Run the model on your own computer and let it look things up on 13F.ai. Your questions stay with your model; only its requests for data reach us. A Basic account can't use them: the app is told it needs the Premium plan. The Basic plan covers Claude, ChatGPT, Gemini and Grok (Microsoft 365 Copilot coming soon).

  1. 1

    Get an API key

    Create one on your account page. It is shown once, so copy it somewhere safe. Treat it like a password.

  2. 2

    Add the server

    Use this address, with your key as a bearer token: https://13f.ai/api/mcp. Set-ups for Open WebUI, LM Studio and Claude Code are below.

  3. 3

    Ask

    Pick a model that supports tools, turn the 13F.ai tools on for the chat and ask.

Try these first

All three work on Basic.

Open WebUI

You need Open WebUI version 0.6.31 or newer.

  1. Open Settings, then Admin Settings, then Integrations and find External Tool Servers.
  2. Click + Add Connection.
  3. Set Type to MCP (Streamable HTTP) and the URL to:
    https://13f.ai/api/mcp
  4. Set Auth to Bearer and paste your key as the token. Give the connection a name such as 13F.ai, then save.
  5. Start a chat, pick a model that supports tools, and turn the 13F.ai tools on for that chat.
LM Studio

Add the server to LM Studio's mcp.json (in the app: Program tab, Install, Edit mcp.json):

{
  "mcpServers": {
    "13f-ai": {
      "url": "https://13f.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer <your key>"
      }
    }
  }
}

Replace <your key> with your key and save the file. LM Studio then loads the server and lists its tools.

Claude Code and other MCP apps

Claude Code takes one command, then you sign in from inside it:

claude mcp add --transport http 13f-ai https://13f.ai/api/mcp

Or send a key in a header, which suits scripts and apps that can't sign in:

claude mcp add --transport http 13f-ai https://13f.ai/api/mcp --header "Authorization: Bearer <your key>"

Any app that supports MCP over Streamable HTTP with a custom header will work. The developer docs describe the tools and the REST API.

A note on small models

Looking things up takes a model that is good at using tools. Many small local models aren't: they skip the lookup, make up details or invent an answer. If answers look wrong, try a larger or newer model that supports tools, and always check the source links in the results.

Third-party apps change often, so check their documentation if a menu has moved.