Connect an AI assistant

What's Moving is also an MCP server: your assistant reads the archive the app shows and answers your questions from it. Every connection is tied to your What's Moving account.

Endpoint

https://api.whatsmoving.app/api/mcp

Sign in (Claude)

Open What's Moving in the Claude directory and press Connect, then sign in with your What's Moving account.

Or add it by hand: in Claude, open Customize → Connectors → Add → Add custom connector, paste the endpoint above, keep the detected defaults, add it and connect.

Sign in (Claude Code)

Add the server:

claude mcp add -s user --transport http whatsmoving https://api.whatsmoving.app/api/mcp

Then start Claude Code, run /mcp, choose whatsmoving and authenticate with your What's Moving account.

Sign in (ChatGPT desktop app and Codex)

In the ChatGPT desktop app, open Settings → Plugins → Add → Add MCP server. Name it, choose Streamable HTTP, paste the endpoint above and save, then press Authenticate on the server row and sign in with your What's Moving account.

In Codex, add the server and sign in:

codex mcp add whatsmoving --url https://api.whatsmoving.app/api/mcp
codex mcp login whatsmoving

Access token (any MCP client)

For any MCP client that sends an Authorization: Bearer header. In the app, go to Account → AI assistants → New token. Treat it like a password: it is shown once, and you revoke it in the app.

Keep it in an environment variable, not in a config file, and set it wherever you start your client. In Claude Code — paste the token at the silent prompt, then press Enter:

read -rs WHATSMOVING_TOKEN && export WHATSMOVING_TOKEN
claude mcp add -s user --transport http whatsmoving https://api.whatsmoving.app/api/mcp --header 'Authorization: Bearer ${WHATSMOVING_TOKEN}'

What you get

The same archive and the same tier as your app account. On the free tier, new content arrives with a 15-minute delay and history reaches back 30 days; Pro has neither limit.

Post-hoc analysis only — no predictions, no investment advice.