What is MCP tool?

Tools are the MCP feature that lets a model do things: query a database, call an API, or run a computation. A server that supports tools declares the tools capability. A client discovers them with a tools/list request and runs one with tools/call, passing the tool name and arguments. Each tool definition has a unique name, a description, and an inputSchema written in JSON Schema. It may also have a title, an outputSchema and annotations.

The specification calls tools model-controlled, meaning the model can find and call them on its own based on context. It still recommends that a person can always refuse a tool call, so applications should show which tools are exposed, signal when one runs, and ask for confirmation.

Results can hold text, images, audio, resource links or embedded resources, plus structured content that follows the output schema. Errors come in two kinds. Protocol errors are standard JSON-RPC errors for things like an unknown tool. Tool execution errors are returned in the result with isError set to true, so the model can read the message and retry.

Clients must treat tool annotations as untrusted unless they come from a trusted server. Servers must validate inputs, apply access controls, rate limit calls and sanitize outputs.