rfs-ai · v1

A common API for agentic AI

One backend for streaming chat, tool calls, hierarchical prompts, and quota-aware inference — wire it to any product, on any platform.

Server-sent events for token-by-token delivery, with non-streaming fallback when you'd rather have the whole reply.

Async tool calls with heartbeats. Wire in any tool source via the MCP gateway — your own backend, any platform.

Per-product, per-tenant, and per-user prompt overrides resolved at request time. Bring your own product.