Every Search Console MCP server does roughly the same thing: it wraps the Search Analytics API in typed tools so Claude or Cursor can query your search data mid-conversation. There are at least eight open-source ones on GitHub. They're free, and for ad-hoc questions about recent months they work fine.
They also share a ceiling, because the live API only reaches back 16 months. Ask one to compare this year against two years ago and the honest answer is that the data no longer exists. A protocol can't fix a retention policy.
Get early access Free during early access. No card required.
Same protocol, different substrate
Month17's MCP server speaks the same protocol but queries a permanent archive instead of the live API. On connect it backfills the full 16 months Google still holds. From then on it syncs daily and never deletes, so the longer it runs, the further back your AI can see.
It's hosted, so there is nothing to run locally. No Node process, no service-account JSON, no laptop that has to stay awake for the cron job.
The tools
22 of them, from raw search_analytics queries with full dimensions and filters to prepared analyses. Four only make sense with an archive underneath:
year_over_year: any metric, any years, side by side.seasonal_patterns: recurring trends across multiple cycles.content_decay: pages with sustained click decline over months.core_update_impact: your traffic overlaid with confirmed Google update dates.
Setup
For Claude Code, one command:
claude mcp add --transport http month17 https://month17.com/mcp/ \
--header "Authorization: Bearer m17_your_api_key"Claude Desktop and Cursor take the JSON equivalent. The setup walkthrough covers both, along with prompts that pull their weight.
When a free server is the right call
If your questions never leave the last few months, run one of the open-source servers and pay nothing. They are genuinely good enough for that job, and pretending otherwise would be silly.
The catch is time. The moment a question spans more than 16 months, no live-API server can answer it, and switching later doesn't recover the months that expired while you waited. That asymmetry is the whole argument for putting the archive in place before you strictly need it. It's free during early access, and the backfill grabs everything Google still has on the day you connect.