dbt MCP Server
Runs dbt commands and answers questions about the project's models and metrics
Add it to your client
Officialuvx dbt-mcp
{
"mcpServers": {
"dbt-mcp": {
"command": "uvx",
"args": [
"dbt-mcp"
]
}
}
}- Transport
- both
- Runtime
- Python
- Auth
- Required
- License
- Apache-2.0
- Stars
- 597
dbt Labs' own server splits its tools into groups that map to how people actually work: the CLI group runs build, run, test and compile against a local project, the Discovery group answers structural questions — lineage, model health, node details, recent performance — from dbt Platform metadata, and the Semantic Layer group lists metrics and dimensions and executes metric queries. Two SQL tools sit alongside them, one executing statements on Platform infrastructure and one generating SQL from a plain-language question using project context. It covers dbt Core, Fusion and Platform, so the same client entry works whether the project runs locally or in the cloud. Installation is a uvx command; dbt Platform features authenticate with a token or OAuth.
Similar MCP servers
GreptimeDB MCP Server
MCP serverReads tables and runs SQL against GreptimeDB, the open-source observability database.
SurrealDB MCP Server
MCP serverCommunity bridge to SurrealDB: run SurrealQL, manage records and relations from an agent.
Cloud Run MCP Server
MCP serverGoogle's own server for deploying a folder or a set of files to Cloud Run and reading back the logs.

Comments(0)
Sign in to comment