DevTools AI is pre-1.0 and developed on a single main branch. Security fixes are applied to the latest commit only — there are no maintained release branches yet.

Please do not open a public GitHub issue for security vulnerabilities.

Instead, report it privately through the Contact page (it opens a pre-filled GitHub issue you can redirect into a private channel) or via a GitHub Security Advisory on this repository.

Please include:

• A description of the vulnerability and its potential impact
• Steps to reproduce it
• Any relevant logs or proof-of-concept code

You should get an initial response within a few days.

• OPENROUTER_API_KEY is read only in src/lib/ai/openrouter-client.ts, a server-only module. It is never included in any response sent to the browser or the CLI's stdout.
• Model allowlisting — the /api/ai/[slug] route rejects any model value not present in src/config/models.ts, so a client can't force arbitrary model IDs or parameters through to OpenRouter.
• No server-side storage — there is no database and no logging of request/response content, so there's nothing to leak on that front. See Privacy for the full picture.
• Input validation — every tool's input is validated with Zod, both on the web API route and in the CLI, before it's used to build a prompt.

• No rate limiting is implemented on /api/ai/[slug]. A naive in-memory limiter would silently stop working across multiple serverless instances, which is worse than not having one — so instead of shipping something that looks safe but isn't, this is left as an explicit gap. If you deploy a public instance, put a real rate limiter (e.g. Upstash, or your platform's edge/WAF rules) in front of it.
• This project has not had a formal third-party security audit. Use your own judgment before deploying a public instance with a funded API key.