Claude-Code-Workflow/docs/commands/claude/idaw.md

IDAW Commands

One-Liner

IDAW (Independent Development Autonomous Workflow) is the batch task execution engine — queue development tasks, execute skill chains serially with per-task git checkpoints, and resume from interruptions.

Core Concepts

Concept	Description	Location
Task	Independent JSON task definition	.workflow/.idaw/tasks/IDAW-*.json
Session	Execution session with progress tracking	.workflow/.idaw/sessions/IDA-*/
Skill Chain	Ordered sequence of skills per task type	Mapped from SKILL_CHAIN_MAP
Checkpoint	Per-task git commit after successful execution	Automatic git add -A && git commit

IDAW vs Issue System

Aspect	Issue System	IDAW
Granularity	Fine-grained, multi-step orchestration	Coarse-grained, batch autonomous
Pipeline	new → plan → queue → execute	add → run (all-in-one)
Execution	DAG parallel	Serial with checkpoints
Storage	.workflow/issues.jsonl	.workflow/.idaw/tasks/IDAW-*.json
Use Case	Individual issue resolution	Batch task queue, unattended execution

Command List

Command	Function	Syntax
add	Create tasks manually or import from issue	/idaw:add [-y] [--from-issue <id>] "description" [--type <type>] [--priority 1-5]
run	Execute task queue with git checkpoints	/idaw:run [-y] [--task <id,...>] [--dry-run]
run-coordinate	Execute via external CLI with hook callbacks	/idaw:run-coordinate [-y] [--task <id,...>] [--tool <tool>]
status	View task and session progress	/idaw:status [session-id]
resume	Resume interrupted session	/idaw:resume [-y] [session-id]

Command Details

add

Function: Create IDAW tasks manually or import from existing ccw issues.

Syntax:

/idaw:add [-y|--yes] [--from-issue <id>[,<id>,...]] "description" [--type <task_type>] [--priority <1-5>]

Options:

• --from-issue <id>: Import from ccw issue (comma-separated for multiple)
• --type <type>: Explicit task type (see Task Types)
• --priority 1-5: Priority (1=critical, 5=low, default=3)

Modes:

Mode	Trigger	Behavior
Manual	No --from-issue	Parse description, generate task
Import	--from-issue	Fetch issue, freeze snapshot, create task

Examples:

# Manual creation
/idaw:add "Fix login timeout bug" --type bugfix --priority 2
/idaw:add "Add rate limiting to API endpoints" --priority 1
/idaw:add "Refactor auth module to use strategy pattern"

# Import from ccw issue
/idaw:add --from-issue ISS-20260128-001
/idaw:add --from-issue ISS-20260128-001,ISS-20260128-002

# Auto mode (skip clarification)
/idaw:add -y "Quick fix for typo in header"

Output:

Created IDAW-001: Fix login timeout bug
  Type: bugfix | Priority: 2 | Source: manual
  → Next: /idaw:run or /idaw:status

---

run

Function: Main orchestrator — execute task skill chains serially with git checkpoints.

Syntax:

/idaw:run [-y|--yes] [--task <id>[,<id>,...]] [--dry-run]

Options:

• --task <id,...>: Execute specific tasks (default: all pending)
• --dry-run: Show execution plan without running
• -y: Auto mode — skip confirmations, auto-skip on failure

6-Phase Execution:

Phase 1: Load Tasks
   └─ Glob IDAW-*.json → filter → sort by priority ASC, ID ASC

Phase 2: Session Setup
   └─ Create session.json + progress.md + TodoWrite

Phase 3: Startup Protocol
   ├─ Check running sessions → offer resume or fresh
   └─ Check git status → stash/continue/abort

Phase 4: Main Loop (serial)
   For each task:
   ├─ Resolve: skill_chain || SKILL_CHAIN_MAP[task_type || inferred]
   ├─ Execute each skill (retry once on failure)
   └─ On error: skip (autoYes) or ask (interactive)

Phase 5: Checkpoint (per task)
   ├─ git add -A && git commit
   ├─ Update task.json + session.json
   └─ Append to progress.md

Phase 6: Report
   └─ Summary: completed/failed/skipped counts + git commits

Examples:

# Execute all pending tasks (auto mode)
/idaw:run -y

# Execute specific tasks
/idaw:run --task IDAW-001,IDAW-003

# Preview execution plan
/idaw:run --dry-run

---

run-coordinate

Function: Coordinator variant of /idaw:run — executes via external CLI with hook callbacks instead of blocking Skill() calls.

Syntax:

/idaw:run-coordinate [-y|--yes] [--task <id>[,<id>,...]] [--dry-run] [--tool <tool>]

Options:

• --tool <tool>: CLI tool to use (claude, gemini, qwen, default: claude)
• --task <id,...>: Execute specific tasks
• --dry-run: Preview plan without executing
• -y: Auto mode

Execution Model:

Launch skill via ccw cli --tool <tool> --mode write (background)
    ↓
★ STOP — wait for hook callback
    ↓
Hook fires → handleStepCompletion()
    ├─ More skills in chain → launch next → STOP
    ├─ Chain complete → git checkpoint → next task → STOP
    └─ All done → Report

When to Use:

Scenario	Command
Standard execution (main process, blocking)	/idaw:run
External CLI, isolated context per skill	/idaw:run-coordinate
Long-running tasks, avoid context pressure	/idaw:run-coordinate
Need specific CLI tool (claude/gemini)	/idaw:run-coordinate --tool gemini

Differences from /idaw:run:

Aspect	/idaw:run	/idaw:run-coordinate
Execution	Skill() blocking	ccw cli background + hook
Context	Shared main context	Isolated per CLI call
Tool selection	N/A	--tool claude|gemini|qwen
State	session.json	session.json + prompts_used

Examples:

# Execute via claude CLI (default)
/idaw:run-coordinate -y

# Use gemini as execution tool
/idaw:run-coordinate -y --tool gemini

# Specific tasks
/idaw:run-coordinate --task IDAW-001,IDAW-003 --tool claude

---

status

Function: Read-only view of IDAW task queue and session progress.

Syntax:

/idaw:status [session-id]

View Modes:

Mode	Trigger	Output
Overview	No arguments	All tasks table + latest session summary
Session Detail	Session ID provided	Task × status × commit table + progress.md

Examples:

# Overview
/idaw:status

# Specific session
/idaw:status IDA-auth-fix-20260301

Output Example:

# IDAW Tasks

| ID       | Title                    | Type    | Priority | Status    |
|----------|--------------------------|---------|----------|-----------|
| IDAW-001 | Fix auth token refresh   | bugfix  | 1        | completed |
| IDAW-002 | Add rate limiting        | feature | 2        | pending   |
| IDAW-003 | Refactor payment module  | refactor| 3        | pending   |

Total: 3 | Pending: 2 | Completed: 1 | Failed: 0

---

resume

Function: Resume an interrupted IDAW session from the last checkpoint.

Syntax:

/idaw:resume [-y|--yes] [session-id]

Options:

• session-id: Resume specific session (default: latest running)
• -y: Auto-skip interrupted task, continue with remaining

Recovery Flow:

1. Find session with status=running
2. Handle interrupted task (in_progress):
   ├─ autoYes → mark as skipped
   └─ interactive → ask: Retry or Skip
3. Build remaining task queue
4. Execute Phase 4-6 from /idaw:run

Examples:

# Resume most recent running session
/idaw:resume

# Resume specific session
/idaw:resume IDA-auth-fix-20260301

# Resume with auto mode
/idaw:resume -y

Task Types

IDAW supports 10 task types, each mapping to a specific skill chain:

Task Type	Skill Chain	Use Case
bugfix	lite-plan → test-fix	Standard bug fixes
bugfix-hotfix	lite-plan (--hotfix)	Urgent production fixes
feature	lite-plan → test-fix	New features
feature-complex	plan → execute → test-fix	Multi-module features
refactor	refactor-cycle	Code restructuring
tdd	tdd-plan → execute	Test-driven development
test	test-fix	Test generation
test-fix	test-fix	Fix failing tests
review	review-cycle	Code review
docs	lite-plan	Documentation

Type Resolution: Explicit task_type field takes priority. When null, the type is inferred from title and description using keyword matching at execution time.

Task Schema

{
  "id": "IDAW-001",
  "title": "Fix auth token refresh race condition",
  "description": "Detailed problem/goal description...",
  "status": "pending",
  "priority": 2,
  "task_type": "bugfix",
  "skill_chain": null,
  "context": {
    "affected_files": ["src/auth/token-manager.ts"],
    "acceptance_criteria": ["No concurrent refresh requests"],
    "constraints": [],
    "references": []
  },
  "source": {
    "type": "manual",
    "issue_id": null,
    "issue_snapshot": null
  },
  "execution": {
    "session_id": null,
    "started_at": null,
    "completed_at": null,
    "skill_results": [],
    "git_commit": null,
    "error": null
  },
  "created_at": "2026-03-01T10:00:00Z",
  "updated_at": "2026-03-01T10:00:00Z"
}

Key Fields:

• task_type: Optional — inferred from title/description when null
• skill_chain: Optional — overrides automatic mapping when set
• source.type: manual or import-issue
• source.issue_snapshot: Frozen copy of original issue data (import only)
• execution: Runtime state populated by /idaw:run

Task Lifecycle

graph TD
    A["/idaw:add"] --> B["pending"]
    B --> C["/idaw:run"]
    C --> D["in_progress"]
    D --> E{"Skill Chain"}
    E -->|Success| F["completed (git commit)"]
    E -->|Retry once| G{"Retry"}
    G -->|Success| F
    G -->|Fail| H{"autoYes?"}
    H -->|Yes| I["failed (skip)"]
    H -->|No| J["Ask: Skip/Abort"]
    J -->|Skip| I
    J -->|Abort| K["Session failed"]

    L["/idaw:resume"] --> M{"Interrupted task"}
    M -->|Retry| B
    M -->|Skip| N["skipped"]

Directory Structure

.workflow/.idaw/
├── tasks/                          # Task definitions (persist across sessions)
│   ├── IDAW-001.json
│   ├── IDAW-002.json
│   └── IDAW-003.json
└── sessions/                       # Execution sessions
    └── IDA-{slug}-YYYYMMDD/
        ├── session.json            # Session state + task queue
        └── progress.md             # Human-readable progress log

Error Handling

Error	Resolution
No tasks found	Suggest /idaw:add
Task JSON parse error	Skip malformed task, log warning
Task type unresolvable	Default to feature chain
Skill failure	Retry once → skip (autoYes) or ask (interactive)
Git commit fails (no changes)	Record no-commit, continue
Dirty git tree	autoYes: proceed; interactive: ask
Session ID collision	Append -2 suffix
Issue fetch fails (import)	Log error, skip issue
Duplicate import (same issue_id)	Warn and skip
No resumable sessions	Suggest /idaw:run

Typical Workflow

# 1. Queue tasks
/idaw:add "Fix login timeout bug" --type bugfix --priority 1
/idaw:add "Add rate limiting to API" --priority 2
/idaw:add --from-issue ISS-20260128-001,ISS-20260128-002

# 2. Preview execution plan
/idaw:run --dry-run

# 3. Execute all (unattended)
/idaw:run -y

# 4. Check progress
/idaw:status

# 5. Resume if interrupted
/idaw:resume -y

Related Documentation

• Issue Commands — Fine-grained issue management
• Core Orchestration — /ccw main orchestrator
• Workflow Commands — Individual workflow skills