What an agent is told about openai.com
This is the public, signed answer Resolve gives any agent that asks about openai.com before it reads, cites, connects or transacts, from a reading 19 hours ago. Every line is a field of that answer; nothing is re-scored here. The decision on reading rests on this: no AI-crawler identity was served the page (gptbot 403, claudebot 403, oaisearch 403, perplexity 403).
What it declares, what it was served, what is absent
Declared is what the site says about itself. Observed is what CrawlCheck was served. Not measured means this record does not carry that reading, and nothing is guessed in its place.
Crawl policy
declared12 of 12 named AI agents may read / under the site’s own robots.txt
Rights
observedtrain not declared · ground not declared · summarize not declared · cite not declared · transact not declared
Delivery
observed5 identities sent from CrawlCheck’s own network; 5 refused: Mobile browser, GPTBot, ClaudeBot, OAI-SearchBot, PerplexityBot
Entity
not measuredthe census carries no entity reading
Control
unprovennobody has proven control of openai.com to CrawlCheck, so who stands behind the site is unconfirmed
Assurance ladder
level: declared✓ declared · – discovered (not measured) · ✗ reachable · ✗ conformant · ✗ owner verified · ✗ authorized · – reliable (not measured) · ✗ action ready
MCP Registry
observedthe official MCP Registry lists no remote MCP endpoint on this domain (import finished 2026-10-14)
Findings
not measurednone open
Absences
observedmcp_registry_listing — looked for and not served; each absence states the request that proved it
Freshness
observedobserved 19 hours ago, fresh until 2026-10-25
The same answer, as calls
Everything on this page is a field of one signed body. An agent reads the body; a person reads this.
GET /api/v1/resolve?domain=openai.comJSON, Ed25519-signed, public, no keyBefore actingPOST /api/v1/preflight {domain, action}allow, warn, require_confirmation or block for one actionIts evidencethe registry rowevery row re-derives from a fetchVerify itnode crawlcheck-verify.mjs --online answer.jsonhash, signature and signing policy, against the keys CrawlCheck publishesWhat comes next
Each step builds on the same signed record. Start wherever you are.
- 01FixFix what holds the grade downSteps for your platform, checked by a re-scan.
- 02WatchWatch it every dayFree for 30 days. Changes and new findings reach your inbox.
- 03ClaimProve the domain is yoursOne DNS record or meta tag. Every signed answer about the site then says control is proven.
- 04Agent viewSee what agents are toldThe signed answer any agent gets before it reads, cites, connects or transacts.You are here
- 05BuildAsk from your own codeOne call before your agent acts: npm, PyPI or plain HTTPS.