Bigeye MCP Server (Beta)

The Bigeye MCP Server connects Bigeye to AI assistants and agents through the
Model Context Protocol (MCP). Once it's connected, your
assistant can work with Bigeye directly: check data quality issues, trace lineage, search the
catalog, profile tables, and manage monitors and dimensions. No more copying results out of the UI
and pasting them into chat.

The quickest way to connect is the Bigeye MCP Gateway, a hosted endpoint at
https://mcpgateway.bigeye.com/mcp. Point your MCP client at that URL, pass your Bigeye credentials
as request headers, and you're connected. There's no Docker image to build and nothing to run
locally.

This guide covers connecting these clients to the gateway:

• Claude Code
• Snowflake Cortex Code (CoCo)
• GitHub Copilot CLI

Want to run the server yourself with Docker or Python instead? See Run it yourself.

What you'll need

Connecting through the gateway takes two values from Bigeye, plus a third in rare cases:

These three values map to the three request headers the gateway uses to authenticate each request:

In the examples below, swap in your own values for <YOUR_API_KEY> and <YOUR_WORKSPACE_ID>.

Connect your client

Claude Code

Register the gateway with one command:

claude mcp add --transport http bigeye https://mcpgateway.bigeye.com/mcp \
  --header "Authorization: apikey <YOUR_API_KEY>" \
  --header "x-bigeye-workspace-id: <YOUR_WORKSPACE_ID>"

On a non-default stack, add one more header:

  --header "x-bigeye-url: https://<your-stack>.bigeye.com"

By default the server is added for the current project only. Use --scope user to make it available
in every project, or --scope project to write it to a shared .mcp.json you can commit for your
team.

To check it worked, run claude mcp list, or type /mcp inside a session, and confirm bigeye
shows as connected.

Snowflake Cortex Code (CoCo)

You can add the gateway through the Desktop UI, the CLI, or by editing the config file.

Desktop UI

1. Open Agent Settings and go to the MCP tab.
2. Click + New.
3. Choose a Configuration Scope: Global (all workspaces) or Workspace (current project only).
4. In the Form view, fill in:
   • Server Name: bigeye
   • Server Type: Remote (HTTP)
   • Server URL: https://mcpgateway.bigeye.com/mcp
5. Under Headers, add:
   • Authorization → apikey <YOUR_API_KEY>
   • x-bigeye-workspace-id → <YOUR_WORKSPACE_ID>
   • On a non-default stack only: x-bigeye-url → https://<your-stack>.bigeye.com
6. Click Save. The server starts and its tools become available right away.

CLI

cortex mcp add bigeye https://mcpgateway.bigeye.com/mcp --type http \
  -H "Authorization: apikey <YOUR_API_KEY>" \
  -H "x-bigeye-workspace-id: <YOUR_WORKSPACE_ID>"

Config file

Edit ~/.snowflake/cortex/mcp.json (or <workspace>/.snowflake/cortex/mcp.json for workspace scope):

{
  "mcpServers": {
    "bigeye": {
      "type": "http",
      "url": "https://mcpgateway.bigeye.com/mcp",
      "headers": {
        "Authorization": "apikey <YOUR_API_KEY>",
        "x-bigeye-workspace-id": "<YOUR_WORKSPACE_ID>"
      }
    }
  }
}

Add an x-bigeye-url header only if you're on a non-default stack.

GitHub Copilot CLI

Edit ~/.copilot/mcp-config.json:

{
  "mcpServers": {
    "bigeye": {
      "type": "http",
      "url": "https://mcpgateway.bigeye.com/mcp",
      "headers": {
        "Authorization": "apikey <YOUR_API_KEY>",
        "x-bigeye-workspace-id": "<YOUR_WORKSPACE_ID>"
      },
      "tools": ["*"]
    }
  }
}

Add an x-bigeye-url header only if you're on a non-default stack.

You can also run /mcp add inside a Copilot CLI session, choose the HTTP server type, paste
https://mcpgateway.bigeye.com/mcp as the URL, and enter the headers as JSON when prompted.

Verify the connection

Ask your assistant something that uses the Bigeye tools, for example:

Are you connected to Bigeye? List my data sources.

If it returns your warehouses and sources, you're connected. You can also use your client's
MCP-status command (/mcp in Claude Code or Copilot CLI) to confirm the bigeye server is healthy.

If this fails with a workspace-related error, you can try asking "Get my current Bigeye user and workspaces". This should output your user info and the workspaces you have access to.

Run it yourself (advanced)

The Bigeye MCP Server is open source. If you'd rather run it locally with Docker or Python, to
self-host, customize behavior, or run fully air-gapped, clone the repository and follow the README:

https://github.com/bigeyedata/bigeye-mcp-server