yao

Why we included this project

Yao suits teams that want one runtime to serve both conversational agents and the web app around them, skipping the usual backend assembly. It ships as a single Go binary with an embedded V8 engine, so you define data models and REST APIs in JSON/YAML, render server-side UI, and route each request to an LLM, a containerized CLI agent, or your own TypeScript code, mixing all three in one pipeline. The agent side is where it stands out: Create and Next hooks run before and after the model, letting you inject context and enforce constraints for deterministic control over open-ended AI behavior. Native MCP support and a four-scope memory API make connecting real tools and keeping state across a session straightforward. For small teams that want to deploy an AI product as a self-contained artifact rather than stitch together multiple services, the single-binary packaging is genuinely convenient.