dbt MCP Server
dbt MCP Server is a Model Context Protocol server that gives AI agents tools to run and explain SQL, query the dbt Semantic Layer, and browse a dbt project's models, sources, macros, exposures and lineage across dbt Core, dbt Fusion and dbt Platform.
dbt MCP Server exposes dbt project context and operations as MCP tools so an AI agent can work with a dbt Core, dbt Fusion or dbt Platform project. SQL tools execute SQL on dbt Platform infrastructure with Semantic Layer support and generate SQL from a natural-language request using the project's own context.
Semantic Layer tools retrieve dimension values, list metrics and saved queries, get compiled SQL for a metric without running it, and query metrics with filtering and grouping. Discovery tools retrieve macros, models, sources with freshness status, exposures, and a bounded lineage graph with configurable type, depth and direction.
The project also publishes an experimental Model Context Protocol Bundle with each release, so bundle-aware clients can import the server without additional setup. It is written in Python.