Files
Rohit Ghumare 79573df7cb Initial release: 100-file Claude Code toolkit
20 specialized agents, 10 skills, 17 slash commands, 6 plugins,
12 hooks with scripts, 8 rule sets, 3 CLAUDE.md templates,
14 MCP server configs, and interactive setup installer.
2026-02-04 18:55:28 +00:00

2.7 KiB

/code-guardian:review

Perform a thorough code review of the current uncommitted changes or a specified file/directory.

Process

  1. Gather the changes to review:

    • Run git diff for unstaged changes and git diff --cached for staged changes
    • If a specific file or directory is provided as an argument, focus the review there
    • If no changes exist, review the most recent commit with git diff HEAD~1
  2. Analyze each changed file across these dimensions:

Correctness

  • Logic errors, off-by-one mistakes, incorrect boolean conditions
  • Unhandled null/undefined values and missing error boundaries
  • Race conditions in async code, missing awaits, unhandled promise rejections
  • Incorrect type assertions or unsafe casts

Security

  • User input flowing to SQL queries, shell commands, or file paths without sanitization
  • Hardcoded credentials, API keys, tokens, or connection strings
  • Missing authentication or authorization checks on new endpoints
  • Unsafe deserialization of external data

Performance

  • N+1 query patterns in database access
  • Unnecessary re-renders in React components (missing memoization, unstable references)
  • Unbounded data fetching without pagination or limits
  • Synchronous blocking operations on hot paths
  • Memory leaks from unclosed resources, event listeners, or subscriptions

Design

  • Single Responsibility violations: functions or classes doing too many things
  • Tight coupling between modules that should be independent
  • Missing abstractions where patterns repeat three or more times
  • Public API surface area: are new exports intentional and well-designed?

Readability

  • Variable and function names that obscure intent
  • Deeply nested conditionals that should be flattened or extracted
  • Magic numbers or strings that should be named constants
  • Missing or misleading type annotations
  1. Classify each finding by severity:

    • CRITICAL - Will cause bugs, data loss, or security vulnerabilities in production
    • WARNING - Likely to cause issues under certain conditions or during maintenance
    • SUGGESTION - Improvement opportunity that is not urgent
  2. For each finding, provide:

    • The file path and line range
    • A clear description of the issue
    • A concrete fix or code suggestion

Output Format

Present findings grouped by severity, then by file. Lead with critical issues. Skip categories that have no findings. End with a brief summary of overall quality.

Rules

  • Be specific and actionable, not vague ("this could be improved")
  • Do not flag style preferences unless they impact readability significantly
  • Acknowledge well-written code briefly at the end
  • Limit total findings to the 15 most impactful to avoid noise
  • If the diff is very large, prioritize new code over modified code