lightdash-mcp-server is an MCP (Model Context Protocol) server that exposes Lightdash's API via a standardized MCP interface, enabling AI assistants to interact with Lightdash data. It provides MCP-compatible access to core Lightdash resources such as projects, spaces, charts, dashboards, catalogs, and metrics, and supports retrieving charts and dashboards as code. The server exposes a set of tools: list_projects, get_project, list_spaces, list_charts, list_dashboards, get_custom_metrics, get_catalog, get_metrics_catalog, get_charts_as_code, and get_dashboards_as_code. It supports two transport modes: Stdio (default) and HTTP. In Stdio mode you run npx lightdash-mcp-server and configure the lightdash section in your MCP config to specify command, args, and environment variables (LIGHTDASH_API_KEY and LIGHTDASH_API_URL). In HTTP mode you run npx lightdash-mcp-server -port 8080 to expose a HTTP endpoint at http://localhost:8080/mcp; clients should use the url field. Programmatic usage uses the StreamableHTTPClientTransport with a standard MCP client. Environment variables must be set on the server side, as they cannot be passed through the MCP client. See examples/list_spaces_http.ts for a complete HTTP connection example.