lightdash

The lightdash-mcp-server exposes a set of MCP-compatible tools that allow an AI assistant to interact with Lightdash data. It provides methods such as listing projects, spaces, charts, dashboards, and catalogs, as well as retrieving custom metrics and charts/dashboards as code. These tools enable streamlined integration with your Lightdash workspace, letting AI agents query metadata and fetch code representations for dashboards and charts in a standardized way. To use it, run the server (via npm or npx) and connect an MCP client to the server transport (stdio by default, or HTTP for remote access). The server supports both a local stdio transport and an HTTP transport, so you can choose the integration style that best fits your environment.

Key capabilities include:

• list_projects, get_project
• list_spaces, list_charts, list_dashboards
• get_custom_metrics, get_catalog, get_metrics_catalog
• get_charts_as_code, get_dashboards_as_code These tools give you programmatic access to Lightdash resources, enabling chat-based or automation workflows to discover and fetch Lightdash data and configurations.

Prerequisites:

• Node.js and npm installed on your machine
• Access to Lightdash with a valid API key (PAT)

Installation steps:

1. Install the MCP server package locally or globally:

# Local project install
npm install lightdash-mcp-server

# Or run directly via npx in a one-off command
npx -y lightdash-mcp-server

2. If you want to run via Smithery for automated client setup, use:

npx -y @smithery/cli install lightdash-mcp-server --client claude

3. Prepare configuration for MCP usage (see mcp_config example below) and set required environment variables (LIGHTDASH_API_KEY and LIGHTDASH_API_URL) when running the server.
4. Start the server in your preferred mode (stdio or HTTP) and connect your MCP client to it using the appropriate transport (default stdio or HTTP if you expose port 8080).

Notes:

• Ensure the Lightdash API key and URL are kept secret and are not committed to source control.
• If using the HTTP transport, you may need to expose port 8080 or configure a reverse proxy as needed.

Tips and common considerations:

• Environment variables: LIGHTDASH_API_KEY and LIGHTDASH_API_URL must be set for HTTP mode and when using Lightdash authentication.
• Transport modes: Default is Stdio. For HTTP, start the server with a port (e.g., -port 8080) and connect clients via http://localhost:8080/mcp.
• When using HTTP mode, clients should use the URL-based MCP configuration (url field) instead of command/args.
• The npm package name is lightdash-mcp-server. Use npx lightdash-mcp-server for quick runs or install locally for persistent use.
• Inspect and update the MCP configuration to map each Lightdash operation to the corresponding MCP tool name (e.g., lightdash.list_projects).
• Refer to Lightdash API rate limits and permissions to avoid hitting quotas during automated queries.