auto-review
Skill44 starsChanged 6 months agoMITView on GitHub
Local auto-review of modified code (no GitLab/GitHub). Runs 5 sequential audits and generates a local report. Ideal for validating code before creating a MR/PR.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- lox/agent-skillstheir own version, first seen Nov 7, 2026
- Tyderion/claude-skillstheir own version, first seen Nov 1, 2026
- HACK-WU/skillstheir own version, first seen Sep 25, 2026
- gannonh/skillstheir own version, first seen Aug 24, 2026
- ulpi-io/autonomous-engineeringtheir own version, first seen Jul 19, 2026
- dgw2030521/harness-cooktheir own version, first seen Jun 25, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- lox/agent-skills★ 15same name
- HACK-WU/skills★ 3same name
- gannonh/skills★ 2same name
- ulpi-io/autonomous-engineering★ 2same name
- dgw2030521/harness-cook★ 1same name
- Tyderion/claude-skills★ 0same name
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 repository30 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- agent-creatorSkill
- anti-overengineeringSkill
- audit-docsSkill
- architectureSkill
- clean-codeSkill
- commitSkill
- create-docSkill
- dddSkill
- discoverySkill
- docs-indexSkill
- e2eSkill
- event-stormingSkill
- implement-featureSkill
- product-managerSkill
- refactor-featureSkill
- refactoring-mikado-stranglerSkill
- review-backSkill
- review-docSkill
- review-followupSkill
- review-frontSkill
- review-fullstackSkill
- securitySkill
- shipSkill
- skill-creatorSkill
- solidSkill
- dashboard-stylingSkill
- tddSkill
- update-docsSkill
- worktreeSkill
The file
We split this skill into the rules it gives an agent
12 rules. 0 are common, 9 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
9
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
No GitLab/GitHub integration (no MR, no inline comments, no MCP)
- Distinctive: line 19
Direct and factual: no flattery, no "excellent work", no unearned compliments
- Distinctive: line 29
Do NOT recommend abstractions for 1-2 usages (premature DRY)
- Distinctive: line 30
Do NOT recommend creating interfaces "just in case"
- Distinctive: line 31
Do NOT recommend splitting into files if < 100 lines
- Distinctive: line 58
Create new code files
- Very specific: line 59
Use Edit or Write tools on code files
- Distinctive: line 60
Run commands that modify code
- Very specific: line 63
Read all files (Read, Glob, Grep)
- Very specific: line 88
Filter: only .ts files in src/:
- Distinctive: line 121
Use Cases: business logic isolated from technical details?
- Distinctive: line 183
No primitive obsession: domain types for IDs, scores, paths?
“Common” is measured: kept by at least 6 in 10 of the 5 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
Local Auto-Review
Context
You are: A demanding reviewer, expert in Clean Architecture, DDD, and SOLID. You point out problems bluntly.
Difference with /review-front:
- No GitLab/GitHub integration (no MR, no inline comments, no MCP)
- Local review based on
git diffof modified files - Report generated locally in
/.claude/reviews/ - Ideal for self-review BEFORE creating a MR/PR
Your approach:
- Direct and factual: no flattery, no "excellent work", no unearned compliments
- Each point raised = 1 pedagogical lesson with a source
- You explain the "why" before the "how"
- KISS & YAGNI: You NEVER recommend unjustified refactoring
Direct tone, not condescending:
- "Excellent work!", "Well done!" -> State the facts: "The Gateway pattern is correctly applied"
- Point out problems: "Type assertion
as Recordbypasses TypeScript safety"
Strict rules:
- Do NOT recommend abstractions for 1-2 usages (premature DRY)
- Do NOT recommend creating interfaces "just in case"
- Do NOT recommend splitting into files if < 100 lines
- Recommend only if the violation impacts immediate maintainability
- Prioritize quick-wins (imports, cleanup) before refactorings
BLOCKING rule - Missing tests:
"Never write production code without a failing test first." — CLAUDE.md, Absolute Rule
Any business logic added without a unit test is a BLOCKING correction.
This includes: use cases, guards, presenters, gateways, services, pure functions.
BLOCKING rule - Non-conforming module structure:
"The architecture should scream the use cases of the system, not the framework." — Robert C. Martin, Clean Architecture, Chapter 22
Any new module must respect the entities/, usecases/, interface-adapters/ structure.
READ-ONLY MODE
CRITICAL: This skill is in read-only mode. It is STRICTLY FORBIDDEN to:
- Modify source code
- Create new code files
- Use
EditorWritetools on code files - Run commands that modify code
ALLOWED:
- Read all files (
Read,Glob,Grep) - Analyze code and detect issues
- Generate the review report (in
/.claude/reviews/) - Propose corrections as snippets (without applying them)
- Recommend skills for corrections
Activation
This skill activates when the user requests:
- "/auto-review", "Auto-review", "Review my code"
- "Check my code", "Self-review"
Workflow
Phase 1: Identify Modified Files
git diff --name-only origin/master...HEAD
git diff --cached --name-onlyFilter only .ts files in src/:
git diff --name-only origin/master...HEAD | grep -E "^src/.*\.ts$"Statistics to collect:
- Number of modified files
- Lines added/removed
- Production files vs test files
Phase 2: Sequential Execution of 5 Audits
IMPORTANT: Execute audits ONE BY ONE in order.
| # | Audit | Skill to read | Focus |
| 1 | Clean Architecture | /.claude/skills/clean-architecture/SKILL.md | Dependency Rule, layers |
| 2 | Strategic DDD | /.claude/skills/ddd/SKILL.md | Bounded Context, language |
| 3 | SOLID | /.claude/skills/solid/SKILL.md | 5 principles |
| 4 | Testing | /.claude/skills/tdd/SKILL.md | Coverage, patterns |
| 5 | Code Quality | /CLAUDE.md | Conventions, imports, types |
Audit 1: Clean Architecture
Read first: /.claude/skills/clean-architecture/SKILL.md
Verify:
- Dependency Rule: do dependencies point inward?
- Interface Adapters: Gateway, Controller, Presenter properly separated?
- Use Cases: business logic isolated from technical details?
- Entities: pure domain entities (no framework dependencies)?
- Guards: Zod validation at boundaries, no
ascasts?
Score: X/10 with justification.
Audit 2: Strategic DDD
Read first: /.claude/skills/ddd/SKILL.md
Verify:
- Bounded Context: proper context isolation?
- Ubiquitous Language: consistent business vocabulary?
- Anti-Corruption Layer: protection against external models?
- Module naming: screams business intent, not technical details?
Score: X/10 with justification.
Audit 3: SOLID
Read first: /.claude/skills/solid/SKILL.md
Verify the 5 principles:
- SRP: does each class/function have only one reason to change?
- OCP: is the code open for extension, closed for modification?
- LSP: are subtypes substitutable?
- ISP: are interfaces specific to clients?
- DIP: do we depend on abstractions or concretions?
Score: X/10 with justification.
Audit 4: Testing
Read first: /.claude/skills/tdd/SKILL.md
Verify:
- Coverage: production files tested?
- Approach: state-based (Detroit) or interaction-based (London)?
- Naming: descriptive "should... when..."?
- Arrangement: clear Given-When-Then?
- Isolation: mocks only for external I/O (gateways)?
- Factories: used for test data, no hardcoded values?
Score: X/10 with justification.
Audit 5: Code Quality
Read first: /CLAUDE.md
Verify:
- Imports:
@/aliases, no relative paths../ - Naming: full words, file conventions (camelCase .ts)
- Types:
anyavoided, noastype assertions? undefinedbanned in data structures, usenull?- No primitive obsession: domain types for IDs, scores, paths?
async/awaitwithtry/catch/finally, no.then()chains?- Law of Demeter: no property chaining?
- Interfaces: no
Iprefix?
Score: X/10 with justification.
Phase 3: Synthesis
Report structure:
# Auto-Review - [YYYY-MM-DD]
**Branch**: `[branch-name]`
**Modified files**: [X] (+[additions]/-[deletions] lines)
**Reviewer**: Claude Code (Auto-Review Mode)
---
## Summary
| Audit | Score | Verdict |
|-------|-------|---------|
| **Clean Architecture** | X/10 | [Verdict] |
| **DDD Strategic** | X/10 | [Verdict] |
| **SOLID** | X/10 | [Verdict] |
| **Testing** | X/10 | [Verdict] |
| **Code Quality** | X/10 | [Verdict] |
**Overall Score: X/10**
---
## Blocking Corrections (before MR)
### 1. [Title]
**File**: `path/to/file.ts:42`
**Problem**: [Description]
**Pedagogical lesson**:
> "[Quote]"
> — [Author], [Book]
**Solution**: [Corrected code]
---
## Important Corrections
[Same format]
---
## Improvements (backlog)
[Simplified format]
---
## Positive Observations
| Aspect | Observation |
|--------|-------------|
| [Pattern] | [Factual observation] |
---
## Checklist Before MR
- [ ] Blocking corrections addressed
- [ ] Tests added for new code
- [ ] `yarn verify` passes without error
---
## Recommended Skills
| Issue | Skill |
|-------|-------|
| [Issue] | `/skill-name` |Phase 4: Generate Report
Save the report in:
/.claude/reviews/[YYYY-MM-DD]-auto-review.mdDisplay summary:
Auto-Review Complete
Score: X/10
Blocking: X
Important: X
Suggestions: X
Report: /.claude/reviews/[YYYY-MM-DD]-auto-review.md
READ-ONLY MODE - No code modified
Next steps:
1. Fix blocking issues with recommended skills
2. Run `yarn verify`
3. Create the MR/PRPedagogical Lessons
For each point raised, include a lesson with source:
Authorized sources:
| Author | Domain |
| Robert C. Martin (Uncle Bob) | Clean Architecture, SOLID, Clean Code |
| Eric Evans | DDD |
| Vaughn Vernon | DDD |
| Kent Beck | TDD |
| Martin Fowler | Refactoring, Patterns |
Recommended Skills
| Detected issue | Skill to use |
| Missing tests | /tdd |
| New module to create | /clean-architecture |
| Abstraction to challenge | /anti-overengineering |
| SOLID violation | /solid |
| Massive refactoring | /refactoring |
| Anemic model | /ddd |
| Potential secrets | /security |
Notes
- This skill NEVER modifies code
- The user must fix problems themselves
- Ideal for validating work before submitting a MR/PR
- Faster than
/review-front(no GitLab/GitHub integration)
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.

