How to Build an AI Agent from Your Existing REST API — Without an MCP Server
How to Build an AI Agent from Your Existing REST API — Without an MCP Server You already have a REST API. You don't want to rewrite it, run a dedicated protocol server, or learn JSON-RPC just to let an AI agent call GET /orders or POST /users . RCP (REST Connector Protocol) is the shortest path: add one GET /manifest route that describes your endpoints as tools. Your AI agent fetches it and calls…
You already possess a REST API but don't want to rewrite it, set up a dedicated protocol server, or learn JSON-RPC just to enable an AI agent to call your existing endpoints like GET /orders or POST /users. RCP (REST Connector Protocol) offers a simple solution: add one GET /manifest route that describes your API endpoints as tools. Your AI agent retrieves this manifest and directly calls your API using stateless HTTP, without requiring a persistent connection.
RCP is a recent development that most AI agents like ChatGPT or Claude don't natively support. However, your AI agent can still handle it by fetching the manifest and exposing the tools to the model through the OpenAI SDK, LangChain, or Gemini. That's all it takes.
For a complete comparison between RCP and MCP (Multi-Modal Connector Protocol), refer to the RCP documentation. MCP is ideal when you need rich, stateful capabilities such as resources, prompts, elicitation, sampling, and a long-lived session. These features come at the cost of a protocol server, JSON-RPC, and transport negotiation.
RCP is suitable for the more common scenario: you have a REST API and want an AI agent to call some of its endpoints as tools, without running anything beyond your existing API plus an additional route that returns JSON. If your needs grow beyond this, consider MCP as a real signal, not a gap RCP tries to fill.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.