claude

Claude Code pre-launch security checklist

Claude Code is increasingly the AI behind serious side projects, but a long session can drift away from security defaults set early. These twelve checks are tied to scanner findings — paste the prompt at the bottom into Claude Code and let it audit its own work.

12 checks

Tick through each one

Click any item to open the detailed fix guide. The detector tag shows which scanner module flips the state pass or fail when you run a SafeToShip scan.

Paste this into claude

One prompt that runs the entire checklist as a code review pass.

Audit the codebase you wrote against this 12-point security checklist: (1) no API keys in client code, (2) .env not deployed, (3) .git not deployed, (4) CSP, HSTS, X-Frame-Options, X-Content-Type-Options headers all set, (5) CORS allowlist explicit origins, (6) cookies Secure+HttpOnly+SameSite, (7) TLS valid, (8) every protected server route runs an auth check, (9) rate limiting on auth + write endpoints. For each item, report PASS/FAIL with the file path of any failure, then propose fixes.

FAQ

Frequently asked questions

Will Claude Code refuse to write insecure code?
Claude tends to follow security best practices when prompted, but if your codebase already has insecure patterns (NEXT_PUBLIC_ secrets, wildcard CORS), it will follow your patterns. Establish secure defaults early and reference them in subsequent prompts.

Run the scan to confirm each check

60 seconds. Free. No account required.