Creating context with dbt MCP Server

Learn how to use the dbt MCP Server to connect to your external AI tool of choice to governed dbt context and actions. In this lab, you’ll configure MCP access, verify connectivity, and then complete a guided workflow where the AI tool pulls project metadata/context via MCP to complete a task (for example: understanding project lineage, querying the semantic layer, triaging production issues, or getting up-to-date docs) more accurately than prompting alone.

After this lab, you will be able to:

- Explain what MCP is and when it’s useful compared to “chat-only” AI workflows

- Configure a working MCP connection to dbt and verify the toolchain is functioning

- Use MCP-enabled context to complete a guided workflow (impact analysis / plan / draft output)

- Apply basic troubleshooting steps for common MCP setup failures (auth, connectivity, permissions)

Prerequisites:

- Basic familiarity with dbt concepts (projects, models, environments)

- Comfort following setup instructions (no coding required beyond copy/paste)

- Access to an AI client (such as Claude Desktop, Claude Code, Codex, ChatGPT, Gemini, dbt Wizard)

What to bring: You must bring your own laptop and AI client to complete the exercises. We will provide all other sandbox environments and any required credentials.

Duration: 90 minutes