review
General code review for staged changes or specified files. Use when reviewing pull requests, commits, or code changes across any language.
When & Why to Use This Skill
This Claude skill automates comprehensive code reviews for staged changes, pull requests, and specific files across any programming language. It systematically analyzes code for security vulnerabilities, runtime safety, logic correctness, and performance bottlenecks, ensuring high-quality software delivery while adhering to language-specific best practices and idiomatic standards.
Use Cases
- Security Auditing: Automatically detect critical vulnerabilities such as SQL injection, hardcoded secrets, and authentication bypasses during the development phase.
- Pull Request Automation: Streamline the peer review process by catching logic errors, edge cases, and style violations before human reviewers step in.
- Performance Optimization: Identify inefficient algorithms, unnecessary memory allocations, and missing caching opportunities in hot code paths.
- Standardization & Compliance: Ensure all code changes follow industry-standard style guides (e.g., PEP 8, Uber Go Style) and maintain high test coverage for new features.
- Runtime Safety Checks: Prevent common production crashes by identifying potential null pointer dereferences, race conditions, and resource leaks.
| name | review |
|---|---|
| description | General code review for staged changes or specified files. Use when reviewing pull requests, commits, or code changes across any language. |
| allowed-tools | ["shell", "read_file", "apply_patch"] |
| short-description | Review code changes comprehensively |
Code Review
Read these references:
~/.config/agent/domain/review/process.md- Review workflow~/.config/agent/domain/review/priorities.md- Review focus areas
Instructions
Review staged changes (or specified files/diff) in this priority order:
Note: When receiving review feedback, use the review-reception skill instead of this one.
Review Focus (Priority Order)
Security issues
- Injection vulnerabilities (SQL, command, XSS)
- Secrets exposure (hardcoded keys, tokens)
- Authentication/authorization bypasses
- Insecure cryptography, weak randomness
Runtime safety
- Nil/null pointer dereferences
- Race conditions, data races
- Resource leaks (file handles, connections, memory)
- Buffer overflows, out-of-bounds access
Code correctness
- Logic errors, incorrect algorithms
- Edge cases not handled
- Error handling: all errors checked and wrapped
- State management issues
Language idiomaticity
- Go: Uber Go Style Guide, effective Go patterns
- Rust: ownership patterns, lifetime management
- Python: PEP 8, pythonic idioms
- Follow language-specific best practices
Performance
- Unnecessary allocations (especially in hot paths)
- Inefficient algorithms (O(n²) where O(n) possible)
- Repeated expensive operations
- Missing caching opportunities
Test coverage
- Missing tests for new functionality
- Untested edge cases
- Insufficient error case testing
Review Process
Per AGENT.md Code Review:
- Be thorough and pedantic - catch every issue
- Present interactively - discuss each issue separately
- Order by severity - most critical first
- Offer fixes - for each issue, offer to make the change
- Prove when possible - run tests/lint to demonstrate issues
Arguments
Target: ${ARGUMENTS:-staged changes}