Is claude.com agent ready?
claude.com is partly agent ready: it scores 59/100 (grade C) on the w2agent rubric. Weakest areas: Agent Protocols (25), Agent Discovery (34), Agent Auth (40).
Changed since last check
+4 points since the previous check (2026-09-16)
Fixed: Appropriate Schema.org type, Schema.org JSON-LD present, nextjs-metadata-api
Regressed: Response time, llms.txt links resolve
Changed: Schema.org completeness (partial → fail)
Category scores
Technical
91
Agent Auth
40
Agent Discovery
34
Agent Protocols
25
Content Quality
74
Structured Data
77
Bot Accessibility
79
Agent protocol panel
- A2A agent cardNot found
- OAuth discoveryNot found
- MCP discoveryNot found
- llms.txt validityPresent
Check results
Bot Accessibility
- ✓No error status codespass
- ✗Response timefail
- ✓XML sitemap existspass
- ✓AI agent accesspass
- ✓Edge treatment of agent user-agentstracked, not scoredpass
- ✓robots.txt existspass
- ✗Content negotiation (Markdown)fail
- —Declared AI usage preferencestracked, not scoredskip
Structured Data
- ✓OpenGraph meta tagspass
- ✓Semantic HTML elementspass
- ✓Appropriate Schema.org typepass
- ~JSON-LD validpartial
- ~Heading hierarchypartial
- ✓Schema.org JSON-LD presentpass
- ✗Schema.org completenessfail
Content Quality
- ✓FAQ contentpass
- ✓Answer-first contentpass
- ✓Content lengthpass
- ✓Token efficiency (HTML-to-text ratio)pass
- ✗Freshness signalsfail
- ✗Author attributionfail
- ✓Citation readinesspass
- ✓Lists and tables for structured infopass
Agent Discovery
- ✓llms.txt follows specpass
- ✓llms.txt existspass
- —MCP server cardtracked, not scoredskip
- ✗Markdown alternate linksfail
- ✗/.well-known/agent-card.jsonfail
- ~llms.txt links resolvepartial
Agent Protocols
- ✗Discoverable API endpointsfail
- ✗MCP endpointfail
- ✗Action schemasfail
- —/.well-known/ucpskip
- ✗Structured formsfail
- —WebMCP imperative APItracked, not scoredskip
- —WebMCP declarative APItracked, not scoredskip
- ✓Reachable as Google-Agentpass
- ✓Reachable as ChatGPT-Userpass
- ✓Reachable as Perplexity-Userpass
Agent Auth
- —MCP authorization (401 challenge → protected resource metadata)skip
- ✗PKCE S256 supportfail
- ✓Public API surfacepass
- —Client ID Metadata Documents (bonus)skip
- ✗OpenAPI specificationfail
- ✓Agent/bot auth documentationpass
- ✗OAuth discovery chain (RFC 9728 → RFC 8414)fail
- ✓Developer portalpass
- ✗Scoped API permissionsfail
- ✗API catalog (RFC 9727)fail
- —Agent operator signing directorytracked, not scoredskip
Technical
- ✓HTTPSpass
- ✓UTF-8 charsetpass
- ✓Canonical URLspass
- ✓Mobile viewportpass
- ✓Content accessible without JSpass
- ✓Security headerspass
Other
- ✓nextjs-metadata-apipass
- ~nextjs-static-generationpartial
- ✗nextjs-well-known-servedfail
- —local-business-hours-jsonldskip
- —local-business-nap-structuredskip
- —local-business-booking-accessibleskip
Score history
- 59/100 · rubric v3.0
- 55/100 · rubric v3.0
- 54/100 · rubric v3.0
- 58/100 · rubric v2.0
- 58/100 · rubric v2.0
- 57/100 · rubric v2.0
- 55/100 · rubric v2.0
- 50/100 · rubric v2.0
- 50/100 · rubric v2.0
- 50/100 · rubric v2.0
- 50/100 · rubric v2.0