review

REVIEW Mode Protocol

You are entering REVIEW mode. Determine which type of review is needed and follow the appropriate protocol.

Detect Review Context

Check board_get_tasks for the active project:

• If tasks exist with status: 'todo' and riper_mode: 'plan' → Post-PLAN Review
• If tasks exist with status: 'done' and riper_mode: 'execute' → Post-EXECUTE Review
• If git diff shows uncommitted changes → Post-EXECUTE Review
• If unclear, ask the user: "Is this a plan review or an implementation review?"

---

Post-PLAN Review

Validate the plan is complete and correct before execution begins.

Checklist

1. Query board tasks: board_get_tasks(project_id, status='todo')
2. For each task, verify:
   • Has an assigned_agent set (FAIL if missing)
   • The assigned agent matches the task type (see /plan skill mapping)
   • Dependencies are correct — no circular deps, no missing prerequisites
   • Priority makes sense — blocking tasks are critical or high
   • Description has enough context for the agent to execute standalone
   • Description includes acceptance criteria (FAIL if missing — "build X" without "with Y behavior, Z config" is insufficient)
   • Acceptance criteria include any user-specified requirements (schedule, trigger method, notification rules, etc.)
3. Check for gaps:
   • Are there missing tasks? (e.g., plan mentions 5 steps but only 4 tasks exist)
   • Is there a test-runner task? (there should be one for integration testing)
   • Is there a code-reviewer task for post-implementation review?
4. Fix issues:
   • Missing agent → board_update_task to set assigned_agent
   • Wrong agent → board_update_task to correct it
   • Missing task → board_create_task to add it
   • Wrong dependency → board_update_task to fix depends_on

Verdict

• PASS: All tasks have correct assignments, dependencies are valid, no gaps. Ready for EXECUTE.
• NEEDS REVISION: List specific issues. Do NOT proceed to EXECUTE until resolved.