b Review Gate

GPTClaudeDeepSeek··▲ 285 copies·updated 2026-07-14

b-review-gate.prompt

<!-- Orchestrator contract:
  role: driver
  reads: [REVIEW_NOTES.md, PLAN.md, TASK_QUEUE.md, CONTEXT.md, DECISIONS.md]
  writes: [FIX_SUMMARY.md]
  never_writes: [REVIEW_NOTES.md, DIFF_REVIEW_NOTES.md, REREVIEW_NOTES.md]
  produces: [STEP_RESULT.json]
-->
# 2b. Review Gate (Re-review Decision)

You are in Driver mode.

Resolve active slice: Read ROADMAP.md -> find "Current Active Slice" -> use as `changes/<slug>/`.

Goal:
Classify step-2 changes as compliance vs direction-change and decide if explicit reviewer re-review is required.

Do this:
1) Read reviewer must-fix items in `REVIEW_NOTES.md`.
2) Compare against Driver changes in `PLAN.md`, `TASK_QUEUE.md`, `CONTEXT.md`, `DECISIONS.md`.
3) Classify each change:
- Compliance: directly implements reviewer request.
- Direction change: adds scope, changes acceptance, changes architecture/data/auth approach, or deviates materially.
4) Write `FIX_SUMMARY.md` with:
- compliance fixes applied
- direction changes and rationale
- `Re-review required: YES | NO`

Rules:
- Do not modify reviewer verdict files.
- Keep this decision file-driven and explicit.

## Output checklist (non-negotiable)
- [ ] `changes/<slug>/FIX_SUMMARY.md`
- [ ] `changes/<slug>/STEP_RESULT.json`

## STEP_RESULT.json (required)

Write `changes/<slug>/STEP_RESULT.json`:

when to use it

Community prompt sourced from the open-source GitHub repo kylehuirevvision/repo-rails (MIT). A "b Review Gate" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.