nestia

Why we included this project

Teams already running NestJS backends get a fairly direct route from their existing typed API to both a client SDK and an LLM assistant. You write ordinary TypeScript controllers, and nestia reads those types to generate a type-safe fetch SDK, a fuller Swagger document, and a mockup simulator for frontend work. On the AI side, the same Swagger document feeds @agentica, which turns it into LLM function calling, so users can get answers by talking to the backend instead of clicking through the UI. The typia-based validation loop is the part worth noticing: it catches the type-level mistakes LLMs tend to make when composing function arguments and retries with detailed errors, which makes the agent noticeably more reliable. If you want a chatbot that actually calls your real endpoints rather than a standalone chat wrapper, this gives you a workable path from controller to agent.