Is on.com ready for AI agents?

AI agent readiness report · platform: Custom · checked 2026-10-24 · log in to re-run
50%
Readiness
4/8
Verified (readable)
Custom
Platform
50%4/8
Partly ready
Mcp
Webmcp
A2a
Anp
Ucp
Ap2
Openapi
Llms Txt
Wikidata
15/18 AI crawlers reached the homepage3 blocked
Ranks above 99% of 171 sites checkedcorpus avg 24/100

Partly ready

on.com has 4 of 8 readable agent standards in place; missing Webmcp, A2a, Anp, Ap2. 1 could not be read because the site refused our request (Openapi), so they are not counted.

Assessment

4 of 8 readable agent-readiness checks verified (1 could not be read: the site refused our request). Our request with an ordinary browser user-agent was refused too, so the crawler results describe our checker being blocked, not rules about AI crawlers. See per-protocol drawers for evidence and fix-it files.

Per-protocol verdicts

Mcpinitialize handshake succeeded at https://www.on.com/mcp,tools=1Verified
Webmcpno navigator.modelContext.register, <script type="webmcp">, or <link rel="modelcontext"> in HTMLNot found
A2ano /.well-known/agent.json or /.well-known/agent-card.jsonNot found
Anpno /.well-known/agent-descriptions and no did:wba identity at /.well-known/did.jsonNot found
Ucp/.well-known/ucp present, UCP version 2026-08-25Verified
Ap2/.well-known/ucp does not declare dev.ucp.shopping.ap2_mandateNot found
Openapicould not read /api/openapi.json (403) - the site refused our request, so this is not a findingCould not read
Llms Txt/llms.txt present, 1 sectionsVerified
WikidataWikidata entity Q43452713 (On AG)Verified

.well-known files

.well-known/mcp.json248 bytes200
Mcp Well Known Alt3049 bytes200
/mcp endpointnot found406
Ai Plugin Legacy3049 bytes200
.well-known/agent.json3049 bytes200
.well-known/agent-card.json3049 bytes200
Anp Discovery3049 bytes200
Anp Did3049 bytes200
Openapi Well Known3049 bytes200
Openapi Root3049 bytes200
Openapi Apicould not read403
Ucp Profile5301 bytes200
Llms Txt800 bytes200
Llms Full Txt3049 bytes200
robots.txt392 bytes200
sitemap.xml460 bytes200

AI crawler audit

15 of 18 AI user-agents reached the homepage.
Our request with an ordinary browser user-agent was refused as well (HTTP 200 before, 0 after). This site was blocking our checker, so read these counts as "could not check", not as rules about AI crawlers.
9 of 18 were tested with the operator's published user-agent string and the rest with the bare token, since no published string was found. A request that only names a user-agent does not prove the real crawler would be treated the same.

What this means for on.com

On on.com (Custom), 4 of 8 readable agent standards are in place: Mcp, Ucp, Llms Txt, Wikidata. Missing: Webmcp, A2a, Anp, Ap2.

Without JavaScript, your brand name does not appear in the raw HTML, so crawlers that do not run JS see an empty shell.

on.com scores higher than 99% of the 171 sites checked here so far; the average readiness score is 24/100.

Make on.com agent-ready

Start with the highest-impact gap: Webmcp. Publish a working MCP endpoint at /.well-known/mcp.json (the initialize handshake must succeed, not just return 200), add an A2A agent card, and keep the crawler path clean.

Frequently asked questions

Are these real probes or AI guesses?

Real probes. We fetch the actual files, run a real MCP initialize handshake, and validate A2A cards against their schema. No model decides a verdict.

Does on.com need every standard to pass?

No. Start with the highest-impact gap above, then add the rest. Most sites win the most by shipping a working MCP endpoint and keeping the crawler path clean.

Check your own site free

Run the same protocol probes on any URL, then sign up free for the copy-paste fix-it files and the full 20-crawler audit.

Run a free check

Agent readiness on Custom

We have checked 74 other sites on Custom. The Custom hub aggregates protocol verification rates and average readiness scores across every Custom site checked here.

Is Custom ready for AI agents?