legion-cc 0.20.1 → 0.21.0

package/agents/architect.md

@@ -97,6 +97,7 @@ When your work is complete, execute these steps IN ORDER:
 
 1. **Write report to task**: `TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your full structured report>")`
 2. **Notify teammates** (if needed): `SendMessage(type: "message", recipient: "<teammate-name>", content: "<report or summary>", summary: "<brief>")`
+3. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after steps 1-2.
 
 Your task ID and team name are provided in your spawn prompt. If not provided, check `TaskList` to find your assigned task.
 

package/agents/code-analyst.md

@@ -92,6 +92,7 @@ When your work is complete, execute these steps IN ORDER:
 
 1. **Write report to task**: `TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your full structured report>")`
 2. **Notify teammates** (if needed): `SendMessage(type: "message", recipient: "<teammate-name>", content: "<report or summary>", summary: "<brief>")`
+3. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after steps 1-2.
 
 Your task ID and team name are provided in your spawn prompt. If not provided, check `TaskList` to find your assigned task.
 

package/agents/context-curator.md

@@ -99,6 +99,7 @@ When your work is complete, execute these steps IN ORDER:
 
 1. **Write report to task**: `TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your full context document>")`
 2. **Notify teammates** (if needed): In quick pipeline, send Bundle Section 3 to architect via `SendMessage`. In cycle pipeline, TaskUpdate is sufficient.
+3. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after steps 1-2.
 
 Your task ID and team name are provided in your spawn prompt. If not provided, check `TaskList` to find your assigned task.
 

package/agents/implementer.md

@@ -113,6 +113,7 @@ When your work is complete, execute these steps IN ORDER:
 
 1. **Write report to task**: `TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your full implementation report>")`
 2. **Notify teammates** (if needed): `SendMessage(type: "message", recipient: "<teammate-name>", content: "<summary>", summary: "<brief>")`
+3. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after steps 1-2.
 
 Your task ID and team name are provided in your spawn prompt. If not provided, check `TaskList` to find your assigned task.
 

package/agents/legion-orchestrator.md

@@ -122,10 +122,10 @@ Cap at 10.
 3. Spawn 2 teammates **in parallel** — include task IDs in prompts:
    ```
    Agent(team_name: ..., name: "context-curator", subagent_type: "context-curator",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-1-id>\n\nCollect project context. Write Bundle Section 3 following the ## Bundle Section 3: Project Context schema (Identity, Rules & Conventions, Build/Run/Test, Code Style, Config & Environment, Key Constraints). CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\n1. TaskUpdate(taskId: '<task-1-id>', status: 'completed', description: '<Bundle Section 3>')\n2. SendMessage(type: 'message', recipient: 'architect', content: '<Bundle Section 3>', summary: 'Bundle Section 3 ready')\nBoth steps REQUIRED.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-1-id>\n\nCollect project context. Write Bundle Section 3 following the ## Bundle Section 3: Project Context schema (Identity, Rules & Conventions, Build/Run/Test, Code Style, Config & Environment, Key Constraints). CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\n1. TaskUpdate(taskId: '<task-1-id>', status: 'completed', description: '<Bundle Section 3>')\n2. SendMessage(type: 'message', recipient: 'architect', content: '<Bundle Section 3>', summary: 'Bundle Section 3 ready')\nBoth steps REQUIRED.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
    Agent(team_name: ..., name: "architect", subagent_type: "architect",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-2-id>\n\nWait for Bundle Section 3 from context-curator via SendMessage. Then write Bundle Sections 1+2 combined (Architecture + Code Analysis, budget: 300-600 lines) following the schemas:\n- ## Bundle Sections 1+2: Architecture + Code Analysis\n- Subsections: 1.1 Approach, 1.2 Change Boundaries, 1.3 Architecture Decisions, 2.1 Entry Points, 2.2 Files to Change, 2.3 Risks\n\nAND produce a lightweight Execution Pack (Plan + Diff + Acceptance Criteria + Rollback). Write both as your task output via TaskUpdate.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-2-id>', status: 'completed', description: '<Bundle Sections 1+2 + Execution Pack>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-2-id>\n\nWait for Bundle Section 3 from context-curator via SendMessage. Then write Bundle Sections 1+2 combined (Architecture + Code Analysis, budget: 300-600 lines) following the schemas:\n- ## Bundle Sections 1+2: Architecture + Code Analysis\n- Subsections: 1.1 Approach, 1.2 Change Boundaries, 1.3 Architecture Decisions, 2.1 Entry Points, 2.2 Files to Change, 2.3 Risks\n\nAND produce a lightweight Execution Pack (Plan + Diff + Acceptance Criteria + Rollback). Write both as your task output via TaskUpdate.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-2-id>', status: 'completed', description: '<Bundle Sections 1+2 + Execution Pack>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
 4. Context Curator collects context → writes Bundle Section 3 to task via TaskUpdate → sends to Architect via SendMessage
 5. Architect receives Bundle Section 3 → writes Bundle Sections 1+2 + Execution Pack to task via TaskUpdate
@@ -134,9 +134,9 @@ Cap at 10.
 8. Spawn Implementer with full Execution Pack text from TaskGet(task-2-id) and task 3's ID:
    ```
    Agent(team_name: ..., name: "implementer", subagent_type: "implementer",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-3-id>\n\n<full text from TaskGet(task-2-id).description>\n\nThe above contains Bundle Sections 1+2 and an Execution Pack. Execute the Plan from the Execution Pack. Use Bundle sections for architectural context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-3-id>', status: 'completed', description: '<implementation report>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-3-id>\n\n<full text from TaskGet(task-2-id).description>\n\nThe above contains Bundle Sections 1+2 and an Execution Pack. Execute the Plan from the Execution Pack. Use Bundle sections for architectural context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-3-id>', status: 'completed', description: '<implementation report>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
-9. Poll TaskList until task 3 completed → Read result from TaskGet(task-3-id) → Shutdown teammates → TeamDelete
+9. Poll TaskList until task 3 completed → Read result from TaskGet(task-3-id) → Call `TeamDelete`. Optionally send `shutdown_request` as best-effort, but do NOT wait for `shutdown_response`.
 
 ## Cycle Pipeline (score 7-10)
 
@@ -154,13 +154,13 @@ Cap at 10.
 3. Spawn 3 teammates **in parallel** — each prompt MUST include task, team name, task ID, and Bundle Section instruction:
    ```
    Agent(team_name: ..., name: "architect", subagent_type: "architect",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-1-id>\n\nAnalyze architecture for this task. Write Bundle Section 1 following the ## Bundle Section 1: Architecture schema. CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-1-id>', status: 'completed', description: '<Bundle Section 1>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-1-id>\n\nAnalyze architecture for this task. Write Bundle Section 1 following the ## Bundle Section 1: Architecture schema. CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-1-id>', status: 'completed', description: '<Bundle Section 1>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
    Agent(team_name: ..., name: "code-analyst", subagent_type: "code-analyst",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-2-id>\n\nAnalyze codebase for this task. Write Bundle Section 2 following the ## Bundle Section 2: Code Analysis schema. CRITICAL: MUST NOT exceed 600 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-2-id>', status: 'completed', description: '<Bundle Section 2>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-2-id>\n\nAnalyze codebase for this task. Write Bundle Section 2 following the ## Bundle Section 2: Code Analysis schema. CRITICAL: MUST NOT exceed 600 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-2-id>', status: 'completed', description: '<Bundle Section 2>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
    Agent(team_name: ..., name: "context-curator", subagent_type: "context-curator",
-        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-3-id>\n\nCollect project context. Write Bundle Section 3 following the ## Bundle Section 3: Project Context schema. CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-3-id>', status: 'completed', description: '<Bundle Section 3>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-3-id>\n\nCollect project context. Write Bundle Section 3 following the ## Bundle Section 3: Project Context schema. CRITICAL: MUST NOT exceed 400 lines.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-3-id>', status: 'completed', description: '<Bundle Section 3>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
 4. Wait for all 3 to complete — poll `TaskList` until tasks 1, 2, 3 are all `status: "completed"`
 
@@ -170,12 +170,12 @@ Cap at 10.
    Spawn Planner:
    ```
    Agent(team_name: ..., name: "planner", subagent_type: "planner",
-        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-4-id>\n\n## Task\n<the user's original request>\n\n## Phase 1 Bundle Section Task IDs\nRead each Bundle Section using TaskGet(taskId) — the section is in the task's `description` field. Each section starts with `## Bundle Section N:`.\n- **Architect (Bundle Section 1)**: TaskGet('<task-1-id>')\n- **Code Analyst (Bundle Section 2)**: TaskGet('<task-2-id>')\n- **Context Curator (Bundle Section 3)**: TaskGet('<task-3-id>')\n\nRead all three Bundle Sections, assemble the Bundle, then produce an Execution Pack with sub-tasks, pseudo-diffs, acceptance criteria, risks, and rollback. Use ToolSearch('+legion-sequential-thinking') and ToolSearch('+legion-memory') before plan construction. Read Infrastructure Notes sections (§1.6, §2.7) from Bundle Sections. Query mcp__legion-memory__read_graph for cross-session context. If legion-memory MCP is unavailable, proceed without. Include User Questions section (≤3) only if critical clarifications are needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-4-id>', status: 'completed', description: '<your full Execution Pack>')\nIf you have user questions, also: SendMessage(type: 'message', recipient: 'legion-orchestrator', content: '<questions>', summary: 'Clarifying questions for user')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-4-id>\n\n## Task\n<the user's original request>\n\n## Phase 1 Bundle Section Task IDs\nRead each Bundle Section using TaskGet(taskId) — the section is in the task's `description` field. Each section starts with `## Bundle Section N:`.\n- **Architect (Bundle Section 1)**: TaskGet('<task-1-id>')\n- **Code Analyst (Bundle Section 2)**: TaskGet('<task-2-id>')\n- **Context Curator (Bundle Section 3)**: TaskGet('<task-3-id>')\n\nRead all three Bundle Sections, assemble the Bundle, then produce an Execution Pack with sub-tasks, pseudo-diffs, acceptance criteria, risks, and rollback. Use ToolSearch('+legion-sequential-thinking') and ToolSearch('+legion-memory') before plan construction. Read Infrastructure Notes sections (§1.6, §2.7) from Bundle Sections. Query mcp__legion-memory__read_graph for cross-session context. If legion-memory MCP is unavailable, proceed without. Include User Questions section (≤3) only if critical clarifications are needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-4-id>', status: 'completed', description: '<your full Execution Pack>')\nIf you have user questions, also: SendMessage(type: 'message', recipient: 'legion-orchestrator', content: '<questions>', summary: 'Clarifying questions for user')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
    Spawn Memory (fire-and-forget):
    ```
-   Agent(team_name: ..., name: "memory", subagent_type: "general-purpose",
-        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-6-id>\n\n## Phase 1 Bundle Section Task IDs\nRead Bundle Sections 1 and 2 using TaskGet(taskId) — the section is in the task's `description` field.\n- **Architect (Bundle Section 1)**: TaskGet('<task-1-id>')\n- **Code Analyst (Bundle Section 2)**: TaskGet('<task-2-id>')\n\nLoad MCP tools first: ToolSearch(\"+legion-memory\"). Then extract key findings and store them in project memory via MCP legion-memory server. Pay special attention to Infrastructure Notes (§1.6, §2.7) — store these as separate named entities using mcp__legion-memory__create_entities and mcp__legion-memory__add_observations. If MCP is unavailable, note the error and complete anyway.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-6-id>', status: 'completed', description: '<summary of stored memories>')\nDo NOT go idle without calling TaskUpdate.")
+   Agent(team_name: ..., name: "memory", subagent_type: "general-purpose", model: "haiku",
+        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-6-id>\n\n## Phase 1 Bundle Section Task IDs\nRead Bundle Sections 1 and 2 using TaskGet(taskId) — the section is in the task's `description` field.\n- **Architect (Bundle Section 1)**: TaskGet('<task-1-id>')\n- **Code Analyst (Bundle Section 2)**: TaskGet('<task-2-id>')\n\nLoad MCP tools first: ToolSearch(\"+legion-memory\"). Then extract key findings and store them in project memory via MCP legion-memory server. Pay special attention to Infrastructure Notes (§1.6, §2.7) — store these as separate named entities using mcp__legion-memory__create_entities and mcp__legion-memory__add_observations. If MCP is unavailable, note the error and complete anyway.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-6-id>', status: 'completed', description: '<summary of stored memories>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
 6. Wait for plan — poll `TaskList` until tasks 4 and 6 are `status: "completed"`
 
@@ -195,10 +195,10 @@ Cap at 10.
 9. Spawn Implementer with full Execution Pack text from TaskGet(task-4-id) and task 5's ID:
    ```
    Agent(team_name: ..., name: "implementer", subagent_type: "implementer",
-        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-5-id>\n\n<full text from TaskGet(task-4-id).description>\n\nThe Execution Pack above contains Plan, Diff, Acceptance Criteria, Risks, and Rollback.\nExecute the Plan. Use Bundle References (TaskGet) for deeper context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-5-id>', status: 'completed', description: '<your implementation report>')\nDo NOT go idle without calling TaskUpdate.")
+        prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-5-id>\n\n<full text from TaskGet(task-4-id).description>\n\nThe Execution Pack above contains Plan, Diff, Acceptance Criteria, Risks, and Rollback.\nExecute the Plan. Use Bundle References (TaskGet) for deeper context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-5-id>', status: 'completed', description: '<your implementation report>')\nDo NOT go idle without calling TaskUpdate.\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
    ```
 10. Wait for implementation — poll `TaskList` until task 5 is `status: "completed"`
-11. Read final report from `TaskGet(task-5-id).description` → Shutdown all teammates → TeamDelete
+11. Read final report from `TaskGet(task-5-id).description` → Call `TeamDelete`. Optionally send `shutdown_request` as best-effort, but do NOT wait for `shutdown_response`.
 
 ## Partial Pipeline (score 4-6)
 
@@ -262,7 +262,7 @@ Agent(
 3. **Create team before spawning** — TeamCreate must come before Agent calls
 4. **Pass task IDs** — give Planner the Phase 1 task IDs to read reports directly
 5. **Monitor progress** — check TaskList between phases
-6. **Clean shutdown** — SendMessage(type: "shutdown_request") to each teammate, then TeamDelete
+6. **Clean shutdown** — Call `TeamDelete` immediately. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 7. **Report to user** — summarize what was done after pipeline completes
 
 ### Agent Tool Reference
@@ -290,7 +290,7 @@ Spawn subagents using the **Agent** tool:
 **Example:**
 ```json
 {
-  "prompt": "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-id>\n\nAnalyze architecture, approach, boundaries, prior art.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-id>', status: 'completed', description: '<your full report>')",
+  "prompt": "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-id>\n\nAnalyze architecture, approach, boundaries, prior art.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-id>', status: 'completed', description: '<your full report>')\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).",
   "description": "Architectural analysis",
   "subagent_type": "architect",
   "model": "opus"
@@ -356,7 +356,7 @@ executionPack = TaskGet(Task2.id).description
 
 // Phase 5: Spawn Implementer with full Execution Pack
 Agent(name: "implementer", subagent_type: "implementer",
-  prompt: "## Assignment\n**Your task ID**: " + Task3.id + "\n\n" + executionPack + "\n\nExecute the Plan from the Execution Pack. Use Bundle sections for context.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + Task3.id + "', ...)")
+  prompt: "## Assignment\n**Your task ID**: " + Task3.id + "\n\n" + executionPack + "\n\nExecute the Plan from the Execution Pack. Use Bundle sections for context.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + Task3.id + "', ...)\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
 // Phase 6: Poll until task 3 completed, read final report
 result = TaskGet(Task3.id).description
@@ -384,10 +384,10 @@ Agent(..., name: "context-curator", subagent_type: "context-curator",
 while not all [t1, t2, t3] completed: check TaskList()
 
 // Spawn Planner AND Memory in parallel (Phase 2)
-Agent(..., name: "planner", prompt: "## Assignment\n**Your task ID**: " + t4.id + "\n\n## Task\n<the user's original request>\n\n## Phase 1 Bundle Section Task IDs\nRead each Bundle Section using TaskGet(taskId). Each starts with ## Bundle Section N:.\n- Architect (§1): TaskGet('" + t1.id + "')\n- Code Analyst (§2): TaskGet('" + t2.id + "')\n- Context Curator (§3): TaskGet('" + t3.id + "')\n\nAssemble Bundle, produce Execution Pack with optional user questions (≤3).\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t4.id + "', ...)")
+Agent(..., name: "planner", prompt: "## Assignment\n**Your task ID**: " + t4.id + "\n\n## Task\n<the user's original request>\n\n## Phase 1 Bundle Section Task IDs\nRead each Bundle Section using TaskGet(taskId). Each starts with ## Bundle Section N:.\n- Architect (§1): TaskGet('" + t1.id + "')\n- Code Analyst (§2): TaskGet('" + t2.id + "')\n- Context Curator (§3): TaskGet('" + t3.id + "')\n\nAssemble Bundle, produce Execution Pack with optional user questions (≤3).\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t4.id + "', ...)\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
-Agent(..., name: "memory", subagent_type: "general-purpose",
-  prompt: "## Assignment\n**Your task ID**: " + t6.id + "\n\n## Phase 1 Bundle Section Task IDs\n- Architect (§1): TaskGet('" + t1.id + "')\n- Code Analyst (§2): TaskGet('" + t2.id + "')\n\nExtract key findings and store in project memory via MCP legion-memory server.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t6.id + "', ...)")
+Agent(..., name: "memory", subagent_type: "general-purpose", model: "haiku",
+  prompt: "## Assignment\n**Your task ID**: " + t6.id + "\n\n## Phase 1 Bundle Section Task IDs\n- Architect (§1): TaskGet('" + t1.id + "')\n- Code Analyst (§2): TaskGet('" + t2.id + "')\n\nExtract key findings and store in project memory via MCP legion-memory server.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t6.id + "', ...)\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 
 // Phase 2.5: Check for Planner questions → AskUserQuestion if present
 
@@ -396,7 +396,7 @@ executionPack = TaskGet(t4.id).description
 // Extract ## Plan section, present via AskUserQuestion
 
 // Spawn Implementer with full Execution Pack
-Agent(..., name: "implementer", prompt: "## Assignment\n**Your task ID**: " + t5.id + "\n\n" + executionPack + "\n\nExecute the Plan. Use Bundle References (TaskGet) for deeper context if needed.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t5.id + "', ...)")
+Agent(..., name: "implementer", prompt: "## Assignment\n**Your task ID**: " + t5.id + "\n\n" + executionPack + "\n\nExecute the Plan. Use Bundle References (TaskGet) for deeper context if needed.\n\n## End of Work Protocol\nTaskUpdate(taskId: '" + t5.id + "', ...)\nIf you receive a shutdown_request message, respond: SendMessage(type: 'shutdown_response', request_id: '<requestId>', approve: true).")
 ```
 
 ## Behavioral Guidelines
@@ -426,5 +426,5 @@ Agent(..., name: "implementer", prompt: "## Assignment\n**Your task ID**: " + t5
 4. **Read reports via TaskGet** — after a task is completed, call `TaskGet(taskId)` to read the report from the `description` field
 5. **Pass task IDs to Planner** — give Planner the Phase 1 task IDs so it can read reports directly via TaskGet
 6. **Create new tasks for each phase** — don't reuse tasks; Phase 2 is a new task (Planner), Phase 3 is another new task (Implementer)
-7. **Shutdown gracefully** — after pipeline completes, send `shutdown_request` to each agent before `TeamDelete`
+7. **Shutdown gracefully** — Call `TeamDelete` immediately after pipeline completes. Optionally send `shutdown_request` as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 8. **Don't rely on stdout or SendMessage for reports** — reports are delivered via TaskUpdate(description). TaskGet reads them.

package/agents/memory.md

@@ -80,8 +80,9 @@ If MCP servers are available in the project, you MUST:
 
 ## End of Work Protocol (MANDATORY)
 
-TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<summary of what was stored in memory, or error if MCP unavailable>")
-Do NOT go idle without calling TaskUpdate.
+1. TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<summary of what was stored in memory, or error if MCP unavailable>")
+2. Do NOT go idle without calling TaskUpdate.
+3. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after step 1.
 
 ## Rules
 

package/agents/planner.md

@@ -114,6 +114,7 @@ When your work is complete, execute these steps IN ORDER:
 1. **Write Execution Pack to task**: `TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your full Execution Pack>")`
 2. **Send questions to orchestrator** (if any): If your Execution Pack contains a `## User Questions` section with 1+ questions, also: `SendMessage(type: "message", recipient: "legion-orchestrator", content: "<questions text>", summary: "Clarifying questions for user")`
 3. **Notify teammates** (if needed): `SendMessage(type: "message", recipient: "<teammate-name>", content: "<summary>", summary: "<brief>")`
+4. **Respond to shutdown** (if received): If a `shutdown_request` message arrives, call `SendMessage(type: "shutdown_response", request_id: "<requestId>", approve: true)`. Do NOT wait actively for it — you may stop after steps 1-3.
 
 Your task ID and team name are provided in your spawn prompt. If not provided, check `TaskList` to find your assigned task.
 

package/bin/install.js

@@ -88,13 +88,21 @@ const COMMAND_FILES = [
   'update.md',
 ];
 
+// Cross-platform helpers
+const isWindows = process.platform === 'win32';
+function npmCmd(cmd) { return isWindows ? cmd + '.cmd' : cmd; }
+function makeHookCmd(hookPath) {
+  const p = isWindows ? hookPath.replace(/\\/g, '/') : hookPath;
+  return `node "${p}"`;
+}
+
 // Claude Code >=2.1 hook format: { matcher: {}, hooks: [{ type, command }] }
 // Statusline is a top-level `statusLine` setting, not a hook event.
-const STATUSLINE_CMD = `node "${path.join(hooksTarget, 'legion-statusline.js')}"`;
-const CONTEXT_MONITOR_CMD = `node "${path.join(hooksTarget, 'legion-context-monitor.js')}"`;
-const TEAMMATE_IDLE_CMD = `node "${path.join(hooksTarget, 'legion-teammate-idle.js')}"`;
-const TASK_COMPLETED_CMD = `node "${path.join(hooksTarget, 'legion-task-completed.js')}"`;
-const MCP_ADVISOR_CMD = `node "${path.join(hooksTarget, 'legion-mcp-advisor.js')}"`;
+const STATUSLINE_CMD = makeHookCmd(path.join(hooksTarget, 'legion-statusline.js'));
+const CONTEXT_MONITOR_CMD = makeHookCmd(path.join(hooksTarget, 'legion-context-monitor.js'));
+const TEAMMATE_IDLE_CMD = makeHookCmd(path.join(hooksTarget, 'legion-teammate-idle.js'));
+const TASK_COMPLETED_CMD = makeHookCmd(path.join(hooksTarget, 'legion-task-completed.js'));
+const MCP_ADVISOR_CMD = makeHookCmd(path.join(hooksTarget, 'legion-mcp-advisor.js'));
 
 // Marker prefix for identifying Legion hook entries in settings.json.
 // All Legion hook commands start with this path prefix.
@@ -105,14 +113,14 @@ const LEGION_MCP_PREFIX = 'legion-';
 
 const LEGION_MCP_SERVERS = {
   'legion-memory': {
-    command: 'npx',
+    command: npmCmd('npx'),
     args: ['-y', '@modelcontextprotocol/server-memory'],
     env: {
       MEMORY_FILE_PATH: path.join(os.homedir(), '.legion', 'memory.jsonl'),
     },
   },
   'legion-sequential-thinking': {
-    command: 'npx',
+    command: npmCmd('npx'),
     args: ['-y', '@modelcontextprotocol/server-sequential-thinking'],
   },
 };
@@ -169,7 +177,7 @@ function copyFile(src, dest, options) {
   const tmpPath = dest + '.tmp.' + process.pid;
   try {
     fs.copyFileSync(src, tmpPath);
-    fs.chmodSync(tmpPath, mode);
+    if (!isWindows) fs.chmodSync(tmpPath, mode);
     fs.renameSync(tmpPath, dest);
     log(`copied ${path.basename(dest)}`);
   } catch (err) {
@@ -387,13 +395,15 @@ function upsertHookEntry(settings, eventName, entry) {
     settings.hooks[eventName] = [];
   }
 
-  const existingIdx = settings.hooks[eventName].findIndex((e) =>
-    e.hooks && e.hooks.some((h) => h.command && h.command.includes(LEGION_HOOK_MARKER))
-  );
-
   // Extract hook filename from command (e.g. 'node "/path/to/legion-foo.js"' → 'legion-foo.js')
   const hookFile = path.basename(entry.hooks[0].command.replace(/"/g, ''));
 
+  const existingIdx = settings.hooks[eventName].findIndex((e) =>
+    e.hooks && e.hooks.some((h) => h.command &&
+      h.command.includes(LEGION_HOOK_MARKER) &&
+      path.basename(h.command.replace(/"/g, '')) === hookFile)
+  );
+
   if (existingIdx === -1) {
     settings.hooks[eventName].push(entry);
     log(`registered hook: ${eventName} → ${hookFile}`);
@@ -472,15 +482,13 @@ function registerHooks() {
     log('updated hook: PostToolUse → legion-context-monitor.js');
   }
 
-  // ── TeammateIdle hook ────────────────────────────────────────────────────
+  // ── TeammateIdle hook (no matcher support — fires on every occurrence) ──
   upsertHookEntry(settings, 'TeammateIdle', {
-    matcher: '*',
     hooks: [{ type: 'command', command: TEAMMATE_IDLE_CMD }],
   });
 
-  // ── TaskCompleted hook ───────────────────────────────────────────────────
+  // ── TaskCompleted hook (no matcher support — fires on every occurrence) ─
   upsertHookEntry(settings, 'TaskCompleted', {
-    matcher: '*',
     hooks: [{ type: 'command', command: TASK_COMPLETED_CMD }],
   });
 

package/commands/legion/auto.md

@@ -78,7 +78,7 @@ Announce your score and selected pipeline to the user before proceeding.
    Agent(team_name: ..., name: "implementer", subagent_type: "implementer",
         prompt: "## Assignment\n**Task**: <task>\n**Team**: <team-name>\n**Your task ID**: <task-3-id>\n\n<full text from TaskGet(task-2-id).description>\n\nThe above contains Bundle Sections 1+2 and an Execution Pack. Execute the Plan from the Execution Pack. Use Bundle sections for architectural context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-3-id>', status: 'completed', description: '<implementation report>')\nDo NOT go idle without calling TaskUpdate.")
    ```
-9. Poll TaskList until task 3 completed → Read result from TaskGet(task-3-id) → Shutdown teammates → TeamDelete
+9. Poll TaskList until task 3 completed → Read result from TaskGet(task-3-id) → Call `TeamDelete`. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 
 ---
 
@@ -153,7 +153,7 @@ Escalation: shutdown quick team → create cycle team → run full Cycle Pipelin
         prompt: "## Assignment\n**Team**: <team-name>\n**Your task ID**: <task-5-id>\n\n<full text from TaskGet(task-4-id).description>\n\nThe Execution Pack above contains Plan, Diff, Acceptance Criteria, Risks, and Rollback.\nExecute the Plan. Use Bundle References (TaskGet) for deeper context if needed.\n\n## End of Work Protocol (MANDATORY)\nTaskUpdate(taskId: '<task-5-id>', status: 'completed', description: '<your implementation report>')\nDo NOT go idle without calling TaskUpdate.")
    ```
 10. Wait for implementation — poll `TaskList` until task 5 is `status: "completed"`
-11. Read final report from `TaskGet(task-5-id).description` → Shutdown all teammates → TeamDelete
+11. Read final report from `TaskGet(task-5-id).description` → Call `TeamDelete`. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 
 ---
 

package/commands/legion/codebase.md

@@ -87,7 +87,7 @@ $ARGUMENTS
    - `architect-brief.md`
    - `code-analyst-brief.md`
    - `context-curator-brief.md`
-11. **Clean shutdown** — shutdown all teammates, TeamDelete
+11. **Clean shutdown** — Call `TeamDelete`. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 12. **Report** — summarize what was generated: 5 codebase analysis files in `.legion/codebase/`, 3 agent briefs in `.legion/codebase/agents/`, and key findings stored in MCP memory
 
 ## Rules

package/commands/legion/cycle.md

@@ -111,7 +111,7 @@ $ARGUMENTS
    TaskUpdate(taskId: "<your-task-id>", status: "completed", description: "<your implementation report>")
    ```
 10. **Wait** for implementation — poll `TaskList` until task 5 is `status: "completed"`
-11. **Clean shutdown** — shutdown all teammates, TeamDelete
+11. **Clean shutdown** — Call `TeamDelete`. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 12. **Report** — read final report from `TaskGet` on task 5, summarize what was done to the user
 
 ## Rules

package/commands/legion/quick.md

@@ -82,7 +82,7 @@ $ARGUMENTS
    ```
 
 8. **Monitor** — poll `TaskList` until Task 3 is `status: "completed"`
-9. **Clean shutdown** — shutdown teammates, TeamDelete
+9. **Clean shutdown** — Call `TeamDelete`. Optionally send `shutdown_request` to teammates as best-effort, but do NOT wait for `shutdown_response`. Idle agents do not process incoming messages.
 10. **Report** — read final report from `TaskGet` on Task 3, summarize what was done to the user
 
 ## Rules

package/hooks/legion-context-monitor.js

@@ -93,6 +93,13 @@ function systemNotify(title, message) {
     } else if (platform === 'linux') {
       spawnSync('notify-send', ['--urgency=critical', title, message],
         { timeout: 3000, stdio: 'ignore' });
+    } else if (platform === 'win32') {
+      const safeMsg = message.replace(/'/g, "''");
+      const safeTitle = title.replace(/'/g, "''");
+      const ps = `Add-Type -AssemblyName System.Windows.Forms; ` +
+        `[System.Windows.Forms.MessageBox]::Show('${safeMsg}', '${safeTitle}')`;
+      spawnSync('powershell', ['-NonInteractive', '-Command', ps],
+        { timeout: 3000, stdio: 'ignore' });
     }
   } catch (_e) { /* best-effort */ }
 

package/hooks/legion-mcp-advisor.js

@@ -1,10 +1,10 @@
 #!/usr/bin/env node
-// MCP & Subagent Advisor — Legion SessionStart/SubagentStart hook
+// MCP & Subagent Advisor — Legion SessionStart hook
 //
 // Reads .legion/config.json and injects MCP server awareness + subagent
 // enforcement into agent context. Searches for config up the directory tree.
 //
-// Events: SessionStart (matcher: "startup"), SubagentStart
+// Events: SessionStart (matcher: "startup")
 // Output: hookSpecificOutput.additionalContext with MCP + subagent rules
 
 'use strict';
@@ -57,6 +57,13 @@ process.stdin.on('end', () => {
       process.exit(0);
     }
 
+    // Only inject enforcement on SessionStart — NOT on SubagentStart
+    // to prevent recursive subagent spawning loops
+    const hookEvent = data.hook_event_name || 'SessionStart';
+    if (hookEvent === 'SubagentStart') {
+      process.exit(0);
+    }
+
     const parts = [];
 
     // ── MCP enforcement ───────────────────────────────────────────────────
@@ -90,23 +97,23 @@ process.stdin.on('end', () => {
 
     // ── Subagent enforcement ──────────────────────────────────────────────
     if (config.subagents && config.subagents.enforce !== false) {
-      const sa = config.subagents;
-      const minCount = sa.minCount || 3;
-      const types = Array.isArray(sa.types) ? sa.types : [];
-      const saMessage = sa.message ||
-        'ALWAYS delegate tasks to subagents via Task tool.';
-
-      let saContext =
-        `[LEGION SUBAGENTS] ${saMessage} ` +
-        `Minimum ${minCount} subagents per task. ` +
-        'Use Task tool with specialized subagent_type for each subtask. ' +
-        'Maximize parallel execution — launch independent agents simultaneously.';
-
-      if (types.length > 0) {
-        saContext += ` Recommended types: ${types.join(', ')}.`;
-      }
-
-      parts.push(saContext);
+        const sa = config.subagents;
+        const minCount = sa.minCount || 3;
+        const types = Array.isArray(sa.types) ? sa.types : [];
+        const saMessage = sa.message ||
+          'ALWAYS delegate tasks to subagents via Task tool.';
+
+        let saContext =
+          `[LEGION SUBAGENTS] ${saMessage} ` +
+          `Minimum ${minCount} subagents per task. ` +
+          'Use Task tool with specialized subagent_type for each subtask. ' +
+          'Maximize parallel execution — launch independent agents simultaneously.';
+
+        if (types.length > 0) {
+          saContext += ` Recommended types: ${types.join(', ')}.`;
+        }
+
+        parts.push(saContext);
     }
 
     // ── Output ────────────────────────────────────────────────────────────
@@ -114,8 +121,6 @@ process.stdin.on('end', () => {
       process.exit(0);
     }
 
-    const hookEvent = data.hook_event_name || 'SessionStart';
-
     const output = {
       hookSpecificOutput: {
         hookEventName:     hookEvent,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "legion-cc",
-  "version": "0.20.1",
+  "version": "0.21.0",
   "description": "Legion — context monitoring hooks and agent teams for Claude Code",
   "type": "commonjs",
   "bin": {
@@ -15,7 +15,7 @@
     "README.md"
   ],
   "scripts": {
-    "postinstall": "node bin/install.js 2>/dev/null || true"
+    "postinstall": "node bin/install.js || exit 0"
   },
   "author": "Vladyslav Blackmoore",
   "repository": {