Last week I built an MCP tool that gives agents access to search for code snippets on nostr.

The change in the output LLMs produce with it is MASSIVE; they no longer worked based on a large blob of unweighted code; now I can point them to see EXACTLY how I do a particular implementation under very precise conditions.

Now whenever I see the LLM making a mistake I correct it, publish a code snippet, and the LLM never makes the mistake again.

It’s absolutely wild to see LLMs using interfaces in NDK I literally just implemented and getting their uses right first time.

This is not only useful for nostr development.

This is useful for ALL development.

And requires cryptographic identity, web-of-trust, censorship-resistance and money.

All building blocks that are in our very robust toolbox.

The future is very bright