MCP overview

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data. Mereon runs an MCP server, so any MCP-capable assistant can search and read your knowledge base without you writing any code.

This is the path to choose if you want an assistant like Claude Code or Cursor to answer using your company’s real processes, rather than guessing.

How it works

You point your assistant at one URL and give it your token. From then on, the assistant has a set of Mereon tools it can call on its own, like search_mereon and get_topic. When you ask it something about how your business works, it looks the answer up in Mereon and cites the source.

• Endpoint: https://api.mereon.ai/v1/mcp
• Transport: Streamable HTTP
• Auth: the same personal access token you would use for the API, sent as an Authorization: Bearer header
• Access: read-only, scoped to exactly what you can see in Mereon

Tip

The MCP server and the REST API are the same surface underneath. Anything a tool can read here, you can also fetch directly from the API, and both respect the same permissions.

Set it up

Connect Claude Code Step-by-step setup for Claude Code, with a ready-to-paste instruction snippet for your repositories.

Tools reference Every tool the assistant gains, what it does, and the inputs it accepts.

Other MCP clients

Any client that supports a remote, Streamable HTTP MCP server with a bearer token will work. Point it at https://api.mereon.ai/v1/mcp and set the header:

Authorization: Bearer mrn_...

The exact steps differ per client (Cursor, Claude Desktop, and others each have their own settings screen for adding a remote MCP server), but the URL, the header, and the tools are the same everywhere.