Connect the MCP server

One remote endpoint, Streamable HTTP, stateless. The data and free tools need no token: paste the URL and ask.

Endpoint: https://mcp.arcadiumdata.io/mcp
Auth: None needed (14 tools, 25 calls a day per IP). Optional Authorization: Bearer <key> with a free Arcadium Data API key for your plan's limits. The live scraping tools need an Apify token instead.
Server: io.arcadiumdata/mcp v0.3.0, server card at /.well-known/mcp/server-card.json

Claude (claude.ai and Claude Desktop)

Open Settings → Connectors and choose Add custom connector.
Name: Arcadium Data. URL: https://mcp.arcadiumdata.io/mcp. Leave the auth fields empty.
In a chat, enable the connector from the tools menu and ask, for example: "What changed at spotify.com in the last 30 days?"

To send your API key from Claude Desktop, use the mcp-remote bridge in claude_desktop_config.json:

{
  "mcpServers": {
    "arcadium-data": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.arcadiumdata.io/mcp",
               "--header", "Authorization: Bearer ${ARCADIUM_API_KEY}"],
      "env": { "ARCADIUM_API_KEY": "<your key>" }
    }
  }
}

Claude Code

claude mcp add --transport http arcadium-data https://mcp.arcadiumdata.io/mcp

# optional: your API key for your plan's limits
claude mcp add --transport http arcadium-data https://mcp.arcadiumdata.io/mcp \
  --header "Authorization: Bearer $ARCADIUM_API_KEY"

Check with claude mcp list; in a session, /mcp shows the tools.

Cursor

Add to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project). Drop the headers line to start without a key.

{
  "mcpServers": {
    "arcadium-data": {
      "url": "https://mcp.arcadiumdata.io/mcp",
      "headers": { "Authorization": "Bearer ${env:ARCADIUM_API_KEY}" }
    }
  }
}

ChatGPT

In Settings → Apps & Connectors → Advanced settings, turn on Developer mode (available on the plans that support custom connectors).
Create a connector: name Arcadium Data, MCP server URL https://mcp.arcadiumdata.io/mcp, authentication No authentication.
Pick it in a chat from the tools menu. Every tool is marked read-only.

VS Code, Windsurf, Cline and your own agent

Any client that speaks remote MCP over Streamable HTTP works with the URL above. VS Code (.vscode/mcp.json):

{
  "servers": {
    "arcadium-data": { "type": "http", "url": "https://mcp.arcadiumdata.io/mcp" }
  }
}

Raw JSON-RPC, to test from a shell:

curl -s https://mcp.arcadiumdata.io/mcp \
  -H 'content-type: application/json' -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"company_pulse","arguments":{"domain":"spotify.com"}}}'

Tag where you found us with ?src=<channel> on the URL or an X-Arcadium-Src header. We count calls per channel; we never log your inputs, key or IP address.

Limits and errors

No key: 25 calls a day per IP for the data and free tools. With an API key: your plan's limits.
Every answer is JSON with status (done, not_found, invalid_input, rate_limited with retryAfterSecs, auth_required), results, a note with the as-of date and coverage, and quota (calls left).
Live tools on Apify's own endpoint (https://arcadium--arcadium-mcp.apify.actor/mcp) need an Apify token on every request; runs are billed per result on that Apify account.

Tool reference

Generated from the server's own tools/list (v0.3.0, 22 tools). Every tool declares readOnlyHint: true and destructiveHint: false.

Data tools: our own database, no account needed

`company_profile` Company profile from our database

One-call company profile read from our own database (not a live scrape; company_hiring is the live one): hiring board and open jobs, daily hiring signals, tools named in its job ads, recent hiring events, Shopify store if any. Answers carry an as-of date; refreshed daily.