claude-autopm 1.17.0 → 1.20.0

package/README.md

@@ -269,6 +269,165 @@ autopm pm:epic-sync user-auth
 
 **That's it!** You're now using structured, spec-driven development with AI agents.
 
+## 📋 Complete PM Workflow Guide
+
+### Understanding the Process: PRD → Epic → Tasks
+
+ClaudeAutoPM follows a structured workflow from requirements to implementation:
+
+```
+PRD (Product Requirements)
+  ↓
+Epic Split (Optional - for complex projects)
+  ↓
+Task Decomposition
+  ↓
+GitHub/Azure Sync
+  ↓
+Development & Tracking
+```
+
+### Quick Decision Guide
+
+**When to use ONE epic (`/pm:epic-decompose`):**
+- ✅ Simple feature (1-2 weeks)
+- ✅ Single component (frontend OR backend)
+- ✅ One developer
+- ✅ Examples: "Add user profile page", "Create REST API endpoint"
+
+**When to use MULTIPLE epics (`/pm:epic-split`):**
+- ✅ Complex project (2+ months)
+- ✅ Multiple components (frontend + backend + infrastructure)
+- ✅ Multiple teams working in parallel
+- ✅ Examples: "E-commerce platform", "Social media dashboard"
+
+### Workflow Examples
+
+<details>
+<summary><b>Simple Feature Workflow</b> - Single epic, quick delivery</summary>
+
+```bash
+# 1. Create PRD
+/pm:prd-new user-profile
+
+# 2. Parse and analyze
+/pm:prd-parse user-profile
+
+# 3. Decompose into tasks (ONE epic)
+/pm:epic-decompose user-profile
+# Creates: .claude/epics/user-profile/
+#   ├── epic.md
+#   ├── 001.md  # Create profile component
+#   ├── 002.md  # Add avatar upload
+#   ├── 003.md  # Implement edit form
+#   └── 004.md  # Add validation
+
+# 4. Sync with GitHub/Azure
+/pm:epic-sync user-profile
+
+# 5. Start working
+/pm:next                    # Get next task
+/pm:issue-start #123        # Begin work
+# ... development ...
+/pm:issue-close #123        # Complete task
+```
+</details>
+
+<details>
+<summary><b>Complex Project Workflow</b> - Multiple epics, phased delivery</summary>
+
+```bash
+# 1. Create PRD
+/pm:prd-new ecommerce-platform
+
+# 2. Parse and analyze
+/pm:prd-parse ecommerce-platform
+
+# 3. Split into multiple epics (AUTOMATIC)
+/pm:epic-split ecommerce-platform
+# Automatically creates 6 epics:
+#   Epic 1: Infrastructure (Docker, DB, Redis) - P0, 1w
+#   Epic 2: Auth Backend (JWT, users, RBAC) - P0, 2w
+#   Epic 3: Product API (catalog, orders) - P0, 3w
+#   Epic 4: Frontend Foundation (React setup) - P0, 1w
+#   Epic 5: E-commerce UI (pages, cart) - P1, 3w
+#   Epic 6: Testing & Deployment (CI/CD) - P1, 1w
+
+# 4. Decompose EACH epic into tasks
+/pm:epic-decompose ecommerce-platform/01-infrastructure    # 12 tasks
+/pm:epic-decompose ecommerce-platform/02-auth-backend      # 15 tasks
+/pm:epic-decompose ecommerce-platform/03-product-api       # 20 tasks
+/pm:epic-decompose ecommerce-platform/04-frontend          # 10 tasks
+/pm:epic-decompose ecommerce-platform/05-ui                # 25 tasks
+/pm:epic-decompose ecommerce-platform/06-testing           #  8 tasks
+# TOTAL: ~90 tasks across 6 epics
+
+# 5. Sync ALL epics at once
+/pm:epic-sync ecommerce-platform
+
+# 6. Parallel team work
+/pm:next  # Team 1: Infrastructure (P0)
+/pm:next  # Team 2: Auth Backend (P0, parallel)
+/pm:next  # Team 3: Frontend Foundation (P0, parallel)
+```
+</details>
+
+<details>
+<summary><b>Multiple PRDs Workflow</b> - Independent features</summary>
+
+```bash
+# Create multiple independent features
+/pm:prd-new login-page
+/pm:prd-new payment-system
+/pm:prd-new notifications
+
+# Process each independently
+/pm:prd-parse login-page
+/pm:epic-decompose login-page      # 5 tasks
+/pm:epic-sync login-page
+
+/pm:prd-parse payment-system
+/pm:epic-decompose payment-system  # 8 tasks
+/pm:epic-sync payment-system
+
+/pm:prd-parse notifications
+/pm:epic-decompose notifications   # 4 tasks
+/pm:epic-sync notifications
+
+# /pm:next automatically picks highest priority across ALL epics
+```
+</details>
+
+### Key Commands Reference
+
+| Command | Purpose | When to Use |
+|---------|---------|-------------|
+| `/pm:prd-new <name>` | Create PRD | Always start here |
+| `/pm:prd-parse <name>` | Analyze PRD | After writing requirements |
+| `/pm:epic-split <name>` | Create multiple epics | Complex, multi-component projects |
+| `/pm:epic-decompose <name>` | Break into tasks | Always (once per epic) |
+| `/pm:epic-sync <name>` | Sync with GitHub/Azure | After decomposition |
+| `/pm:context` | Show current state | Check progress anytime |
+| `/pm:next` | Get next task | Ready to work |
+| `/pm:issue-start #N` | Start task | Before coding |
+| `/pm:issue-close #N` | Complete task | After finishing |
+
+### FAQ
+
+**Q: Can I create multiple PRDs at once?**
+A: Yes! Each PRD is independent. Create as many as you need.
+
+**Q: How do I decide between epic-split and epic-decompose?**
+A: Use epic-split for multi-component projects (frontend + backend + infra). Use epic-decompose for single-component features.
+
+**Q: Do I need to decompose ALL epics after split?**
+A: Yes, before syncing. But you can decompose them one at a time as needed.
+
+**Q: How do I check my progress?**
+A: Use `/pm:context` to see current project, epics, and task progress.
+
+📖 **Full Workflow Guide:** See [PM-WORKFLOW-GUIDE.md](PM-WORKFLOW-GUIDE.md) for detailed examples and decision flows.
+
 ## 🎯 Core Workflow: Crawl → Walk → Run
 
 ### Crawl: Basic Commands

package/autopm/.claude/agents/core/mcp-manager.md

@@ -28,7 +28,7 @@ Each MCP server is defined in a Markdown file with:
 - **monitoring**: Observability and metrics
 
 ### Configuration Management
-- Servers defined in `autopm/.claude/mcp/*.md`
+- Servers defined in `.claude/mcp/*.md`
 - Active servers tracked in `.claude/config.json`
 - Runtime config generated to `.claude/mcp-servers.json`
 - Environment variables in `.claude/.env`

package/autopm/.claude/commands/pm/context.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: Bash
+---
+
+Run `node .claude/scripts/pm/context.js` using the Bash tool and show me the complete output.
+
+- DO NOT truncate.
+- DO NOT collapse.
+- DO NOT abbreviate.
+- Show ALL lines in full.
+- DO NOT print any other comments.

package/autopm/.claude/commands/pm/epic-decompose.md

@@ -50,6 +50,19 @@ Do not bother the user with preflight checks progress ("I'm not going to ...").
    - For each epic, check if status is already "completed"
    - If any epic is completed, warn user: "⚠️ Epic(s) marked as completed. Are you sure you want to decompose again?"
 
+## ⚠️ TDD REMINDER
+
+**CRITICAL: All tasks MUST follow Test-Driven Development (TDD).**
+
+When creating tasks, ensure each task includes:
+- TDD Requirements section (RED-GREEN-REFACTOR cycle)
+- "Tests written FIRST" as first item in Definition of Done
+- References to `.claude/rules/tdd.enforcement.md`
+
+Every generated task file will remind developers to write tests first.
+
+---
+
 ## Instructions
 
 You are decomposing epic(s) into specific, actionable tasks for: **$ARGUMENTS**
@@ -123,6 +136,14 @@ conflicts_with: []  # Tasks that modify same files, e.g., [003, 004]
 ## Description
 Clear, concise description of what needs to be done
 
+## ⚠️ TDD Requirements
+**This project uses Test-Driven Development. You MUST:**
+1. 🔴 RED: Write failing test first
+2. 🟢 GREEN: Write minimal code to make test pass
+3. 🔵 REFACTOR: Clean up code while keeping tests green
+
+See `.claude/rules/tdd.enforcement.md` for complete requirements.
+
 ## Acceptance Criteria
 - [ ] Specific criterion 1
 - [ ] Specific criterion 2
@@ -143,8 +164,10 @@ Clear, concise description of what needs to be done
 - Parallel: true/false (can run in parallel with other tasks)
 
 ## Definition of Done
-- [ ] Code implemented
-- [ ] Tests written and passing
+- [ ] Tests written FIRST (RED phase)
+- [ ] Code implemented (GREEN phase)
+- [ ] Code refactored (REFACTOR phase)
+- [ ] All tests passing
 - [ ] Documentation updated
 - [ ] Code reviewed
 - [ ] Deployed to staging

package/autopm/.claude/commands/pm/epic-oneshot.md

@@ -11,6 +11,19 @@ Decompose epic into tasks and sync to GitHub in one operation.
 /pm:epic-oneshot <feature_name>
 ```
 
+## ⚠️ TDD REMINDER
+
+**CRITICAL: All tasks generated will include TDD requirements.**
+
+This command will create tasks that REQUIRE Test-Driven Development:
+- Each task includes TDD Requirements section
+- Definition of Done starts with "Tests written FIRST"
+- All agents must follow RED-GREEN-REFACTOR cycle
+
+See `.claude/rules/tdd.enforcement.md` for complete requirements.
+
+---
+
 ## Instructions
 
 ### 1. Validate Prerequisites

package/autopm/.claude/commands/pm/epic-start.md

@@ -33,6 +33,25 @@ Launch parallel agents to work on epic tasks using the unified branch strategy.
    ```
    If output is not empty: "❌ You have uncommitted changes. Please commit or stash them before starting an epic"
 
+## ⚠️ TDD REMINDER - READ THIS FIRST
+
+**CRITICAL: This project follows Test-Driven Development (TDD).**
+
+Before ANY coding work begins on this epic, you MUST follow the RED-GREEN-REFACTOR cycle:
+
+1. **RED Phase**: Write failing test that describes the desired behavior
+2. **GREEN Phase**: Write minimum code to make test pass
+3. **REFACTOR Phase**: Clean up code while keeping tests green
+
+**For this epic:**
+- All tasks have TDD requirements section
+- Every agent must start with test creation
+- No implementation without tests first
+
+See `.claude/rules/tdd.enforcement.md` for complete TDD requirements.
+
+---
+
 ## Instructions
 
 ### 1. Create or Enter Branch

package/autopm/.claude/commands/pm/epic-sync-modular.md

@@ -33,7 +33,7 @@ This is handled automatically by our modular scripts, but you can run the check
 
 ```bash
 # Check repository protection (built into all scripts)
-bash autopm/.claude/scripts/lib/github-utils.sh
+bash .claude/scripts/lib/github-utils.sh
 ```
 
 The scripts will automatically:
@@ -47,7 +47,7 @@ Create the main GitHub issue for the epic:
 
 ```bash
 # Create epic issue with proper stats and labels
-epic_number=$(bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$ARGUMENTS")
+epic_number=$(bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$ARGUMENTS")
 
 echo "✅ Epic issue created: #$epic_number"
 ```
@@ -65,7 +65,7 @@ Create GitHub issues for all tasks with automatic parallel processing:
 
 ```bash
 # Create task issues (automatically chooses sequential/parallel)
-task_mapping_file=$(bash autopm/.claude/scripts/pm/epic-sync/create-task-issues.sh "$ARGUMENTS" "$epic_number")
+task_mapping_file=$(bash .claude/scripts/pm/epic-sync/create-task-issues.sh "$ARGUMENTS" "$epic_number")
 
 echo "✅ Task issues created. Mapping: $task_mapping_file"
 ```
@@ -90,7 +90,7 @@ Update all task dependencies and rename files to use GitHub issue numbers:
 
 ```bash
 # Update references and rename files
-bash autopm/.claude/scripts/pm/epic-sync/update-references.sh "$ARGUMENTS" "$task_mapping_file"
+bash .claude/scripts/pm/epic-sync/update-references.sh "$ARGUMENTS" "$task_mapping_file"
 
 echo "✅ Task references updated and files renamed"
 ```
@@ -109,7 +109,7 @@ Update the epic.md file with GitHub information and real task IDs:
 
 ```bash
 # Update epic file with GitHub info and real task IDs
-bash autopm/.claude/scripts/pm/epic-sync/update-epic-file.sh "$ARGUMENTS" "$epic_number"
+bash .claude/scripts/pm/epic-sync/update-epic-file.sh "$ARGUMENTS" "$epic_number"
 
 echo "✅ Epic file updated with GitHub information"
 ```
@@ -151,7 +151,7 @@ echo "🚀 Starting modular epic sync for: $EPIC_NAME"
 
 # Step 1: Create epic issue
 echo "📝 Creating epic issue..."
-epic_number=$(bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME")
+epic_number=$(bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME")
 
 if [[ -z "$epic_number" ]]; then
     echo "❌ Failed to create epic issue"
@@ -162,7 +162,7 @@ echo "✅ Epic issue created: #$epic_number"
 
 # Step 2: Create task issues
 echo "📋 Creating task issues..."
-task_mapping_file=$(bash autopm/.claude/scripts/pm/epic-sync/create-task-issues.sh "$EPIC_NAME" "$epic_number")
+task_mapping_file=$(bash .claude/scripts/pm/epic-sync/create-task-issues.sh "$EPIC_NAME" "$epic_number")
 
 if [[ ! -f "$task_mapping_file" ]]; then
     echo "❌ Failed to create task issues"
@@ -174,13 +174,13 @@ echo "✅ Created $task_count task issues"
 
 # Step 3: Update references
 echo "🔗 Updating task references..."
-bash autopm/.claude/scripts/pm/epic-sync/update-references.sh "$EPIC_NAME" "$task_mapping_file"
+bash .claude/scripts/pm/epic-sync/update-references.sh "$EPIC_NAME" "$task_mapping_file"
 
 echo "✅ Task references updated"
 
 # Step 4: Update epic file
 echo "📄 Updating epic file..."
-bash autopm/.claude/scripts/pm/epic-sync/update-epic-file.sh "$EPIC_NAME" "$epic_number"
+bash .claude/scripts/pm/epic-sync/update-epic-file.sh "$EPIC_NAME" "$epic_number"
 
 echo "✅ Epic file updated"
 
@@ -303,7 +303,7 @@ Enable debug logging for detailed troubleshooting:
 
 ```bash
 export AUTOPM_LOG_LEVEL=0  # Enable debug logging
-bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME"
+bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME"
 ```
 
 ## Migration from Legacy Epic Sync

package/autopm/.claude/commands/pm/epic-sync.md

@@ -52,7 +52,7 @@ This is handled automatically by our modular scripts, but you can run the check
 
 ```bash
 # Check repository protection (built into all scripts)
-bash autopm/.claude/scripts/lib/github-utils.sh
+bash .claude/scripts/lib/github-utils.sh
 ```
 
 The scripts will automatically:
@@ -66,7 +66,7 @@ Create the main GitHub issue for the epic:
 
 ```bash
 # Create epic issue with proper stats and labels
-epic_number=$(bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$ARGUMENTS")
+epic_number=$(bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$ARGUMENTS")
 
 echo "✅ Epic issue created: #$epic_number"
 ```
@@ -84,7 +84,7 @@ Create GitHub issues for all tasks with automatic parallel processing:
 
 ```bash
 # Create task issues (automatically chooses sequential/parallel)
-task_mapping_file=$(bash autopm/.claude/scripts/pm/epic-sync/create-task-issues.sh "$ARGUMENTS" "$epic_number")
+task_mapping_file=$(bash .claude/scripts/pm/epic-sync/create-task-issues.sh "$ARGUMENTS" "$epic_number")
 
 echo "✅ Task issues created. Mapping: $task_mapping_file"
 ```
@@ -109,7 +109,7 @@ Update all task dependencies and rename files to use GitHub issue numbers:
 
 ```bash
 # Update references and rename files
-bash autopm/.claude/scripts/pm/epic-sync/update-references.sh "$ARGUMENTS" "$task_mapping_file"
+bash .claude/scripts/pm/epic-sync/update-references.sh "$ARGUMENTS" "$task_mapping_file"
 
 echo "✅ Task references updated and files renamed"
 ```
@@ -128,7 +128,7 @@ Update the epic.md file with GitHub information and real task IDs:
 
 ```bash
 # Update epic file with GitHub info and real task IDs
-bash autopm/.claude/scripts/pm/epic-sync/update-epic-file.sh "$ARGUMENTS" "$epic_number"
+bash .claude/scripts/pm/epic-sync/update-epic-file.sh "$ARGUMENTS" "$epic_number"
 
 echo "✅ Epic file updated with GitHub information"
 ```
@@ -172,7 +172,7 @@ echo "🚀 Starting modular epic sync for: $EPIC_NAME"
 
 # Step 1: Create epic issue
 echo "📝 Creating epic issue..."
-epic_number=$(bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME")
+epic_number=$(bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME")
 
 if [[ -z "$epic_number" ]]; then
     echo "❌ Failed to create epic issue"
@@ -183,7 +183,7 @@ echo "✅ Epic issue created: #$epic_number"
 
 # Step 2: Create task issues
 echo "📋 Creating task issues..."
-task_mapping_file=$(bash autopm/.claude/scripts/pm/epic-sync/create-task-issues.sh "$EPIC_NAME" "$epic_number")
+task_mapping_file=$(bash .claude/scripts/pm/epic-sync/create-task-issues.sh "$EPIC_NAME" "$epic_number")
 
 if [[ ! -f "$task_mapping_file" ]]; then
     echo "❌ Failed to create task issues"
@@ -195,13 +195,13 @@ echo "✅ Created $task_count task issues"
 
 # Step 3: Update references
 echo "🔗 Updating task references..."
-bash autopm/.claude/scripts/pm/epic-sync/update-references.sh "$EPIC_NAME" "$task_mapping_file"
+bash .claude/scripts/pm/epic-sync/update-references.sh "$EPIC_NAME" "$task_mapping_file"
 
 echo "✅ Task references updated"
 
 # Step 4: Update epic file
 echo "📄 Updating epic file..."
-bash autopm/.claude/scripts/pm/epic-sync/update-epic-file.sh "$EPIC_NAME" "$epic_number"
+bash .claude/scripts/pm/epic-sync/update-epic-file.sh "$EPIC_NAME" "$epic_number"
 
 echo "✅ Epic file updated"
 
@@ -269,7 +269,7 @@ for epic_dir in $epic_dirs; do
 
     # Step 1: Create epic issue
     echo "📝 Creating epic issue..."
-    epic_number=$(bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$FEATURE_NAME/$epic_name")
+    epic_number=$(bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$FEATURE_NAME/$epic_name")
 
     if [[ -z "$epic_number" ]]; then
         echo "⚠️  Failed to create epic issue for $epic_name"
@@ -281,7 +281,7 @@ for epic_dir in $epic_dirs; do
 
     # Step 2: Create task issues for this epic
     echo "📋 Creating task issues..."
-    task_mapping_file=$(bash autopm/.claude/scripts/pm/epic-sync/create-task-issues.sh "$FEATURE_NAME/$epic_name" "$epic_number")
+    task_mapping_file=$(bash .claude/scripts/pm/epic-sync/create-task-issues.sh "$FEATURE_NAME/$epic_name" "$epic_number")
 
     if [[ -f "$task_mapping_file" ]]; then
         task_count=$(wc -l < "$task_mapping_file")
@@ -290,10 +290,10 @@ for epic_dir in $epic_dirs; do
 
         # Step 3: Update references
         echo "🔗 Updating task references..."
-        bash autopm/.claude/scripts/pm/epic-sync/update-references.sh "$FEATURE_NAME/$epic_name" "$task_mapping_file"
+        bash .claude/scripts/pm/epic-sync/update-references.sh "$FEATURE_NAME/$epic_name" "$task_mapping_file"
 
         # Step 4: Update epic file
-        bash autopm/.claude/scripts/pm/epic-sync/update-epic-file.sh "$FEATURE_NAME/$epic_name" "$epic_number"
+        bash .claude/scripts/pm/epic-sync/update-epic-file.sh "$FEATURE_NAME/$epic_name" "$epic_number"
     fi
 done
 
@@ -418,7 +418,7 @@ Enable debug logging for detailed troubleshooting:
 
 ```bash
 export AUTOPM_LOG_LEVEL=0  # Enable debug logging
-bash autopm/.claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME"
+bash .claude/scripts/pm/epic-sync/create-epic-issue.sh "$EPIC_NAME"
 ```
 
 ## Migration from Legacy Epic Sync

package/autopm/.claude/commands/pm/issue-start.md

@@ -33,6 +33,26 @@ Begin work on a GitHub issue with parallel agents based on work stream analysis.
    ` ``
    If no analysis exists and no --analyze flag, stop execution.
 
+## ⚠️ TDD REMINDER - READ THIS FIRST
+
+**CRITICAL: This project follows Test-Driven Development (TDD).**
+
+Before ANY coding work begins, you MUST follow the RED-GREEN-REFACTOR cycle:
+
+1. **RED Phase**: Write failing test that describes the desired behavior
+2. **GREEN Phase**: Write minimum code to make test pass
+3. **REFACTOR Phase**: Clean up code while keeping tests green
+
+**For this issue:**
+- Read the task requirements from the task file
+- Identify what tests are needed BEFORE any implementation
+- All agents must start with test creation
+- No implementation without tests first
+
+See `.claude/rules/tdd.enforcement.md` for complete TDD requirements.
+
+---
+
 ## Instructions
 
 ### 1. Ensure Branch Exists
@@ -103,12 +123,30 @@ Task:
   description: "Issue #$ARGUMENTS Stream {X}"
   subagent_type: "{agent_type}"
   prompt: |
-    **CRITICAL RULE: This project uses 'Docker-first development'.**
+    **🚨 CRITICAL RULE #1: Test-Driven Development (TDD) is MANDATORY**
+
+    You MUST follow the RED-GREEN-REFACTOR cycle:
+    1. **RED**: Write a FAILING test first that describes the desired behavior
+    2. **GREEN**: Write MINIMUM code to make the test pass
+    3. **REFACTOR**: Clean up code while keeping all tests green
+
+    **NO CODE WITHOUT TESTS FIRST.** Zero exceptions.
+    - Every function starts with a test
+    - Every bug fix starts with a test that reproduces it
+    - Every feature starts with failing acceptance tests
+
+    See `.claude/rules/tdd.enforcement.md` for complete requirements.
+
+    ---
+
+    **CRITICAL RULE #2: This project uses 'Docker-first development'.**
     - All commands (dependency installation, tests, running the application) MUST be executed inside a Docker container using `docker compose run --rm <service_name> <command>`.
     - DO NOT run `npm`, `pip`, `pytest`, etc., directly on the host.
     - The source code is mounted as a VOLUME, so file changes will be immediately visible in the container (hot-reloading).
     - Full rules can be found in `.claude/rules/docker-first-development.md`.
 
+    ---
+
     You are working on Issue #$ARGUMENTS in the epic branch.
 
     Branch: epic/{epic_name}
@@ -120,10 +158,12 @@ Task:
     
     Requirements:
     1. Read full task from: .claude/epics/{epic_name}/{task_file}
-    2. Work ONLY in your assigned files
-    3. Commit frequently with format: "Issue #$ARGUMENTS: {specific change}"
-    4. Update progress in: .claude/epics/{epic_name}/updates/$ARGUMENTS/stream-{X}.md
-    5. Follow coordination rules in /rules/agent-coordination.md
+    2. **START WITH TESTS**: Write failing tests BEFORE any implementation
+    3. Work ONLY in your assigned files
+    4. Follow TDD cycle: RED (test fails) → GREEN (minimal code) → REFACTOR (cleanup)
+    5. Commit frequently with format: "Issue #$ARGUMENTS: {specific change}"
+    6. Update progress in: .claude/epics/{epic_name}/updates/$ARGUMENTS/stream-{X}.md
+    7. Follow coordination rules in /rules/agent-coordination.md
     
     If you need to modify files outside your scope:
     - Check if another stream owns them
@@ -156,6 +196,11 @@ Launching {count} parallel agents:
 Progress tracking:
   .claude/epics/{epic_name}/updates/$ARGUMENTS/
 
+⚠️  TDD CHECKLIST - All agents MUST follow:
+  1. ❌ RED: Write failing test
+  2. ❌ GREEN: Make test pass (minimal code)
+  3. ❌ REFACTOR: Clean up code
+
 Monitor with: /pm:epic-status {epic_name}
 Sync updates: /pm:issue-sync $ARGUMENTS
 ` ``