Just launched Vard - TypeScript prompt injection detection

What is it: Pattern-based prompt injection detection for TypeScript. Zod-inspired API, <0.5ms latency, ~90% detection rate.

Why I built it: Kept writing the same validation code in every LLM app.

Open source, MIT licensed. Built for developers who need fast local validation without LLM API calls.

Would love feedback from other founders building with LLMs - what security measures are you using?

GitHub: github.com/andersmyrmel/vard

npm: @andersmyrmel/vard