Thanks for checking out SchemaPin 🧷 — a free and Open Source simple but powerful protocol to secure your AI agents from a growing threat: tool schema tampering.

As LLM agents get smarter, they increasingly rely on external tools (APIs, scripts, functions). But here’s the problem: most agents blindly trust these tools — even if the schema changes unexpectedly or maliciously. We call one of the worst scenarios the "MCP Rug Pull", where a tool’s schema is subtly altered after initial approval.

SchemaPin fixes this with cryptographic signing and key pinning.
It’s like HTTPS for your agent’s toolchain.

We’d love your thoughts on the idea, the security model, or even your experiences building agents.
If you’re working on AI infra, LLM agents, or just care about supply-chain security, say hi — we’re listening!

— Jascha / ThirdKey.ai

SchemaPin Protocol

Like HTTPS - but for AI Agent tool schemas.