speccrew 0.6.44 → 0.6.46

package/.speccrew/agents/speccrew-team-leader-xml.md

@@ -409,6 +409,38 @@ When user reports problems ("something went wrong", "error occurred", "not worki
 - Do not handle business development requests (feature requirements, code modifications, bug fixes), should prompt user to talk directly to Qoder
 - Do not delete or modify WORKFLOW-PROGRESS.json directly (read-only for status display)
 
+## ACTION EXECUTION RULES
+
+When executing XML workflow blocks, map actions to IDE tools as follows:
+- `action="run-skill"` → Use **Skill tool** (pass skill name only, do NOT browse for files)
+- `action="dispatch-to-worker"` → Use **Task tool** (create Task for worker agent)
+- `action="run-script"` → Use **Bash/Terminal tool**
+- `action="read-file"` → Use **Read tool**
+- `action="write-file"` → Use **Write/Edit tool**
+
+**FORBIDDEN**: Do NOT manually search directories for SKILL.md files. Do NOT execute worker tasks yourself — always delegate via Task tool.
+
+## DISPATCH SKILL EXECUTION PROTOCOL
+
+When you load a dispatch skill (e.g., `speccrew-knowledge-bizs-dispatch-xml` or `speccrew-knowledge-techs-dispatch-xml`) via the Skill tool:
+
+1. **You ARE the executor** — do NOT delegate the entire workflow to someone else
+2. **Read the XML workflow** in the loaded SKILL.md from top to bottom
+3. **Execute each block** according to its `action` attribute:
+   - `run-script` → Terminal tool
+   - `run-skill` → Skill tool (do NOT browse directories for SKILL.md)
+   - `dispatch-to-worker` → Task tool (create Task for speccrew-task-worker)
+4. **For `<loop parallel="true">` with `dispatch-to-worker`**: create ALL worker Tasks in ONE batch
+5. **For `<event action="confirm">`**: present to user and wait
+6. **For `<checkpoint>`**: verify condition before proceeding to next stage
+7. **Execute ALL stages in sequence** — Stage 0 → 1a → 1b → 1c → 2 → 3 → 4 (or as defined)
+
+**CRITICAL FORBIDDEN BEHAVIORS**:
+- Do NOT run terminal commands as a substitute for calling a Skill via Skill tool
+- Do NOT manually create JSON files when a Skill should generate them
+- Do NOT execute analysis work that should be dispatched to a Worker via Task tool
+- Do NOT stop between stages to ask the user for direction
+
 ## CONTINUOUS EXECUTION RULES
 
 This agent MUST execute tasks continuously without unnecessary interruptions.

package/.speccrew/skills/speccrew-knowledge-bizs-dispatch-xml/SKILL.md

@@ -4,6 +4,20 @@ description: Dispatch bizs knowledge base generation tasks with 5-stage pipeline
 tools: Read, Write, Task, Bash
 ---
 
+> **⚠️ MANDATORY EXECUTION PROTOCOL — READ BEFORE EXECUTING ANY BLOCK**
+>
+> **Step 1**: Load XML workflow specification: `docs/rules/xml-workflow-spec.md` — this defines all block types and action-to-tool mappings
+>
+> **Step 2**: Execute this SKILL.md's XML workflow **block by block in document order**:
+> - `action="run-script"` → Execute via **Terminal tool** (PowerShell/Bash)
+> - `action="run-skill"` → Invoke via **Skill tool** (do NOT manually read SKILL.md files or browse directories)
+> - `action="dispatch-to-worker"` → Create **Task** via **Task tool** for `speccrew-task-worker`
+> - `action="confirm"` (event) → Present to user and wait for response
+>
+> **Step 3**: Execute ALL stages sequentially without pausing (only stop at explicit `<event action="confirm">` blocks)
+>
+> **FORBIDDEN**: Do NOT run terminal commands as substitute for Skill tool calls. Do NOT do Worker's job yourself. Do NOT skip blocks or improvise.
+
 # Bizs Knowledge Dispatch (XML Block Version)
 
 Orchestrate **bizs knowledge base generation** with a 5-stage pipeline using **XML Block system**: Feature Inventory → Feature Analysis + Graph Write → Module Summarize → UI Style Pattern Extract → System Summary.

package/.speccrew/skills/speccrew-knowledge-techs-dispatch-xml/SKILL.md

@@ -4,6 +4,20 @@ description: Dispatch techs knowledge base generation tasks with 3-stage pipelin
 tools: Read, Write, Task, Bash
 ---
 
+> **⚠️ MANDATORY EXECUTION PROTOCOL — READ BEFORE EXECUTING ANY BLOCK**
+>
+> **Step 1**: Load XML workflow specification: `docs/rules/xml-workflow-spec.md` — this defines all block types and action-to-tool mappings
+>
+> **Step 2**: Execute this SKILL.md's XML workflow **block by block in document order**:
+> - `action="run-script"` → Execute via **Terminal tool** (PowerShell/Bash)
+> - `action="run-skill"` → Invoke via **Skill tool** (do NOT manually read SKILL.md files or browse directories)
+> - `action="dispatch-to-worker"` → Create **Task** via **Task tool** for `speccrew-task-worker`
+> - `action="confirm"` (event) → Present to user and wait for response
+>
+> **Step 3**: Execute ALL stages sequentially without pausing (only stop at explicit `<event action="confirm">` blocks)
+>
+> **FORBIDDEN**: Do NOT run terminal commands as substitute for Skill tool calls. Do NOT do Worker's job yourself. Do NOT skip blocks or improvise.
+
 # Techs Knowledge Dispatch (XML Block Version)
 
 Orchestrate **techs knowledge base generation** with a 3-stage pipeline using **XML Block system**: Platform Detection → Tech Doc Generation → Root Index.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "speccrew",
-  "version": "0.6.44",
+  "version": "0.6.46",
   "description": "Spec-Driven Development toolkit for AI-powered IDEs",
   "author": "charlesmu99",
   "repository": {

package/workspace-template/docs/rules/xml-workflow-spec.md

@@ -329,6 +329,59 @@ Use `${variable_name}` syntax throughout block attributes and content.
 </block>
 ```
 
+## Dispatch Skill Execution Model
+
+Dispatch skills (naming pattern: `*-dispatch-xml`) are **orchestration playbooks** loaded directly by Team Leader via the Skill tool. They contain the complete multi-stage workflow definition.
+
+### Execution Protocol
+
+When Team Leader loads a dispatch skill via `action="run-skill"`:
+
+1. **Load**: Team Leader invokes the Skill tool with the dispatch skill name (e.g., `speccrew-knowledge-bizs-dispatch-xml`)
+2. **Parse**: Read and understand the complete XML workflow in the loaded SKILL.md
+3. **Execute block-by-block**: Walk through each Stage's blocks in document order, mapping each `action` attribute to the correct IDE tool:
+
+| Block Action | What Team Leader MUST Do |
+|---|---|
+| `action="run-script"` | Execute the command via **Terminal tool** (PowerShell/Bash) |
+| `action="run-skill"` | Invoke the skill via **Skill tool** — do NOT manually read SKILL.md files |
+| `action="dispatch-to-worker"` | Create Task via **Task tool**, assign to `speccrew-task-worker` |
+| `action="analyze"` | Perform analysis directly (read files, extract info) |
+| `action="generate"` | Generate content directly (use templates, write output) |
+| `action="confirm"` (event) | Present confirmation to user, wait for response |
+
+4. **Respect control flow**: Honor `<gateway>` branches, `<loop parallel="true">` concurrency, and `<checkpoint>` verification
+5. **Continuous execution**: Proceed from Stage N to Stage N+1 automatically — ONLY pause at explicit `<event action="confirm">` blocks
+
+### FORBIDDEN During Dispatch Execution
+
+- **NEVER** execute commands manually when a block says `action="run-skill"` — use the Skill tool
+- **NEVER** do a worker's job yourself when a block says `action="dispatch-to-worker"` — use the Task tool
+- **NEVER** skip stages or blocks
+- **NEVER** pause between stages to ask "Should I continue?"
+- **NEVER** improvise your own approach — follow the XML blocks exactly
+
+## Action to IDE Tool Mapping
+
+When executing `<block type="task">` blocks, the `action` attribute determines which **IDE tool** to invoke. Do NOT interpret actions freely — use the exact tool specified below:
+
+| Action | IDE Tool to Use | How to Invoke |
+|--------|----------------|---------------|
+| `run-skill` | **Skill tool** | Call the IDE Skill tool with the skill name from `<field name="skill">`. Do NOT manually browse directories or read SKILL.md files — the Skill tool resolves paths automatically. |
+| `dispatch-to-worker` | **Task tool** | Create a new Task via the IDE Task tool, assigning it to the worker agent specified in `<field name="agent">`. Pass all context fields as task parameters. For `<loop parallel="true">`, create ALL tasks in a single batch. |
+| `run-script` | **Bash / Terminal tool** | Execute the command from `<field name="command">` using the terminal tool. Use PowerShell syntax on Windows, Bash on Unix. |
+| `read-file` | **Read tool** | Read the file at the path specified in `<field name="path">`. |
+| `write-file` | **Write / Edit tool** | Write content to the file at `<field name="path">`. For new files use create_file; for modifications use search_replace. |
+| `analyze` | **Direct execution** | Perform the analysis described in the block's `desc` attribute. Read relevant source files, extract information, and store results in the specified output variables. |
+| `generate` | **Direct execution** | Generate the content described in the block. Use templates when specified, write output to the target path. |
+
+### Critical Rules
+
+1. **`run-skill` MUST use the Skill tool** — NEVER manually search for or read SKILL.md files. The IDE Skill tool handles path resolution across different IDE directories (.qoder/, .cursor/, .claude/).
+2. **`dispatch-to-worker` MUST use the Task tool** — NEVER execute worker tasks yourself. Create a Task and let the worker agent handle it.
+3. **`<loop parallel="true">` with `dispatch-to-worker`** — Create ALL worker tasks in ONE batch call, not sequentially. This enables true parallel execution.
+4. **Variable binding** — After a tool call completes, bind the result to the variable specified in `<field name="output" var="..."/>` for use in subsequent blocks.
+
 ## Execution Rules
 
 1. **NEVER skip a block** — execute every block in document order