If you already work in an editor or coding agent that supports the Model Context Protocol, you do not need to switch tools to use Emolysis. This tutorial adds the Emolysis MCP server, emo mcp, to that tool.
What you need
- An editor or coding agent that supports MCP servers over standard input and output.
- The
emocommand line on your path; it comes with the desktop app. - An Emolysis account; the Free plan is one. Every use of Emolysis, local files included, needs one.
Step 1: Check the engine
Run emo doctor in a terminal. It checks the engine, DuckDB and the credential store, and reports anything missing.
Step 2: Sign in
Run emo login. It opens your browser, where you sign in with your GitHub, Microsoft or Google account; the MCP server uses that sign-in. If you skip this, or your sign-in expires, a tool returns a short code and a link instead of a result: open the link, sign in and confirm the code, then ask the agent to retry.
Step 3: Add the server
Add Emolysis to your tool's MCP configuration. Most tools take an entry like this one, but check yours for the exact file: { "mcpServers": { "emolysis": { "command": "emo", "args": ["mcp"] } } }.
Restart the tool. It starts emo mcp as a local process; nothing listens on a network port. Ready-made configs and skills for common tools are in the public emolysis/integrations repository on GitHub.
Step 4: See the tools
Your tool should now list the Emolysis tools. Each one is an engine command, named with underscores: query_run, template_list, template_render, template_test, dialect_transpile and so on. Their input schemas describe the parameters.
Step 5: Try it on local files
Signed in, the server works on local files on DuckDB alone, with no platform connected. Put a CSV or Parquet file in your project and ask your coding agent something like: "Using Emolysis, show the schema of data/orders.parquet and total amount by month." It calls the query tool and returns the result. Then try: "Using Emolysis, list the templates for the Aggregation / gold shape."
Step 6: Connect a platform
Ask: "Using Emolysis, connect Databricks." The connect flow opens Databricks' own sign-in in your browser; you then pick a workspace and compute. Setup buttons on the web dashboard that write the MCP configuration for common tools are planned.
Step 7: Use the skills
The integrations include skills for common requests, so you can type them directly:
/dashboard "your question"to build a dashboard from a question./pipeline shape from source to targetto fill and test a template./explain pipeline, and/migrate table to platform(planned).
A note on data
Your coding agent sends tool results to its own model provider, under that provider's terms. Tool results can include schemas, query results and sample values. Point it at data you are comfortable sharing that way. See what the model provider sees.
If the tools do not appear
Check that emo is on the path your tool uses and that the entry runs emo mcp, then restart the tool. Run emo doctor to check the engine itself; each failed check says what to fix. The MCP server docs and troubleshooting list common problems.