code-assistant_skill

Overview

This skill is a programming assistant that analyzes code, finds bugs, suggests optimizations, refactors, and generates documentation and tests automatically. It supports multiple languages (Python, JavaScript/TypeScript, Go, Java, Rust, C/C++) and integrates with code agents for complex tasks. The tool focuses on practical fixes: security, performance, style, and logic issues.

How this skill works

The assistant inspects files or directories, reporting metrics (lines, functions, complexity) and highlighting code smells and vulnerabilities like SQL injection, race conditions, and hardcoded secrets. It can propose or apply fixes, suggest refactors, generate docstrings/README, and create unit tests or edge-case lists. For heavy operations it delegates to external coding agents (Codex, Claude, etc.) and follows configurable thresholds and ignore rules.

When to use it

• Perform automated code reviews before merges or releases
• Identify and remediate security vulnerabilities and sensitive leaks
• Reduce technical debt via targeted refactoring runs
• Generate documentation and unit tests for legacy modules
• Optimize hotspots: slow algorithms, large bundles, or heavy queries

Best practices

• Run analyses as part of CI to catch issues early
• Set sensible complexity and style thresholds (e.g., CODE_MAX_COMPLEXITY)
• Use dry-run mode before auto-fixing to review changes
• Limit scope with ignore patterns to avoid noise (node_modules, dist)
• Combine automated suggestions with human review for design changes

Example use cases

• Scan a repository to list potential SQL injection and race conditions
• Refactor legacy codebase to enforce PEP8/ESLint rules and extract functions
• Generate docstrings and a README for an unmaintained project
• Produce unit tests and edge-case suggestions for critical modules
• Optimize SQL queries and reduce frontend bundle size