Overview
mcp-server-gsc-pro is a production-ready MCP server that gives AI agents full programmatic access to Google Search Console. 36 tools total — raw API coverage plus a computed intelligence layer that runs analysis AI agents can’t easily derive from raw data alone.
Published on npm as mcp-server-gsc-pro, currently at v1.2.2.
What it does
The server is split into four layers:
Core GSC API (13 tools) — direct wrappers for everything GSC exposes: search analytics with full dimension/filter support, URL inspection, sitemap management, and site verification.
Computed intelligence (7 tools) — this is where it earns its name. Tools that do the heavy lifting before the model even sees the data: period-over-period comparison with statistical significance, content decay detection, keyword cannibalization scoring, CTR benchmarking against positional norms, and traffic drop alerts with likely-cause attribution.
Multi-API dashboards (5 tools) — unified views that pull across GSC, PageSpeed, and CrUX in a single call. Page health dashboard, indexing health overview, and cannibalization resolver that surfaces the winning URL.
Adjacent Google APIs (10 tools) — PageSpeed Insights, CrUX field data, Google Indexing API, and mobile-friendly test. Useful on their own; more useful alongside GSC data.
Reliability features baked in throughout: automatic retry with exponential backoff, concurrency limits to avoid quota burns, and structured error responses the model can act on rather than just display.
Status
Production-ready. 166 Vitest tests, CI runs on every PR across Node 18, 20, and 22. No known regressions. The npm package is published and actively maintained.
Next focus: expanding the intelligence layer — more anomaly detection patterns and cross-property analysis for sites with multiple GSC properties.