example

Example run — plan-review-implement-audit

This directory contains the real captured output of running the plan-review-implement-audit template end-to-end against gpt-4o via OpenAI.

How this was generated

relay init --template plan-review-implement-audit
# write context.md and acceptance_checklist.md
relay run --loop --backend openai
relay distill

What's here

• artifacts/context.md — feature request (add GET /healthz endpoint to a FastAPI service)
• artifacts/acceptance_checklist.md — explicit acceptance criteria the reviewer + auditor check against
• artifacts/plan.md — implementation plan
• artifacts/plan_review.md — reviewer's verdict (APPROVE on second pass after addressing required changes)
• artifacts/build_log.md — what the implementer changed
• artifacts/build_review.md — auditor's verification (APPROVE)
• LESSONS.md, lessons.json — distilled from this run plus a second deliberately-sparse-context run that hit the implement iteration cap. The auditor's ## MUST_FIX and ## SHOULD_FIX sections from that run are what compiled into the implementer- audience lessons here.

Notes

The plan-review-implement-audit template's reviewer and auditor roles were calibrated in v0.2 to default toward APPROVE when explicit acceptance criteria are met (rather than always asking for more polish). Without that calibration, LLM reviewers tend to interpret "be strict but fair" as "always REQUEST_CHANGES" — see the v0.2 design spec at docs/specs/2026-04-28-v0.2-design.md for the reasoning.