laylzj777-analytics-mcp

An independently maintained fork of Google's experimental Google Analytics MCP server, based on upstream version 0.7.0, commit a8ca729d4a8fa99bffe87962c17c0539c6aa9da7. This fork is not an official Google release. The upstream Apache-2.0 license and copyright notices are retained.

Install and run

Requires Python 3.10 or newer. Install into a dedicated Python environment:

python -m pip install laylzj777-analytics-mcp==0.7.0
analytics-mcp

google-analytics-mcp is an equivalent entry point. Both start the Python server directly; Node.js, npm, and --managed-python are not needed. The distribution installs its Python dependencies itself. Do not install the upstream analytics-mcp distribution in the same environment: both provide the analytics_mcp import package and the same commands.

For a managed plugin host, declare a single pip runtime dependency on laylzj777-analytics-mcp==0.7.0, resolve its analytics-mcp entry point, and preserve the Python runtime environment supplied by the host.

Credentials and validation

Enable the Google Analytics Admin and Data APIs and grant the account access to the required Analytics properties. Configure Google Application Default Credentials using GOOGLE_APPLICATION_CREDENTIALS (a local credential JSON file path) and provide GOOGLE_PROJECT_ID as required by your host. Never put credential files or tokens in this package or its configuration examples.

Set MCP_VALIDATE_CREDENTIALS=true to validate credentials before tools/list returns tools. Validation is disabled by default. Validation refreshes ADC and performs a read-only Admin API request; an empty account list is valid. A successful validation is cached for the server process.

Failures use sanitized JSON-RPC errors with data.kind = "credential_validation":

Error messages do not include credential paths, tokens, or account data. This is a fork-specific error contract, not a standard MCP authentication extension. The fork retains the upstream read-only Analytics reporting and account tools.