ayoubben18/ab-method / .agents/skills/review-implementation
review-implementation
UnverifiedPost-implementation review. Spins up three read-only critics on a completed task's diff — cleaner-architecture, slop-defender, reusability-inspector — that push back ONLY on real issues. Autonomous runs apply safe fixes (tests-green-gated) and write everything to review.md next to progress-tracker.md; interactive runs present findings to pick. Use after a task's missions are done (from start-task / start-roadmap / create-task) or standalone on a diff.
Review Implementation (post-implementation)
Review a completed task's diff through three lenses, each a read-only critic subagent. The counterpart to ../critique-plan/SKILL.md: that guards the plan before coding; this guards the result after.
Silence is the default output. A clean diff produces no findings — that is the normal case. A finding survives only if a senior engineer, looking at this diff, would actually make the change. State each as a concrete cost, not a preference. Never invent refactors to look thorough.
ALWAYS check .ab-method/structure/index.yaml FIRST for paths (task location, domain model,
review.md). Review only what this task changed — the cohesive diff across its missions (git diff
for the commit range). This is not a whole-codebase audit; that's /improve-codebase-architecture.
Look for it under the project root (the current working directory) first — a project's own copy is how it
customises its paths, so it always wins. Only if the project has none (AB Method installed as a plugin rather than
with npx ab-method), read the bundled default: ../../../.ab-method/structure/index.yaml relative to this
SKILL.md. Either way, every path the index names is relative to the project root, never to the folder the
bundled file lives in.
Process
1. Gather diff + context
The changed files and their git diff, plus (read only what exists): UBIQUITOUS_LANGUAGE.md /
CONTEXT.md (canonical terms, spotting reinvented concepts), the pattern docs for the layers the diff
touches (docs/architecture/<layer>/index.md, then only the relevant topic files), docs/adr/ (don't
propose what an ADR settled), .ab-method/rules/code.md (the rules the code was written under), the
task's progress-tracker.md sections ## Placement and ## Protected (## Test Plan in legacy trackers), and its
unresolved-questions.md if it has one.
Black boxes are deliberate — seed every critic with the file. A TODO(UQ-n) seam whose entry is
recorded there is a decision the user signed off on: ship a placeholder rather than guess. No lens may
flag it as slop, a shallow module, a speculative abstraction, or dead code, and none may propose
"just implement it properly" — the answer isn't theirs to pick. Two things are fair game and should be
reported: a TODO(UQ-n) marker with no matching entry (an orphan black box nobody recorded), and a
placeholder that leaked past its named seam into several call sites — the seam was supposed to contain
it, and containing it again is a safe fix.
2. Spin up THREE read-only critics — in parallel
Spawn three subagents in one batch, named exactly:
| Agent | Lens | Brief |
|---|---|---|
cleaner-architecture | depth / deepening | ARCHITECTURE.md |
slop-defender | AI code-slop | SLOP.md |
reusability-inspector | duplication / reuse | REUSE.md |
Each is read-only — analyses the diff, returns a findings list (often empty), edits nothing. Seed
each with the diff + context + its lens file — including the task's unresolved-questions.md when it
exists, since every lens would otherwise read its placeholders as defects (each lens file says so too). On Codex (spawn_agent is one level deep) spawn them
at the orchestrator's own level — flat; on Claude they may nest. Flat works on both.
3. Collect + classify
Merge the lists. Drop anything that's a preference (not a concrete cost), an ADR already settled, or a design change bigger than this task (note it open, don't act). Classify each survivor:
safe-fix— confirmed, mechanical, covered by existing tests, no interface/behavior change (delete dead code, inline a pass-through, drop a redundant comment, call an existing util of identical semantics).needs-judgment— real but design-level, risky, behavior/interface-affecting, or not test-covered.
4. Act — by mode
Interactive (manual create-task tail, or standalone): present findings grouped by lens, each marked
safe-fix/needs-judgment; apply what the user approves; run tests; report.
Autonomous (start-task / start-roadmap — afk): the orchestrator (never the parallel critics)
applies fixes, so there are no concurrent writes:
- For each
safe-fix: apply it, re-run the focused tests for what it touched. Green → keep. Red → revert it, downgrade toneeds-judgmentwith a note (attempted, reverted — broke <test>). If any fix was kept, run the full suite (command fromtech-stack.md) once at the end; red → revert the fixes one by one until green, downgrading each reverted one. - Commit kept fixes as one
refactor(<task>): post-review cleanup(repo convention). Nothing kept → no commit. - Write
docs/tasks/<task>/review.md(below) — every finding, applied or open. Never prompt; anything uncertain stays open rather than being changed.
5. review.md format
Written next to progress-tracker.md. Always write it in autonomous mode — even when clean — so the afk
user knows the review ran.
# Post-Implementation Review: <Task Name>
**Reviewed**: YYYY-MM-DD **Scope**: missions <a–z> / commit <range>
## Applied — safe fixes ✅ (commit <hash>)
- [slop-defender] removed pass-through wrapper `fooProxy` — src/foo.ts
- [cleaner-architecture] inlined shallow `formatName` into its one caller — src/user.ts
## Open — need your judgment ⬜
### [reusability-inspector] duplicates `paymentService.calculateTax`
- **Files**: src/checkout/tax.ts
- **Why it matters**: two tax formulas will drift; a rate change must be made twice.
- **Suggested change**: call the existing `paymentService.calculateTax` instead.
## Clean lenses
- slop-defender: no further findings
## Checked, dropped
- [cleaner-architecture] `RefundRow` looks shallow, but it is the table's row adapter — two adapters exist
## Checked, dropped (optional) lists near-misses a critic examined and rejected, with the reason, so
later reviews (review-pr) don't re-raise them. Each lens's evidence line (Checked: ...) goes under its
heading or after its clean line.
If all three lenses are clean and nothing was applied, the whole body is one line:
All three lenses clean — no findings.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |