---
requires: scratch/wave-mutate.md
produces: scratch/wave-review.md
interactive: true
---
Review what the chord already played.
## Reviewer mode
The launch prompt identifies the reviewer for this exercise.
- **Human reviewer:** conduct the retrospective checkpoint below and record
the human's reaction.
- **Parent reviewer:** independently audit the played mutations against their
evidence and wave objectives. Record each keep/amend/revert verdict as
`parent reviewer judgment`, never as human reaction. Send concrete amendment
or revert work to the Task through the review protocol and verify its reply;
do not implement the Task's changes yourself.
## Orientation
Before starting, orient yourself in this branch:
- Read `scratch/` — design docs and notes for the current work live here
(`scratch/<branch>.md` is this PR's design; `scratch/questions.md` holds open
questions and assumptions).
- Read wave/PM context only when the seed names the exact wave, task, project,
or a concrete coordination question; never infer it or repair access as a
prerequisite.
- Read the repo's agent doc (`CLAUDE.md` / `AGENTS.md`) for conventions.
Write design artifacts, notes, and open questions under `scratch/`. Don't
re-derive what these already record.
## Goal
This is a retrospective human checkpoint, not an approval gate.
Show the human what changed, why it changed, and what it means for the member
waves. If they want to amend or revert something, do that deliberately and
record it.
## Opening
Orient the human quickly:
1. What assessment drove this chord
2. Which waves changed
3. The main tradeoff or risk introduced by the mutations
## Walkthrough
For each applied or skipped mutation:
- show the concrete before / after
- explain the rationale plainly
- note any risks or follow-up
- ask whether to **keep**, **amend**, or **revert** it
If the human wants an amendment or revert, make the smallest clean change that
matches their intent and update the chord record if needed.
## Cross-cutting questions
After the per-mutation walkthrough, zoom out:
- Did the chord address the real pressure?
- Did anything surprising happen during play?
- Is there a mutation that should be undone or added next cycle?
- Are the silent waves still correctly silent?
## Output
Write `scratch/wave-review.md`:
```markdown
# Chord Review — <date>
## Summary
<overall human reaction or explicitly labeled parent reviewer judgment>
## Decisions
### <mutation title>
**Follow-up**: <if any>
## Session Notes
<trajectory observations, calibration notes, context to remember>
```
If the human asks for an amendment or revert, apply it before finishing and note
exactly what changed.
## What to avoid
**Pretending this is pre-approval.** The chord already landed. Review what
actually happened.
**Defensiveness.** If the human dislikes a mutation, treat that as signal.
**Vague summaries.** Tie each decision back to a concrete change.