Answer a few targeted questions about your workflow. Get a production-ready skill file for Claude, Copilot, ChatGPT, and every major AI agent.
---
name: ts-code-review
description: Reviews TypeScript and Next.js
code for type safety, naming
conventions, and project rules.
Triggers on "review this" or
"check this PR".
triggers:
- review this
- check this PR
- audit my code
---
# TypeScript Code Review
## What to check
- No implicit `any` — use proper generics
- Named exports only (no defaults in lib/)
- Conventional Commits for all messages
- No direct DOM manipulation in React
## Hard stops
- Never modify __tests__/ or *.test.ts files
- Never disable TypeScript strict mode
- Never commit directly to main
## Output format
Return a summary with:
- Issues found (severity: warn/error)
- Suggested fix for each issue
- Conventional Commit message draftExample output · TypeScript / code review
The process
Five steps. Under two minutes. The specificity of the questions is what separates production-ready output from generic boilerplate.
Pick from six workflow types — Development, Frontend, Content, Data, Project, or Custom. Each route has its own question set tailored to that domain.
Choose from Claude Code, Cursor, GitHub Copilot, ChatGPT, Windsurf, Codex CLI, Gemini CLI, or Universal. Platform drives the file format, install path, and trigger language.
Three to five questions specific to your category. No generic text boxes. The specificity of the questions is what makes the output production-ready.
SkillDraft calls the Anthropic API and assembles your skill file in real time. Takes about 10 seconds.
Drop it in your .claude/skills/ folder and the agent knows exactly what to do — and what never to do.