dbt MCP Server
dbt Labs' official MCP server. The AI can read model definitions in your project, run dbt commands, trace upstream and downstream lineage, and query defined business metrics through the semantic layer.
Lineage is the standout — "which downstream reports break if I change this staging model?" used to mean clicking around dbt docs. Assessing blast radius before a change gets much faster.
Install / Connect
# No npm package — install from source.
# See the repo's README for the exact command:
# https://github.com/dbt-labs/dbt-mcpFollow the official README; env and args vary per server.
Why use it
Good for: analytics engineers working in dbt, particularly once the project has grown past a manageable number of models.
Before connecting: it can execute dbt commands like run and test. Point it at a development target — don't let the AI run against production schemas.
Similar / alternative servers
airtable-mcp-server
🗂️🤖 Airtable Model Context Protocol Server, for allowing AI systems to interac
ai.dinglebear/soma
Soma
linkedin-mcp-server
Open-source MCP server for LinkedIn. Give Claude and any MCP-compatible AI agent
jupyter-mcp-server
🪐 🔧 Model Context Protocol (MCP) Server for Jupyter.
Frequently Asked Questions
Does it work with both dbt Core and dbt Cloud?
Both, with different configuration. Semantic layer features require dbt Cloud.
Will the AI edit my model files?
Reading is the default; whether it writes depends on your configuration. Commit to git before letting it change models.