Web Agent-Readiness Leaderboard
Agent-readiness scores for websites and their MCP servers, scored against the same eight principles as the CLI leaderboard. See the methodology for how the web audit probes MCP shape, discovery surfaces, and machine-readable content.
| # | Site | Description | Global | Relative |
|---|---|---|---|---|
| 1 | anc.dev | The agent-native CLI standard and the agent-readiness auditor itself. | 57 | 73 |
How web scoring works
Each website is probed for its MCP server shape, MCP and agent discovery surfaces, machine-readable content (llms.txt, OpenAPI, JSON Schemas), root-HTML affordances, and crawl policy. Checks that do not apply to a site (no MCP server, no API surface, a different declared site type) are excluded rather than counted against it. Global measures absolute agent capability against a maximally agent-ready site, so exposing and nailing more surfaces ranks higher; Relative measures how agent-ready a site is for the checks that apply to it, so a site perfect for its type approaches 100%. The board sorts by Global; each result page headlines Relative.
The board is curated. To score any public site on demand, use the web audit or the audit_website MCP tool.