Skip to content

Guides

What is a trading MCP, and how does it give your AI agent a plan?

MCP is the open standard that connects AI assistants to tools. Here's how a trading MCP works, what it can and can't do, and how it fits with your broker.

MCP in one paragraph

The Model Context Protocol (MCP) is an open standard for connecting AI assistants such as Grok Bot, OpenCode, OpenClaw, ChatGPT or Claude to external tools. A tool publishes an MCP server; the assistant connects to it with a URL and a sign-in, discovers what the server offers, and calls it when it needs to.

Two MCPs: the plan and the execution

Trading with an agent usually takes two MCP servers. One gives the plan: what to buy or sell, when and with which exits. The other is your broker's MCP, which reads your account and sends the orders. Órbita is the first kind: it never touches your money or your broker keys.

In practice, the agent reads your cash and positions from the broker, asks Órbita for the next orders with those numbers, and sends them through the broker's MCP. Órbita answers with exact quantities and prices, adapted to what your broker accepts (stop orders, OCO, fractional shares).

What a trading MCP doesn't do

It doesn't guarantee results and it isn't personalized advice: Órbita publishes the same plan for everyone and applies fixed, public rules to your figures. You decide whether to connect an agent, and you're responsible for supervising it.

General information, not personalized investment advice. Trading involves risk of loss. Past or simulated results don't guarantee future results.