Skip to content

README.md

Latest commit

 

History

History
46 lines (28 loc) · 2.32 KB

README.md

File metadata and controls

46 lines (28 loc) · 2.32 KB
Raw

SourceMedium for Gemini CLI

Query and analyze the commerce data your organization maintains in SourceMedium, from Gemini CLI. The extension installs SourceMedium's remote MCP server and gives the agent a data-analysis workflow that preserves SourceMedium's metric definitions and data boundaries.

What it provides

  • SourceMedium MCP tools for warehouse, Shopify, Meta Ads, health, catalog, and documentation data.
  • OAuth connection through SourceMedium's existing authorization flow, discovered automatically from the server.
  • An agent skill that starts with data context, prefers canonical metrics, bounds custom SQL, and reports assumptions.

The server is read-only. It cannot change provider settings, credentials, connections, or warehouse data. No Google Cloud project, gcloud login, or BigQuery credential is required: your SourceMedium organization membership determines what the server can reach.

Install

gemini extensions install https://github.com/source-medium/sourcemedium-gemini

Restart Gemini CLI, then authenticate:

/mcp auth sourcemedium

Complete the SourceMedium sign-in and consent flow in the browser. Verify with /extensions list and /mcp.

An active SourceMedium account is required. If sign-in succeeds but every tool denies access, your user is not yet entitled to an organization; contact your SourceMedium administrator.

Local development

gemini extensions validate .
gemini extensions link "$PWD"

Restart Gemini CLI and confirm the extension appears in /extensions list, the SourceMedium server appears in /mcp, and OAuth completes. Unlink the local copy and install from GitHub in a clean environment before tagging a release, so the path a first-time user takes is the one that was tested.

Data-use workflow

For an analysis session, the agent should first inspect the available data with get_data_context, use search_data_catalog before querying, and prefer query_metrics for SourceMedium-defined metrics. It should only use custom SQL after inspecting the relevant tables and columns, and should state the source, date range, freshness, and assumptions in its answer.

Support

Documentation: Connect an AI assistant

For help, contact SourceMedium support.