NewAgent-ready skill for Claude Code, Cursor and Codex →
AgentsReady
Agent-ready analytics tools · #3 of 11

Fathom

Run the agent liveReport a change
81/100Agent readiness
#3 of 11Agent-ready analytics tools
NoBot wall
NoRemote MCP

Biggest gap: Remote MCP server.

Core 58/60, plus the average of sign-up (30/40) and API (16/40). Scanned 3 Nov 2026.

Latest live check, 12 Nov 2026

74Agent readiness / 100
Mostly readyScored for: Get an API key
Core58/60
Get an API key16/40

What an agent gets measured, not scored

7 of 7AI agents that get the real page
≈32kTokens to read the homepage
0.1 sHomepage load for an agent
Same pageAgents see what browsers see

About 10% of the raw HTML is text (≈3.3k tokens). The rest is markup and scripts.

Agent by agent (7 of 7 get in)
  • INClaude-User sent by a person200
  • INChatGPT-User sent by a person200
  • INPerplexity-User sent by a person200
  • INOAI-SearchBot answer engine200
  • INPerplexityBot answer engine200
  • INGPTBot training crawler200
  • INClaudeBot training crawler200

We send each agent's name from our own servers. A site that checks IP addresses may treat the real agent differently.

Fix these first

  1. +8 Remote MCP server. Host a remote MCP server (mcp.<domain>/mcp, Streamable HTTP) for your core actions.
  2. +5 Docs as markdown. Serve every docs page as .md, or honour Accept: text/markdown.
  3. +5 OAuth client self-registration. Support dynamic client registration (RFC 7591) so MCP clients connect in one click.
  4. +3 OAuth discovery metadata. Publish RFC 8414 and RFC 9728 metadata.
  5. +3 SDKs and a CLI. Ship official SDKs and a CLI with device login.
All 22 checks
  • PASSAgents get the real page, not a bot wall12
  • PASSrobots.txt lets user-triggered agents in6
  • PASSAnswer-engine crawlers are allowed4
  • PASSContent is in the raw HTML6
  • PASSSemantic HTML (main, nav, header)4
  • PASSIcon buttons and links have names3
  • PASSForm fields have labels3
  • PASSImages have alt text3
  • PASSThe page has an H12
  • PASSStructured data (JSON-LD)5
  • PASSMeta description and Open Graph tags3
  • PASSsitemap.xml3
  • PASSllms.txt4
  • FAILDates show the content is fresh2
  • PASSOpenAPI spec7
  • FAILRemote MCP server8
  • FAILDocs as markdown5
  • PASSSelf-serve API keys6
  • FAILOAuth client self-registration5
  • FAILOAuth discovery metadata3
  • FAILSDKs and a CLI3
  • PASSFree tier or test keys3
Newer standards (1/9, not scored yet)
  • NOMarkdown for agents Return markdown when an agent sends Accept: text/markdown. Fewer tokens, cleaner answers. How–
  • NOContent Signals in robots.txt Say what AI may do with your content (search, ai-input, ai-train), separately from crawling. How–
  • NOAgent Skills index Publish /.well-known/agent-skills/index.json so agents can find skills for using your product. How–
  • NOAPI catalog (RFC 9727) Publish /.well-known/api-catalog listing your API specs, so agents find them without guessing. How–
  • NOLink headers to machine-readable versions A Link header on the homepage pointing at your API description, docs or markdown version. How–
  • NOWebMCP tools on the page Register your page's actions as tools (navigator.modelContext) so browser agents call them instead of guessing at buttons. How–
  • NOauth.md for agent sign-up Publish /auth.md saying how an agent registers and gets credentials for a user. How–
  • NOAI Catalog Publish /.well-known/ai-catalog.json listing your MCP servers and agents, so clients find them from the domain alone. How–
  • YESMissing pages return 404 A made-up URL should return 404, not 200. Agents treat a 200 as a real page and quote it. How–

Full checklist for usefathom.comReport page and badgeShare: X / LinkedIn /

How to fix these gaps

Add the badge to your README or site

AgentsReady badge for usefathom.com
[![AgentsReady](https://agentsready.dev/badge/usefathom.com)](https://agentsready.dev/c/usefathom.com)