---
deferred_work_file: '{implementation_artifacts}/deferred-work.md'
---

# Step 4: Review

## RULES

- YOU MUST ALWAYS SPEAK OUTPUT in your Agent communication style with the config `{communication_language}`
- No human interaction: do not ask questions or wait for approval in this step.
- Review subagents get no prior session context.
- All review subagents must run at the same model capability as the current session.

## INSTRUCTIONS

Change `{spec_file}` status to `in-review` in the frontmatter before continuing.

### Construct Diff

Read `{baseline_revision}` from `{spec_file}` frontmatter. If `{baseline_revision}` is missing or `NO_VCS`, use best effort to determine what changed. Otherwise, construct `{diff_output}` covering all changes — tracked and untracked — since `{baseline_revision}`.

Do NOT `git add` anything — this is read-only inspection.

### Review

Launch two subagents without prior session context.

- **Blind hunter** — receives inline `{diff_output}` only. No spec, no context docs, no project access. Invoke via the `bmad-review-adversarial-general` skill.
- **Edge case hunter** — receives `{diff_output}` and read access to the project. Invoke via the `bmad-review-edge-case-hunter` skill.

### Classify

1. Deduplicate all review findings.
2. Classify each finding. The first three categories are **this story's problem** — caused or exposed by the current change. The last two are **not this story's problem**.
   - **intent_gap** — caused by the change; cannot be resolved from the spec because the captured intent is incomplete. Do not infer intent unless there is exactly one possible reading.
   - **bad_spec** — caused by the change, including direct deviations from spec. The spec should have been clear enough to prevent it. When in doubt between bad_spec and patch, prefer bad_spec — a spec-level fix is more likely to produce coherent code.
   - **patch** — caused by the change; trivially fixable without human input. Just part of the diff.
   - **defer** — pre-existing issue not caused by this story, surfaced incidentally by the review. Collect for later focused attention.
   - **reject** — noise. Drop silently. When unsure between defer and reject, prefer reject — only defer findings you are confident are real.
3. Process findings in cascading order. If intent_gap exists, lower findings are moot; follow the intent_gap branch below. If bad_spec exists, lower findings are moot since code will be re-derived. If neither exists, process patch and defer normally. Before each bad_spec loopback, read `{spec_file}` frontmatter `review_loop_iteration` (missing means `0`), increment it by 1, and write it back. If it exceeds 5, HALT with status `blocked` and blocking condition `review repair loop exceeded 5 iterations`.
   - **intent_gap** — Root cause is inside `<intent-contract>`. Revert code changes. HALT with status `blocked`, blocking condition `intent gap in intent contract`, and include the intent-gap findings.
   - **bad_spec** — Root cause is outside `<intent-contract>`. Do not modify content inside `<intent-contract>`. Before reverting code: extract KEEP instructions for positive preservation (what worked well and must survive re-derivation). Revert code changes. Read the `## Spec Change Log` in `{spec_file}` and strictly respect all logged constraints when amending the sections outside `<intent-contract>` that contain the root cause. Append a new change-log entry recording: the triggering finding, what was amended, the known-bad state avoided, and the KEEP instructions. Read fully and follow `./step-03-implement.md` to re-derive the code, then this step will run again.
   - **patch** — Auto-fix. These are the only findings that survive loopbacks.
   - **defer** — Append one new entry to `{deferred_work_file}` using this format. Do not modify existing entries or look for duplicates.
     ```markdown
     - source_spec: `{spec_file}`
       summary: <one sentence>
       evidence: <why this is real>
     ```
   - **reject** — Drop silently.

## Finalize

Prepare `Auto Run Result` details:
- Summary of implemented change
- Files changed with one-line descriptions
- Review findings breakdown: patches applied, items deferred, items rejected
- Verification performed, including command outcomes or manual inspection notes
- Any residual risks

HALT with status `done`.