thoreinstein/agents/code-review
Claude Code and agent skill by thoreinstein/agents
- community
Checking that the SKILL.md file is still in the repository.
Comprehensive code review with security, performance, and maintainability focus. Produces structured review with APPROVE, NEEDS WORK, or BLOCK verdict.
Install
npx vskill@latest install thoreinstein/agents/code-reviewSKILL.md
Code Review Mode
Perform a comprehensive code review with structured analysis across multiple dimensions.
When to Use This Skill
- When reviewing pull requests or merge requests
- Before merging feature branches into main/trunk
- When asked to review specific files or changes
- During pair programming review sessions
- For post-implementation quality checks
Workflow
Phase 1: Context Gathering
- Identify the scope:
- Determine which files/commits to review (diff, specific files, or branch comparison)
- Get the diff:
git difffor the relevant range
- Understand the change:
- Review commit messages and PR description (if available)
- Check git history for related recent changes
- Identify the purpose and expected behavior of the change
- Map dependencies:
- Identify what other code depends on the changed files
- Check for related test files
- Note any configuration or schema changes
Phase 2: Multi-Dimensional Review
Evaluate the code against each dimension in the Review Dimensions section below. For each dimension:
- Walk through the relevant checklist
- Note any issues with severity (Critical/High/Medium/Low)
- Record the specific location (file:line)
- Document suggested fixes
Phase 3: Synthesis
Produce a structured review following the template in references/code-review-template.md.
- Determine verdict:
- APPROVE - No critical/high issues, code is ready to merge
- NEEDS WORK - Has issues that must be addressed before merge
- BLOCK - Has critical issues, security vulnerabilities, or fundamental design problems
- Organize findings by severity
- Acknowledge strengths - what's done well
- Provide actionable fixes - not just problems, but solutions
Review Dimensions
Correctness
- [ ] Logic is correct and handles all expected cases
- [ ] Edge cases are handled (null, empty, boundary values)
- [ ] Assumptions are documented or validated
- [ ] Error states are handled appropriately
- [ ] Types are correct and conversions are safe
Security
- [ ] No injection vulnerabilities (SQL, command, XSS)
- [ ] Input is validated and sanitized
- [ ] No secrets, credentials, or keys in code
- [ ] Authentication/authorization checks are correct
- [ ] Dependencies are secure and up-to-date
Performance
- [ ] No unnecessary loops or redundant operations
- [ ] Algorithms are appropriate for the data size
- [ ] Memory usage is reasonable
- [ ] No N+1 queries or unbounded fetches
- [ ] Caching is used appropriately
Reliability
- [ ] Errors are caught and handled gracefully
- [ ] Failures don't leave system in bad state
- [ ] Retry logic has backoff and limits
- [ ] Resources are properly cleaned up (connections, files, etc.)
- [ ] Timeouts are set for external calls
Maintainability
- [ ] Code is readable and self-documenting
- [ ] Functions/methods have single responsibility
- [ ] Naming is clear and consistent
- [ ] No unnecessary duplication
- [ ] Follows project conventions and patterns
Testing
- [ ] Tests exist for new/changed behavior
- [ ] Edge cases and error paths are tested
- [ ] Tests are deterministic (not flaky)
- [ ] Test names clearly describe what they verify
- [ ] Mocks/stubs are appropriate and not excessive
Escalation Criteria
Flag for additional human review when:
| Concern | Escalation Trigger | |
|---|---|---|
| Security | Any auth changes, crypto usage, user data handling, or potential vulnerabilities | |
| Reliability | Changes to error handling, retry logic, or failure recovery paths | |
| Performance | Changes to hot paths, database queries, or algorithms with scale concerns | |
| Architecture | New patterns, significant structural changes, or cross-cutting concerns | |
| Testing | Reduced coverage, disabled tests, or changes to test infrastructure |
Constraints
- Review only what's in scope - don't expand to unrelated code unless it's directly affected
- Be specific - reference exact file:line locations for all findings
- Provide fixes, not just criticisms - every issue should have a suggested resolution
- Calibrate severity appropriately - not everything is critical
- Acknowledge good work - positive feedback matters too
Examples
Example: Reviewing a New API Endpoint
Scope: PR adds POST /api/users endpoint in internal/api/users.go
Phase 1 output:
Phase 2 findings:
| Severity | Issue | Location | Fix | |
|---|---|---|---|---|
| High | SQL injection via unsanitized email | db/users.go:45 | Use parameterized query | |
| Medium | Missing rate limiting | api/users.go:23 | Add rate limiter middleware | |
| Low | Error message exposes internal details | api/users.go:67 | Return generic error to client |
Phase 3 verdict:
Begin by gathering context on the target files/commits before conducting the multi-dimensional review.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/code-review/SKILL.md
Popularity
- Repository stars
- 3
- 7-day trend score
- 7
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 24 Mar 2026, 7 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/thoreinstein/agents/code-review)