moonlight-lupin/agent-skills / plugins/dev-pipeline/skills/review-lessons
review-lessons
VerifiedLibrary of lessons from past build-review loops — not a stage. Use when a review or fix round hits a specific case: a reviewer BLOCKING that contradicts the brief, a test that passes but checks nothing, a feature dead in production, SQLite/WAL migrations, MCP tool surfaces, REST API authz, app ↔ external-source sync, Android/APK verification, fill-template extraction, CLI/git mechanics, reviewer outages. Load the one file the case needs; never load the whole library.
references/planning-with-files-adaptation.md
View on GitHub# Adaptation Reference: planning-with-files Port (2026-09-12)
Source: https://github.com/OthmanAdi/planning-with-files (OthmanAdi, MIT,
26.8k stars, v3.17.2; reviewed Sep 2026). Verdict on the repo: Watch —
strong engineering (706 tests, honest evals, real security-incident
disclosure), and it ships a NATIVE Hermes plugin (verified against Hermes
0.19.1, skills-guard SAFE). Decision (user, Sep 2026): option C — port the
concepts into this skill dependency-free; do NOT install the plugin.
## Ported
1. **3-file persistent loop state** — their `task_plan.md` / `findings.md`
/ `progress.md` pattern, adapted: renamed to `loop-plan.md` /
`loop-findings.md` / `loop-progress.md` (distinct from repo docs),
gitignored working state per this skill's release-hygiene rules,
promoted into the repo at delivery. `review-commit/references/persistent-loop-plan.md`.
2. **Phase-status + completion gate** — their `**Status:**` grammar and
gated-stop concept, implemented as `review-commit/scripts/loop_plan_check.py` with a
`--gate` exit code instead of a `pre_verify` hook. Runs at session
resume (resume pointer) and before loop completion.
3. **SHA-256 attestation** — their plan-tamper detection, applied to
builder briefs and Hermes-written test/plan files:
`scripts/attest_file.py attest|verify`. Turns pitfall #20's ad-hoc
diff-reading into a mechanical check.
## Explicitly NOT adopted
- **The native Hermes plugin** — per-turn `pre_llm_call` injection,
write reminders, `pre_verify` gate, `/pwf` commands. Standing ~1k
token/turn cost on EVERY session to defend against context rot we only
incur inside long loops; our loops read state on resume instead.
- **Injection-time attestation enforcement** — their hooks refuse to
inject a tampered plan. We attest on demand (attest/verify around
dispatches); no hook chain to enforce at.
- **Parallel-session write guard** — detects completed phases going DOWN
between turns from concurrent sessions. Our loops are
single-orchestrator; parallel BUILDERS never touch loop files.
- **PWF env knobs / .planning/<slug> isolation / .active_plan pointers** —
multi-plan fleet machinery for multi-project agent seats; overkill here.
- **KV-cache-stable injection formatting** — no injection, so moot.
## Revisit trigger
If a real loop shows context rot despite this pattern (orchestrator
resumes wrong, loses a review verdict, or re-derives settled decisions),
re-evaluate installing the plugin (option A):
`hermes skills install OthmanAdi/planning-with-files/.hermes/skills/planning-with-files --yes`
then the plugin route per their docs/hermes.md. Verify with `/plan-doctor`.
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: 323k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 323k | 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: 198k | Updated | Claude CodeCodex | Verified | Install |