MCP setup

Use the Trace MCP server when you want an AI agent to query Trace directly as tools instead of calling the REST API by hand.

Server settings

Setting	Value
Server URL	`https://mcp.tracedata.ai/mcp`
Transport	Streamable HTTP
Authentication	OAuth 2.1 authorization-code flow with PKCE
Client registration	Dynamic Client Registration
Required scope	`mcp:read`

There is no API key field for MCP. The client signs you in through your Trace account in a browser, stores the OAuth token, and attaches it on tool calls. Your Trace email must be verified. MCP is available on every plan.

Available tools

The server exposes tools for:

Company lookup, technology stacks, jobs, partnerships, and data flags.
Technology and vendor search, plus reverse “companies using” queries.
Lookalike target-list generation.
Saved Lists: list discovery, member reads, definition authoring, materialization, permissions, ownership transfer, and domain removal.

See MCP tool reference for the full list, arguments, return shapes, and metering.

Claude Desktop

Use Claude Desktop’s native custom connector flow when available:

Open Settings -> Connectors.
Click Add custom connector.
Enter https://mcp.tracedata.ai/mcp.
Complete the Trace OAuth sign-in.

If you need the mcp-remote fallback, add this to claude_desktop_config.json:

{
  "mcpServers": {
    "trace": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.tracedata.ai/mcp",
        "--transport",
        "http-only"
      ]
    }
  }
}

Claude Code

Add the remote HTTP server:

claude mcp add --transport http trace https://mcp.tracedata.ai/mcp

Then run /mcp inside a Claude Code session or run:

claude mcp login trace

The first connection opens a browser sign-in. After approval, Claude Code stores the token in your OS keychain.

Other clients

Trace also works with Claude.ai, Cursor, VS Code, and ChatGPT where custom MCP connectors are available. Use the client-specific guides under Connect your client.