUser, "<style>") {
 		t.Errorf("Expected script/style tags to be removed, got: %s", result.ForUser)
 	}
 }
 
 // TestWebFetchTool_extractText verifies text extraction preserves newlines
+
 func TestWebFetchTool_extractText(t *testing.T) {
 	tool := &WebFetchTool{}
 
 	tests := []struct {
-		name     string
-		input    string
+		name string
+
+		input string
+
 		wantFunc func(t *testing.T, got string)
 	}{
 		{
-			name:  "preserves newlines between block elements",
+			name: "preserves newlines between block elements",
+
 			input: "<html><body><h1>Title</h1>\n<p>Paragraph 1</p>\n<p>Paragraph 2</p></body></html>",
+
 			wantFunc: func(t *testing.T, got string) {
 				lines := strings.Split(got, "\n")
+
 				if len(lines) < 2 {
 					t.Errorf("Expected multiple lines, got %d: %q", len(lines), got)
 				}
+
 				if !strings.Contains(got, "Title") || !strings.Contains(got, "Paragraph 1") ||
+
 					!strings.Contains(got, "Paragraph 2") {
 					t.Errorf("Missing expected text: %q", got)
 				}
 			},
 		},
+
 		{
-			name:  "removes script and style tags",
+			name: "removes script and style tags",
+
 			input: "<script>alert('x');</script><style>body{}</style><p>Keep this</p>",
+
 			wantFunc: func(t *testing.T, got string) {
 				if strings.Contains(got, "alert") || strings.Contains(got, "body{}") {
 					t.Errorf("Expected script/style content removed, got: %q", got)
 				}
+
 				if !strings.Contains(got, "Keep this") {
 					t.Errorf("Expected 'Keep this' to remain, got: %q", got)
 				}
 			},
 		},
+
 		{
-			name:  "collapses excessive blank lines",
+			name: "collapses excessive blank lines",
+
 			input: "<p>A</p>\n\n\n\n\n<p>B</p>",
+
 			wantFunc: func(t *testing.T, got string) {
 				if strings.Contains(got, "\n\n\n") {
 					t.Errorf("Expected excessive blank lines collapsed, got: %q", got)
 				}
 			},
 		},
+
 		{
-			name:  "collapses horizontal whitespace",
+			name: "collapses horizontal whitespace",
+
 			input: "<p>hello     world</p>",
+
 			wantFunc: func(t *testing.T, got string) {
 				if strings.Contains(got, "     ") {
 					t.Errorf("Expected spaces collapsed, got: %q", got)
 				}
+
 				if !strings.Contains(got, "hello world") {
 					t.Errorf("Expected 'hello world', got: %q", got)
 				}
 			},
 		},
+
 		{
-			name:  "empty input",
+			name: "empty input",
+
 			input: "",
+
 			wantFunc: func(t *testing.T, got string) {
 				if got != "" {
 					t.Errorf("Expected empty string, got: %q", got)
@@ -318,15 +402,19 @@ func TestWebFetchTool_extractText(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			got := tool.extractText(tt.input)
+
 			tt.wantFunc(t, got)
 		})
 	}
 }
 
 // TestWebTool_WebFetch_MissingDomain verifies error handling for URL without domain
+
 func TestWebTool_WebFetch_MissingDomain(t *testing.T) {
 	tool := NewWebFetchTool(50000)
+
 	ctx := context.Background()
+
 	args := map[string]any{
 		"url": "https://",
 	}
@@ -334,11 +422,13 @@ func TestWebTool_WebFetch_MissingDomain(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Should return error result
+
 	if !result.IsError {
 		t.Errorf("Expected error for URL without domain")
 	}
 
 	// Should mention missing domain
+
 	if !strings.Contains(result.ForLLM, "domain") && !strings.Contains(result.ForUser, "domain") {
 		t.Errorf("Expected domain error message, got ForLLM: %s", result.ForLLM)
 	}
@@ -349,14 +439,17 @@ func TestCreateHTTPClient_ProxyConfigured(t *testing.T) {
 	if err != nil {
 		t.Fatalf("createHTTPClient() error: %v", err)
 	}
+
 	if client.Timeout != 12*time.Second {
 		t.Fatalf("client.Timeout = %v, want %v", client.Timeout, 12*time.Second)
 	}
 
 	tr, ok := client.Transport.(*http.Transport)
+
 	if !ok {
 		t.Fatalf("client.Transport type = %T, want *http.Transport", client.Transport)
 	}
+
 	if tr.Proxy == nil {
 		t.Fatal("transport.Proxy is nil, want non-nil")
 	}
@@ -365,10 +458,12 @@ func TestCreateHTTPClient_ProxyConfigured(t *testing.T) {
 	if err != nil {
 		t.Fatalf("http.NewRequest() error: %v", err)
 	}
+
 	proxyURL, err := tr.Proxy(req)
 	if err != nil {
 		t.Fatalf("transport.Proxy(req) error: %v", err)
 	}
+
 	if proxyURL == nil || proxyURL.String() != "http://127.0.0.1:7890" {
 		t.Fatalf("proxy URL = %v, want %q", proxyURL, "http://127.0.0.1:7890")
 	}
@@ -376,6 +471,7 @@ func TestCreateHTTPClient_ProxyConfigured(t *testing.T) {
 
 func TestCreateHTTPClient_InvalidProxy(t *testing.T) {
 	_, err := createHTTPClient("://bad-proxy", 10*time.Second)
+
 	if err == nil {
 		t.Fatal("createHTTPClient() expected error for invalid proxy URL, got nil")
 	}
@@ -388,17 +484,21 @@ func TestCreateHTTPClient_Socks5ProxyConfigured(t *testing.T) {
 	}
 
 	tr, ok := client.Transport.(*http.Transport)
+
 	if !ok {
 		t.Fatalf("client.Transport type = %T, want *http.Transport", client.Transport)
 	}
+
 	req, err := http.NewRequest("GET", "https://example.com", nil)
 	if err != nil {
 		t.Fatalf("http.NewRequest() error: %v", err)
 	}
+
 	proxyURL, err := tr.Proxy(req)
 	if err != nil {
 		t.Fatalf("transport.Proxy(req) error: %v", err)
 	}
+
 	if proxyURL == nil || proxyURL.String() != "socks5://127.0.0.1:1080" {
 		t.Fatalf("proxy URL = %v, want %q", proxyURL, "socks5://127.0.0.1:1080")
 	}
@@ -406,9 +506,11 @@ func TestCreateHTTPClient_Socks5ProxyConfigured(t *testing.T) {
 
 func TestCreateHTTPClient_UnsupportedProxyScheme(t *testing.T) {
 	_, err := createHTTPClient("ftp://127.0.0.1:21", 10*time.Second)
+
 	if err == nil {
 		t.Fatal("createHTTPClient() expected error for unsupported scheme, got nil")
 	}
+
 	if !strings.Contains(err.Error(), "unsupported proxy scheme") {
 		t.Fatalf("error = %q, want to contain %q", err.Error(), "unsupported proxy scheme")
 	}
@@ -416,12 +518,19 @@ func TestCreateHTTPClient_UnsupportedProxyScheme(t *testing.T) {
 
 func TestCreateHTTPClient_ProxyFromEnvironmentWhenConfigEmpty(t *testing.T) {
 	t.Setenv("HTTP_PROXY", "http://127.0.0.1:8888")
+
 	t.Setenv("http_proxy", "http://127.0.0.1:8888")
+
 	t.Setenv("HTTPS_PROXY", "http://127.0.0.1:8888")
+
 	t.Setenv("https_proxy", "http://127.0.0.1:8888")
+
 	t.Setenv("ALL_PROXY", "")
+
 	t.Setenv("all_proxy", "")
+
 	t.Setenv("NO_PROXY", "")
+
 	t.Setenv("no_proxy", "")
 
 	client, err := createHTTPClient("", 10*time.Second)
@@ -430,9 +539,11 @@ func TestCreateHTTPClient_ProxyFromEnvironmentWhenConfigEmpty(t *testing.T) {
 	}
 
 	tr, ok := client.Transport.(*http.Transport)
+
 	if !ok {
 		t.Fatalf("client.Transport type = %T, want *http.Transport", client.Transport)
 	}
+
 	if tr.Proxy == nil {
 		t.Fatal("transport.Proxy is nil, want proxy function from environment")
 	}
@@ -441,6 +552,7 @@ func TestCreateHTTPClient_ProxyFromEnvironmentWhenConfigEmpty(t *testing.T) {
 	if err != nil {
 		t.Fatalf("http.NewRequest() error: %v", err)
 	}
+
 	if _, err := tr.Proxy(req); err != nil {
 		t.Fatalf("transport.Proxy(req) error: %v", err)
 	}
@@ -451,9 +563,11 @@ func TestNewWebFetchToolWithProxy(t *testing.T) {
 	if err != nil {
 		t.Fatalf("NewWebFetchToolWithProxy() error: %v", err)
 	}
+
 	if tool.maxChars != 1024 {
 		t.Fatalf("maxChars = %d, want %d", tool.maxChars, 1024)
 	}
+
 	if tool.proxy != "http://127.0.0.1:7890" {
 		t.Fatalf("proxy = %q, want %q", tool.proxy, "http://127.0.0.1:7890")
 	}
@@ -462,6 +576,7 @@ func TestNewWebFetchToolWithProxy(t *testing.T) {
 	if err != nil {
 		t.Fatalf("NewWebFetchToolWithProxy() error: %v", err)
 	}
+
 	if tool.maxChars != 50000 {
 		t.Fatalf("default maxChars = %d, want %d", tool.maxChars, 50000)
 	}
@@ -470,18 +585,24 @@ func TestNewWebFetchToolWithProxy(t *testing.T) {
 func TestNewWebSearchTool_PropagatesProxy(t *testing.T) {
 	t.Run("perplexity", func(t *testing.T) {
 		tool, err := NewWebSearchTool(WebSearchToolOptions{
-			PerplexityEnabled:    true,
-			PerplexityAPIKey:     "k",
+			PerplexityEnabled: true,
+
+			PerplexityAPIKey: "k",
+
 			PerplexityMaxResults: 3,
-			Proxy:                "http://127.0.0.1:7890",
+
+			Proxy: "http://127.0.0.1:7890",
 		})
 		if err != nil {
 			t.Fatalf("NewWebSearchTool() error: %v", err)
 		}
+
 		p, ok := tool.provider.(*PerplexitySearchProvider)
+
 		if !ok {
 			t.Fatalf("provider type = %T, want *PerplexitySearchProvider", tool.provider)
 		}
+
 		if p.proxy != "http://127.0.0.1:7890" {
 			t.Fatalf("provider proxy = %q, want %q", p.proxy, "http://127.0.0.1:7890")
 		}
@@ -489,18 +610,24 @@ func TestNewWebSearchTool_PropagatesProxy(t *testing.T) {
 
 	t.Run("brave", func(t *testing.T) {
 		tool, err := NewWebSearchTool(WebSearchToolOptions{
-			BraveEnabled:    true,
-			BraveAPIKey:     "k",
+			BraveEnabled: true,
+
+			BraveAPIKey: "k",
+
 			BraveMaxResults: 3,
-			Proxy:           "http://127.0.0.1:7890",
+
+			Proxy: "http://127.0.0.1:7890",
 		})
 		if err != nil {
 			t.Fatalf("NewWebSearchTool() error: %v", err)
 		}
+
 		p, ok := tool.provider.(*BraveSearchProvider)
+
 		if !ok {
 			t.Fatalf("provider type = %T, want *BraveSearchProvider", tool.provider)
 		}
+
 		if p.proxy != "http://127.0.0.1:7890" {
 			t.Fatalf("provider proxy = %q, want %q", p.proxy, "http://127.0.0.1:7890")
 		}
@@ -508,17 +635,22 @@ func TestNewWebSearchTool_PropagatesProxy(t *testing.T) {
 
 	t.Run("duckduckgo", func(t *testing.T) {
 		tool, err := NewWebSearchTool(WebSearchToolOptions{
-			DuckDuckGoEnabled:    true,
+			DuckDuckGoEnabled: true,
+
 			DuckDuckGoMaxResults: 3,
-			Proxy:                "http://127.0.0.1:7890",
+
+			Proxy: "http://127.0.0.1:7890",
 		})
 		if err != nil {
 			t.Fatalf("NewWebSearchTool() error: %v", err)
 		}
+
 		p, ok := tool.provider.(*DuckDuckGoSearchProvider)
+
 		if !ok {
 			t.Fatalf("provider type = %T, want *DuckDuckGoSearchProvider", tool.provider)
 		}
+
 		if p.proxy != "http://127.0.0.1:7890" {
 			t.Fatalf("provider proxy = %q, want %q", p.proxy, "http://127.0.0.1:7890")
 		}
@@ -526,50 +658,69 @@ func TestNewWebSearchTool_PropagatesProxy(t *testing.T) {
 }
 
 // TestWebTool_TavilySearch_Success verifies successful Tavily search
+
 func TestWebTool_TavilySearch_Success(t *testing.T) {
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		if r.Method != "POST" {
 			t.Errorf("Expected POST request, got %s", r.Method)
 		}
+
 		if r.Header.Get("Content-Type") != "application/json" {
 			t.Errorf("Expected Content-Type application/json, got %s", r.Header.Get("Content-Type"))
 		}
 
 		// Verify payload
+
 		var payload map[string]any
+
 		json.NewDecoder(r.Body).Decode(&payload)
+
 		if payload["api_key"] != "test-key" {
 			t.Errorf("Expected api_key test-key, got %v", payload["api_key"])
 		}
+
 		if payload["query"] != "test query" {
 			t.Errorf("Expected query 'test query', got %v", payload["query"])
 		}
 
 		// Return mock response
+
 		response := map[string]any{
 			"results": []map[string]any{
 				{
-					"title":   "Test Result 1",
-					"url":     "https://example.com/1",
+					"title": "Test Result 1",
+
+					"url": "https://example.com/1",
+
 					"content": "Content for result 1",
 				},
+
 				{
-					"title":   "Test Result 2",
-					"url":     "https://example.com/2",
+					"title": "Test Result 2",
+
+					"url": "https://example.com/2",
+
 					"content": "Content for result 2",
 				},
 			},
 		}
+
 		w.Header().Set("Content-Type", "application/json")
+
 		w.WriteHeader(http.StatusOK)
+
 		json.NewEncoder(w).Encode(response)
 	}))
+
 	defer server.Close()
 
 	tool, err := NewWebSearchTool(WebSearchToolOptions{
-		TavilyEnabled:    true,
-		TavilyAPIKey:     "test-key",
-		TavilyBaseURL:    server.URL,
+		TavilyEnabled: true,
+
+		TavilyAPIKey: "test-key",
+
+		TavilyBaseURL: server.URL,
+
 		TavilyMaxResults: 5,
 	})
 	if err != nil {
@@ -577,6 +728,7 @@ func TestWebTool_TavilySearch_Success(t *testing.T) {
 	}
 
 	ctx := context.Background()
+
 	args := map[string]any{
 		"query": "test query",
 	}
@@ -584,17 +736,21 @@ func TestWebTool_TavilySearch_Success(t *testing.T) {
 	result := tool.Execute(ctx, args)
 
 	// Success should not be an error
+
 	if result.IsError {
 		t.Errorf("Expected success, got IsError=true: %s", result.ForLLM)
 	}
 
 	// ForUser should contain result titles and URLs
+
 	if !strings.Contains(result.ForUser, "Test Result 1") ||
+
 		!strings.Contains(result.ForUser, "https://example.com/1") {
 		t.Errorf("Expected results in output, got: %s", result.ForUser)
 	}
 
 	// Should mention via Tavily
+
 	if !strings.Contains(result.ForUser, "via Tavily") {
 		t.Errorf("Expected 'via Tavily' in output, got: %s", result.ForUser)
 	}

pkg/tools/workspace_ctx.go

@@ -8,52 +8,72 @@ import (
 
 type (
 	workspaceOverrideKey struct{}
-	overrideFsKey        struct{}
+
+	overrideFsKey struct{}
 )
 
 // WithWorkspaceOverride returns a context carrying a workspace override path
+
 // and a pre-built sandboxFs for that workspace. Tools will resolve file
+
 // operations against this path instead of the original workspace.
+
 // The cached sandboxFs is reused across all resolveFS calls on the same context,
+
 // avoiding per-operation allocation.
+
 func WithWorkspaceOverride(ctx context.Context, workspace string) context.Context {
 	ctx = context.WithValue(ctx, workspaceOverrideKey{}, workspace)
+
 	ctx = context.WithValue(ctx, overrideFsKey{}, &sandboxFs{workspace: workspace})
+
 	return ctx
 }
 
 // WorkspaceOverrideFromCtx extracts the workspace override from context, or "".
+
 func WorkspaceOverrideFromCtx(ctx context.Context) string {
 	if v, ok := ctx.Value(workspaceOverrideKey{}).(string); ok {
 		return v
 	}
+
 	return ""
 }
 
 // resolveFS returns a fileSystem applying workspace override from context.
+
 // Paths under "memory/" are excluded (always use original workspace).
+
 // For sandboxFs: returns the cached override instance from context.
+
 // For hostFs (unrestricted): returns as-is.
+
 func resolveFS(ctx context.Context, fs fileSystem, path string) fileSystem {
 	override := WorkspaceOverrideFromCtx(ctx)
+
 	if override == "" {
 		return fs
 	}
 
 	// memory/ paths always use original workspace
+
 	if isMemoryPath(path) {
 		return fs
 	}
 
 	// Only sandboxFs supports workspace override
+
 	if sfs, ok := fs.(*sandboxFs); ok {
 		if sfs.workspace == override {
 			return fs
 		}
+
 		// Use cached sandboxFs from context
+
 		if cached, ok := ctx.Value(overrideFsKey{}).(*sandboxFs); ok {
 			return cached
 		}
+
 		return &sandboxFs{workspace: override}
 	}
 
@@ -61,16 +81,20 @@ func resolveFS(ctx context.Context, fs fileSystem, path string) fileSystem {
 }
 
 // isMemoryPath returns true for paths under the memory/ directory.
+
 // Matches: "memory/MEMORY.md", "memory", "/workspace/memory/notes.md"
+
 func isMemoryPath(path string) bool {
 	p := filepath.ToSlash(filepath.Clean(path))
 
 	// Relative path starting with memory/
+
 	if strings.HasPrefix(p, "memory/") || p == "memory" {
 		return true
 	}
 
 	// Absolute path containing /memory/ or ending with /memory
+
 	if strings.Contains(p, "/memory/") || strings.HasSuffix(p, "/memory") {
 		return true
 	}

todo/TASKS-2.md

@@ -1,4 +1,4 @@
-# TASKS-2: Subagent Orchestration (Container Model)
+# TASKS-2: Subagent Orchestration (Container Model) ✅ 実装済み
 
 ## TASKS-1 反映メモ (2026-03-05)
 
@@ -78,15 +78,15 @@ TASKS-2 の下地はかなり実装済み。以下を前提として差分のみ
 | **Async Callback** | `pkg/agent/loop.go` `processRequest` | spawn 完了 → MessageBus → conductor に結果注入 |
 | **Orchestration Nudge** | `pkg/agent/loop.go` `buildOrchReminder()` | plan 実行中に spawn/subagent 使用を促すリマインダ |
 
-### 未実装 ❌ → TASKS-2 スコープ
+### 実装完了 ✅ (2026-03-05)
 
 | # | 要素 | 概要 |
 |---|---|---|
-| 1 | **ContainerMessage channel** | subagent→conductor の question/status/result 双方向通信 |
-| 2 | **Escalation chain** | subagent question → conductor 回答 or → human escalate |
-| 3 | **Deliberate Plan Mode** | coder/worker/coordinator の clarifying→review→executing 状態遷移 |
-| 4 | **SubagentEnvironment injection** | MEMORY.md からの自動コンテキスト注入 |
-| 5 | **MEMORY.md Orchestration section** | conductor guidance に delegated/findings/decisions 追記 |
+| 1 | **ContainerMessage channel** | ✅ `ContainerMessage` + `inCh`/`outCh` on `SubagentTask` |
+| 2 | **Escalation chain** | ✅ `ask_conductor` / `answer_subagent` tools + conductor question injection |
+| 3 | **Deliberate Plan Mode** | ✅ `SubagentPlanState` + `runDeliberateTask()` (clarifying→review→executing) |
+| 4 | **SubagentEnvironment injection** | ✅ `extractPlanContext()` + `buildSubagentSystemPrompt()` |
+| 5 | **MEMORY.md Orchestration section** | ✅ `orchestrationGuidance` 拡張 (Delegated/Findings/Decisions) |
 
 ---