review-fix
Skill19 starsChanged 9 months agoMITView on GitHub
Read the latest code review, plan fixes for all findings, then execute the fixes — all in one command. Produces a dated plan and execution log in .reviews/.
11 other projects use a version of this.
Adopted by
Most recently taken up by
- davidvornholt/mail-mcptheir own version, first seen Nov 15, 2026
- davidvornholt/mailboxtheir own version, first seen Nov 15, 2026
- nwiizo/ccswarmtheir own version, first seen Nov 6, 2026
- raine/skillstheir own version, first seen Nov 1, 2026
- 200iqlabs/shared-skillstheir own version, first seen Oct 31, 2026
- pcode-xyz/simple-skillstheir own version, first seen Oct 20, 2026
Used with
Projects with it also use review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
SentrySkillreview
Fixture review skill.
Also in 5 projects that use this
Other versions11 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 11
- nwiizo/ccswarm★ 156same name
- DasDigitaleMomentum/opencode-processing-skills★ 60same name
- enes-alatas/spool★ 48same name
- EternallLight/tgc-skills★ 11same name
- 200iqlabs/shared-skills★ 4same name
- pcode-xyz/simple-skills★ 4same name
- davidvornholt/mail-mcp★ 2same name
- davidvornholt/mailbox★ 2same name
- raine/skills★ 2same name
- marcuskrogh/skills★ 1same name
1 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- cachibot-api-routeSkill
- cachibot-frontend-viewSkill
- cachibot-full-stack-entitySkill
- cachibot-pluginSkill
- cachibot-websocket-eventSkill
- code-reviewSkill
- create-prSkill
- lint-fixSkill
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 4 are distinctive, and 6 are very specific to this project.
0
Common
Most of the 10 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Read the most recent .reviews/YYYY-MM-DD-review.md, plan fixes for all findings grouped by priority, then execute every planned fix with …
- Very specific: line 17
Read the full review file.
- Very specific: line 19
If no review file exists, tell the user to run /code-review first and stop.
- Very specific: line 51
After all 4 agents return, compile results into .reviews/YYYY-MM-DD-review-plan.md:
- Distinctive: line 95
Read the plan file. Execute fixes wave by wave — all groups in a wave run in parallel, but waves run sequentially (wave 2 waits for wave 1 …
- Distinctive: line 109
After each wave completes:
- Very specific: line 111
Verify no lint errors remain (ruff check cachibot/ and cd frontend && npm run lint).
- Distinctive: line 112
If lint fails, spawn a fix agent for the failing files before continuing.
- Very specific: line 123
After all waves complete, write .reviews/YYYY-MM-DD-review-execution.md:
- Distinctive: line 191
If the review file has a "Previous Reviews" section noting still-open findings from prior reviews, include those in the plan too.
“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Review Fix — Plan & Execute from Last Review
Read the most recent .reviews/YYYY-MM-DD-review.md, plan fixes for all findings
grouped by priority, then execute every planned fix with atomic commits. Produces
two artifacts: a review-plan and a review-execution log.
Instructions
Phase 0 — Find the Latest Review
- Glob for
.reviews/*-review.mdand pick the most recent file by date. - Read the full review file. Parse every finding into a list of
{ severity, reviewer, file, line, description }entries. - If no review file exists, tell the user to run
/code-reviewfirst and stop.
Phase 1 — Plan
Spawn 4 parallel planning agents (Task tool, subagent_type: Plan) that each
receive the full findings list filtered to their scope. Each agent returns a
structured list of fix actions.
Agent scopes
| Agent | Scope |
| backend | Findings from Tina, Marcus, Priya, Carlos, Oscar that touch cachibot/ |
| frontend | Findings from Tina, Zoe, Luna that touch frontend/src/ |
| data | Findings from Carlos (storage/db layer) |
| tests | Findings from Derek (test coverage gaps) |
Findings that span multiple scopes go to every relevant agent.
Each planning agent must:
- Group related findings that can be fixed together (same file, same pattern).
- For each group, produce:
- What: one-line summary of the change
- Files: list of files to modify
- How: 2-3 sentence implementation approach
- Severity: highest severity in the group (Fix > Improve > Note)
- Risk: Low / Medium / High (does this change behavior or just style?)
- Depends on: other group IDs this group must wait for (if any)
- Order groups by: Fix severity first, then by dependency order.
- Skip any finding rated Note — notes are informational only.
After all 4 agents return, compile results into .reviews/YYYY-MM-DD-review-plan.md:
# Review Fix Plan — YYYY-MM-DD
**Source review:** `.reviews/YYYY-MM-DD-review.md`
**Generated:** YYYY-MM-DD HH:MM
## Stats
| Severity | Findings | Planned groups |
|----------|----------|----------------|
| Fix | N | N |
| Improve | N | N |
| Note | N | (skipped) |
## Execution Waves
Wave 1 (no dependencies):
- [B1] Backend: <summary>
- [F1] Frontend: <summary>
Wave 2 (depends on wave 1):
- [B2] Backend: <summary> (depends on B1)
...
## Detailed Plan
### [B1] <Summary>
- **Severity:** Fix
- **Risk:** Low
- **Files:** `path/to/file.py`
- **How:** Description of the fix approach
- **Findings addressed:**
- Tina: `path/to/file.py:42` — original finding text
- Marcus: `path/to/file.py:50` — original finding textPrint the plan summary (wave count, group count, severity breakdown) to the user, then immediately proceed to Phase 2 without waiting for approval.
Phase 2 — Execute
Read the plan file. Execute fixes wave by wave — all groups in a wave run in parallel, but waves run sequentially (wave 2 waits for wave 1 to finish).
For each group, spawn a Task agent (subagent_type: general-purpose) with:
- The group's detailed plan (What, Files, How)
- The original finding text for context
- Instruction to read each file before editing
- Instruction to make the minimal change that addresses the finding
- Instruction to run
ruff check --fixandruff formaton any modified Python file - Instruction to run
npm run lint -- --fixon any modified TypeScript file (fromfrontend/) - Instruction to NOT create new files unless the plan explicitly says to
- Instruction to NOT add tests (Derek findings are tracked but test writing is a separate task)
After each wave completes:
- Verify no lint errors remain (
ruff check cachibot/andcd frontend && npm run lint). - If lint fails, spawn a fix agent for the failing files before continuing.
- Stage all changed files and create one commit per wave:
Review fixes wave N: <comma-separated group IDs> Addresses findings from YYYY-MM-DD code review. Groups: [B1] summary, [F2] summary, ... Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After all waves complete, write .reviews/YYYY-MM-DD-review-execution.md:
# Review Fix Execution — YYYY-MM-DD
**Plan:** `.reviews/YYYY-MM-DD-review-plan.md`
**Source review:** `.reviews/YYYY-MM-DD-review.md`
## Results
| Wave | Groups | Status | Commit |
|------|--------|---------|---------|
| 1 | B1, F1 | Done | abc1234 |
| 2 | B2 | Done | def5678 |
## Findings Addressed
| Severity | Planned | Fixed | Skipped | Reason |
|----------|---------|-------|---------|-----------------|
| Fix | N | N | N | |
| Improve | N | N | N | |
## Group Details
### [B1] <Summary> — Done
- **Commit:** abc1234
- **Files changed:** `path/to/file.py`
- **Findings fixed:**
- Tina: `path/to/file.py:42` — fixed
- Marcus: `path/to/file.py:50` — fixed
### [F1] <Summary> — Skipped
- **Reason:** File was recently refactored, finding no longer applies
- **Findings skipped:**
- Zoe: `frontend/src/Component.tsx:100` — N/A after refactor
## Remaining Work
- Derek's test coverage findings were not addressed (test writing is separate)
- N findings skipped due to: ...Phase 3 — Summary
Print a final summary to the user:
Review fix complete.
Waves executed: N
Commits created: N
Findings fixed: N / M planned (X Fix, Y Improve)
Skipped: N (reasons listed in execution log)
Plan: .reviews/YYYY-MM-DD-review-plan.md
Execution: .reviews/YYYY-MM-DD-review-execution.mdImportant Rules
- Never skip a Fix-severity finding without logging a clear reason.
- Read before edit — every agent must read the target file before modifying it.
- Minimal changes — fix exactly what the finding describes, nothing more.
- No new features — this is a fix pass, not a feature pass.
- No test writing — Derek's findings are logged as "remaining work" but not executed here. Test creation deserves its own focused session.
- Preserve behavior — if a fix would change external API behavior, mark it as Risk: High in the plan and add a note in the execution log.
- If the review file has a "Previous Reviews" section noting still-open findings from prior reviews, include those in the plan too.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

