MCP servers ranked by tool surface quality

Every server below is audited with the same 19 static rules: what its tools/list costs in tokens, how tightly its schemas are typed, and whether its surface is shaped around user intents or around REST endpoints. Nothing here calls a tool or runs an LLM. All 18 carry a published scan as of 2026.

Audited MCP servers, highest composite score first.
#ServerCategoryScoreToolsTokensFindingsScanned
1DeepWikiDeveloper Tools10033560
2ExaSearch and Browsing10025380
3FetchSearch and Browsing10012380
4Microsoft LearnProductivity and Docs10039731
5Cloudflare DocsCloud and Infrastructure9924213
6Context7Developer Tools9921,0532
7PostgreSQLData and Databases991322
8SlackCommunication9986811
9Hugging FaceSearch and Browsing9844,9175
10Brave SearchSearch and Browsing96825,4988
11FilesystemDeveloper Tools96142,79724
12MemoryProductivity and Docs9692,36010
13PlaywrightDeveloper Tools96264,3494
14SupabaseData and Databases96193,32827
15Chrome DevToolsDeveloper Tools95295,63032
16MongoDBData and Databases94205,768142
17NotionProductivity and Docs862417,47695
18GrafanaObservability776516,79486

By category

Scores describe a published tool surface on a given date, not the quality or security of the product behind it. How this is measured.