speccrew 0.2.1 → 0.2.2

package/.speccrew/agents/speccrew-system-designer.md

@@ -221,9 +221,9 @@ When there is only one Feature Spec and one platform:
 
 ### 5.4 Parallel Execution (Feature × Platform)
 
-> **IMPORTANT**: Use the **Skill tool** (not the Agent tool) to invoke each design skill.
+> **IMPORTANT**: Dispatch `speccrew-task-worker` agents (via Agent tool) for parallel design execution. Do NOT call design skills directly — each Feature×Platform combination MUST run in an independent Worker Agent for progress visibility and error isolation.
 
-When multiple Feature Specs and/or multiple platforms exist, create a matrix of **Feature × Platform** and use the **Skill tool** to invoke per-platform design skills in parallel:
+When multiple Feature Specs and/or multiple platforms exist, create a matrix of **Feature × Platform** and dispatch `speccrew-task-worker` agents in parallel:
 
 **Worker Matrix:**
 
@@ -245,19 +245,19 @@ Each worker receives:
 
 **Before dispatch**: Update each task status to `in_progress` with `started_at` timestamp.
 
-**Parallel execution example** (2 features × 3 platforms = 6 skill invocations):
-- Skill 1: speccrew-sd-frontend for Feature A on web-vue → 03.system-design/web-vue/
-- Skill 2: speccrew-sd-backend for Feature A on backend-spring → 03.system-design/backend-spring/
-- Skill 3: speccrew-sd-mobile for Feature A on mobile-uniapp → 03.system-design/mobile-uniapp/
-- Skill 4: speccrew-sd-frontend for Feature B on web-vue → 03.system-design/web-vue/
-- Skill 5: speccrew-sd-backend for Feature B on backend-spring → 03.system-design/backend-spring/
-- Skill 6: speccrew-sd-mobile for Feature B on mobile-uniapp → 03.system-design/mobile-uniapp/
+**Parallel execution example** (2 features × 3 platforms = 6 workers):
+- Worker 1: speccrew-sd-frontend for Feature A on web-vue → 03.system-design/web-vue/
+- Worker 2: speccrew-sd-backend for Feature A on backend-spring → 03.system-design/backend-spring/
+- Worker 3: speccrew-sd-mobile for Feature A on mobile-uniapp → 03.system-design/mobile-uniapp/
+- Worker 4: speccrew-sd-frontend for Feature B on web-vue → 03.system-design/web-vue/
+- Worker 5: speccrew-sd-backend for Feature B on backend-spring → 03.system-design/backend-spring/
+- Worker 6: speccrew-sd-mobile for Feature B on mobile-uniapp → 03.system-design/mobile-uniapp/
 
-All skills execute simultaneously to maximize efficiency.
+All workers execute simultaneously to maximize efficiency.
 
 ### 5.5 Update DISPATCH-PROGRESS.json
 
-After each skill completes, parse its **Task Completion Report** and update:
+After each worker completes, parse its **Task Completion Report** and update:
 
 ```json
 {
@@ -275,7 +275,7 @@ After each skill completes, parse its **Task Completion Report** and update:
 }
 ```
 
-Wait for all skills to complete before proceeding to Phase 6.
+Wait for all workers to complete before proceeding to Phase 6.
 
 ## Phase 6: Joint Confirmation
 

package/.speccrew/agents/speccrew-system-developer.md

@@ -184,7 +184,7 @@ If any pre-check fails:
 
 ## Step 4: Dispatch Per-Module Dev Skills
 
-> **IMPORTANT**: Use the **Skill tool** to dispatch dev skills for parallel execution.
+> **IMPORTANT**: Dispatch `speccrew-task-worker` agents (via Agent tool) for parallel module development. Do NOT call dev skills directly — each module MUST run in an independent Worker Agent for progress visibility and error isolation.
 
 ### 4.0 Initialize DISPATCH-PROGRESS.json
 
@@ -282,11 +282,11 @@ for each platform_id:
    }
    ```
 
-### 4.3 Dispatch Skills with Concurrency Limit
+### 4.3 Dispatch Workers with Concurrency Limit
 
-**Max concurrent child agents: 10**
+**Max concurrent workers: 10**
 
-Process `task_list` using a queue-based concurrency limit model:
+Process `task_list` using a queue-based concurrency limit model. Each task runs in an independent `speccrew-task-worker` agent:
 
 ```
 MAX_CONCURRENT = 10
@@ -303,24 +303,24 @@ while pending is not empty or running is not empty:
       - Set task.status = "in_progress"
       - Set task.started_at = current timestamp
     
-    // Use Skill tool (not Agent tool)
-    skill_result = invoke Skill tool with:
-      - skill: {task.skill_name}
-      - args: |
-          platform_id: {task.platform_id}
-          iteration_path: {task.iteration_path}
-          design_doc_path: {task.module_design_path}
-          api_contract_path: {task.api_contract_path}
-          techs_knowledge_paths: {task.techs_knowledge_paths}
-          task_id: {task.id}  // Pass task ID for completion report
+    // Dispatch speccrew-task-worker agent (NOT Skill tool directly)
+    Invoke `speccrew-task-worker` agent with:
+      - skill_name: {task.skill_name}
+      - context:
+        - platform_id: {task.platform_id}
+        - iteration_path: {task.iteration_path}
+        - design_doc_path: {task.module_design_path}
+        - api_contract_path: {task.api_contract_path}
+        - techs_knowledge_paths: {task.techs_knowledge_paths}
+        - task_id: {task.id}  // Pass task ID for completion report
     
-    running.add({task_id: task.id, skill_handle: skill_result})
+    running.add({task_id: task.id})
   
-  wait until at least one skill in running completes
+  wait until at least one worker in running completes
   
-  // Process completed skill result
-  for each finished skill in running:
-    Parse Task Completion Report from skill output
+  // Process completed worker result
+  for each finished worker in running:
+    Parse Task Completion Report from worker output
     
     if report.status == "SUCCESS":
       Update DISPATCH-PROGRESS.json:
@@ -339,17 +339,17 @@ while pending is not empty or running is not empty:
 ```
 
 **Dispatch rules:**
-- Each skill invocation handles **one module** on **one platform** (not all modules)
+- Each worker handles **one module** on **one platform** (not all modules)
 - Pass complete context including `design_doc_path`, `skill_name`, platform info, and `task_id`
-- Up to 10 skills execute simultaneously (concurrency limit)
+- Up to 10 workers execute simultaneously (concurrency limit)
 - Update DISPATCH-PROGRESS.json **before** dispatch (status → "in_progress")
 - Update DISPATCH-PROGRESS.json **after** completion based on Task Completion Report
 - Track all dispatched tasks: completed / failed / pending counts
-- If a skill fails, log the failure and continue with remaining tasks
-- Wait for all skills to complete before proceeding to Step 5 (Integration Check)
+- If a worker fails, log the failure and continue with remaining tasks
+- Wait for all workers to complete before proceeding to Step 5 (Integration Check)
 
 **Progress Update After Each Batch:**
-After processing a batch of completed skills:
+After processing a batch of completed workers:
 1. Read current DISPATCH-PROGRESS.json
 2. Update counts: `completed`, `failed`, `pending`
 3. Write updated DISPATCH-PROGRESS.json

package/.speccrew/agents/speccrew-test-manager.md

@@ -163,7 +163,7 @@ Design test cases based on loaded knowledge:
 
 After reading `DESIGN-OVERVIEW.md`:
 - **Single Platform**: Invoke Skill directly
-- **Multiple Platforms**: Use **Skill tool** to invoke per-platform skills in parallel
+- **Multiple Platforms**: Dispatch `speccrew-task-worker` agents in parallel (via Agent tool)
 
 ### 3.2 Single Platform Execution
 
@@ -177,7 +177,7 @@ Invoke Skill directly with parameters:
 
 ### 3.3 Multi-Platform Parallel Execution
 
-> **IMPORTANT**: Use the **Skill tool** (not the Agent tool) to invoke each test skill.
+> **IMPORTANT**: Dispatch `speccrew-task-worker` agents (via Agent tool) for parallel test execution. Do NOT call test skills directly — each platform MUST run in an independent Worker Agent for progress visibility and error isolation.
 
 > **DISPATCH-PROGRESS Strategy**: Append mode — each test phase appends its tasks to the existing DISPATCH-PROGRESS.json with a distinct `phase` field. Previous phase records are preserved for full traceability.
 
@@ -210,7 +210,7 @@ Before dispatching, create or update `DISPATCH-PROGRESS.json`:
 
 **Dispatch Workers:**
 
-Use the **Skill tool** to invoke `speccrew-test-case-design` for each platform in parallel:
+Dispatch `speccrew-task-worker` agents for `speccrew-test-case-design` for each platform in parallel:
 - Each worker receives:
   - `skill_name`: `speccrew-test-case-design`
   - `context`:
@@ -321,7 +321,7 @@ Update `DISPATCH-PROGRESS.json` with new phase:
 
 **Dispatch Workers:**
 
-Use the **Skill tool** to invoke `speccrew-test-code-gen` for each platform in parallel:
+Dispatch `speccrew-task-worker` agents for `speccrew-test-code-gen` for each platform in parallel:
   - `context`:
     - `test_cases_path`: Path to the platform-specific test cases document
     - `system_design_path`: Path to the platform system design document
@@ -430,7 +430,7 @@ Update `DISPATCH-PROGRESS.json` with new phase:
 
 **Dispatch Workers:**
 
-Use the **Skill tool** to invoke `speccrew-test-execute` for each platform in parallel:
+Dispatch `speccrew-task-worker` agents for `speccrew-test-execute` for each platform in parallel:
   - `context`:
     - `test_code_path`: Path to the platform test code directory
     - `platform_id`: Platform identifier
@@ -575,7 +575,7 @@ Update `WORKFLOW-PROGRESS.json`:
 **Must do:**
 - Execute three phases in strict order: test case design → code generation → test execution
 - Each phase must have a Checkpoint with user confirmation before proceeding
-- Multi-platform scenarios must use the **Skill tool** to invoke per-platform skills in parallel
+- Multi-platform scenarios must dispatch `speccrew-task-worker` agents (via Agent tool) for parallel execution per platform
 - Test cases must be traceable to Feature Spec requirements
 - Bug reports must reference specific test case IDs
 - Use platform_id from design overview as directory names

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "speccrew",
-  "version": "0.2.1",
+  "version": "0.2.2",
   "description": "Spec-Driven Development toolkit for AI-powered IDEs",
   "author": "charlesmu99",
   "repository": {