How we test
Our rankings answer one question: which MCP server should an AI agent use for a given SEO job? Here is how we get there.
1. Find
We track servers from vendor documentation, GitHub and MCP registries, and re-check for new, renamed and retired servers. Every MCP server that does SEO work is eligible: official or community, paid or free. We review the ones we wouldn't recommend too, and say why.
2. Verify
For each server we record the maker, hosting (remote or local), sign-in method, cost, documented tools and release date, each from a primary source: the vendor's docs, its official announcement or its code repository. Third-party write-ups are never enough on their own. Anything we can't verify stays blank and is marked "—".
3. Judge
We rank servers by SEO job rather than overall, because no server does everything. To be recommended, a server must be official or recently maintained, document its tools in public, and state its sign-in and cost plainly.
Hands-on testing
We are rolling out hands-on tests: connecting each server to the same agent and running the same SEO tasks. Results will appear on each review, with the date, once a server has been tested. Until then, reviews are based on verified documentation.
4. Re-check
Pricing, tool lists and endpoints change often. Every page shows the date its facts were last verified, and we update pages when vendors change.