code-review_skill

Overview

This skill reviews TypeScript code changes against hard rules, conventions, and security checks to catch blocking issues before they reach the repository. It runs as a pre-push validator and produces a concise, actionable summary highlighting critical failures, warnings, and suggestions. The goal is fast, deterministic feedback so maintainers and contributors fix the root cause before merging.

How this skill works

The skill inspects staged or changed files and enforces build, lint, and test gates first (typecheck, lint, tests). It then scans diffs for hard rules (no any, no console.log, component size limits, no inline styles) and evaluates code quality, conventions, and security patterns. Results are categorized by severity and formatted into a short review summary with file/line references and remediation hints.

When to use it

• Automatically before every commit/push (pre-push hook)
• When creating or reviewing a pull request
• During CI to prevent merging broken changes
• As a self-review step before pushing large refactors
• When onboarding contributors to enforce project conventions

Best practices

• Run npm run typecheck && npm run lint locally before committing
• Fix critical issues listed under Critical before proceeding
• Keep components concise (prefer <200 lines) and extract helpers
• Avoid any and prefer precise TypeScript types or narrow unknown
• Store secrets in environment variables and validate inputs

Example use cases

• Rejecting a commit that introduces console.log or any in a lib file
• Catching a component that grew past 200 lines and suggesting refactor
• Detecting hardcoded API keys and flagging them as critical
• Ensuring new API code follows the repository's 3-layer pattern
• Adding lint and type errors to the review summary so CI can block merge