vet

AI coding tools are simultaneously the most loved developer tools and the largest source of new production bugs. vet audits the gap.

what it checks

check	what
ready	is your codebase AI-friendly? scans structure, docs, types, tests
diff	AI anti-patterns in git diffs: wholesale rewrites, orphaned imports, removed error handling
models	deprecated AI model strings across OpenAI, Anthropic, Google, Cohere
config	agent config completeness — CLAUDE.md, .cursorrules, copilot-instructions
history	git commit churn, AI attribution, large changes
scan	malicious patterns in cloned agent configs (.claude/, .cursorrules, .mcp/)
secrets	leaked API keys in build output + dev environment
receipt	post-session audit — what did the agent actually do?

usage

npx @safetnsr/vet                    # run all 8 checks
npx @safetnsr/vet --fix              # auto-fix configs, models
npx @safetnsr/vet --ci               # CI mode — exit 1 if below threshold
npx @safetnsr/vet --watch            # live monitoring during AI sessions
npx @safetnsr/vet receipt            # post-session agent receipt

zero deps. works with Claude Code, Cursor, Copilot, Codex, Aider, Windsurf, Cline.

links

• github
• npm