Our Google Search Console MCP is a local, read-only 📝MCP server that exposes 📝Google Search Console search-performance data — queries, pages, clicks, impressions, 📝Click-Through Rate (CTR), and position — to an AI agent as structured tools.
It runs locally as a stdio server and gives Claude read-only access to any Google Search Console property the authenticated Google account owns. Inside 📝MythOS publishing loops it acts as the measurement layer, so the agent can pull live search performance directly rather than guess at what is working.
It was built to make three questions answerable at a glance: which content is ranking, which content needs improvement, and which content needs to be bolstered by supporting, internally linked content. By surfacing Search Console data inside the writing loop, decisions about what to publish next and which posts to refresh become data-driven instead of intuitive.
Its internals — the tool surface, authentication model, and engineering decisions — are documented separately in 📝Google Search Console MCP — Technical Architecture.
We built it so we can see at a glance which keywords sit in striking distance — ranking just off the front page — and then push them over with minor on-page optimization or a little link building.