Without one, an AI agent has to click through the website like a person.
Checked 3 days ago. Based on ARC's automated checks, not a live agent session. How we check
Can AI agents find your products? No.
No. Your site doesn't offer AI agents a product catalog they can search.
Can they read prices and stock? Not tested yet.
Not tested. We couldn't open a product page on your site to check.
Can they get to checkout? No.
No. Without a product catalog to search, our agent had no way to reach checkout.
Want a deeper test? Run a free full scan with five test shoppers.
Add a product feed AI agents can read
Publish a product catalog agents can search, using the open Universal Commerce Protocol.
How it worksPut this free badge on your site. It shows your latest result, links back here and updates after each new check. It only says “Agent-ready” when our agent actually reached checkout.
Is this your store? Claim this page to add a note. Claims never change our results.
The raw results behind the answer above, for developers.
Agent Reachability Matrix
Existing ARC evidence shows discovery rails, checkout paths, and known block risks. A detected rail is not a promise that a named agent can use it.
Measured: ARC probe or browser cart test · Inferred: possible surface path · Unknown: no conclusion · Public report: external source.
ARC persona (alias id muse) · shared rails + ARC browser cart test
Not confirmed
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
ARC persona (alias id instinct) · shared rails + ARC browser cart test
Not confirmed
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
ARC persona (alias id grok) · shared rails + ARC browser cart test
Not confirmed
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
ARC persona (alias id openai-dots) · shared rails + ARC browser cart test
Not confirmed
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
ARC persona (alias id openclaw) · shared rails + ARC browser cart test
Not confirmed
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
Universal Commerce Protocol surface · eligibility unconfirmed
Not confirmed
Google surface eligibility not tested.
Not confirmed
Not confirmed
Watching for PAP v0.1; identity and permission rails not confirmed.
Rail / discovery evidence checked 2026-10-08 19:42 UTC. Index cart evidence uses the published scan below.
Measured means an ARC HTTP rail check or an Index browser cart test, not a live Muse, Instinct, ChatGPT, Gemini, Grok, or OpenClaw session. Checkout handoff is not a completed purchase. ARC is independent and not affiliated with Meta, Sierra, Shopify, OpenAI, or Google. PAP support is not confirmed.
Real HTTP agent-catalog check. Does not change the Index score.
No agent catalog found in this check.
Checked 2026-10-08 19:42 UTC
The fix
Expose a catalog agents can query over HTTP. On Shopify, enable search_catalog or search_shop_catalog at /api/ucp/mcp or /api/mcp and publish the /.well-known/ucp profile.
Real HTTP check of the store's agent catalog, UCP, and storefront WebMCP adapter, plus Shopify-only Checkout WebMCP HTML detection when applicable (get/update/complete, buyer approval required). Not a browser session and not any third-party agent product. No model call or purchase. The full score comes from ARC browser shoppers A–E (real browser cart tests); this check does not change the score.
Public checks need attention
Not measured items need a successful re-check before readiness can be established.
native_commerce(checkout_eligibility): Can't be checked from outside — set in your Merchant Center feed
These are the public parts of Google's checklist that ARC can check from outside. This is not Google approval or certification. See Google's Merchant Center help on UCP / agentic checkout.
Platform: Not detected (No supported homepage fingerprint found).
What robots.txt says about the homepage and tested product path. Bots may ignore robots.txt, and firewalls may block allowed bots. This is not proof of fetch access.
Answer fetchers: 8/8 allowed
Search indexers: 6/6 allowed
Training crawlers: 7/7 allowed
From the page's structured data · server HTML JSON-LD only.
Unavailable: Product page not reached (bot check, waiting room or not a product page). No score measured.
From the store's pages and structured data. Presence does not verify the policy's terms.
Returns: not found
Shipping: not found
Add your actual return policy to Organization JSON-LD. Replace placeholders with truthful terms.
<script type="application/ld+json">
{"@context":"https://schema.org","@type":"Organization","hasMerchantReturnPolicy":{"@type":"MerchantReturnPolicy","merchantReturnLink":"https://YOUR_STORE/policies/refund-policy","returnPolicyCategory":"https://schema.org/MerchantReturnFiniteReturnWindow","merchantReturnDays":"YOUR_RETURN_DAYS"}}
</script>Checked 2026-10-08 19:42 UTC
https://synology.com/PAP (announced Oct 2026 by Meta and Sierra with industry partners) is an open standard for how personal agents discover and work with businesses — website discovery, OAuth session, then site / APIs (MCP, OpenAPI) / company agent. ARC already answers the two-lane question: can agents discover you, and can they buy from you. Until the v0.1 spec ships, we probe sensible discovery candidates and label unknown clearly. ARC is not affiliated with Meta or Sierra.
Watching for PAP v0.1
No PAP-specific discovery markers detected yet. Spec v0.1 is not published; ARC is watching, not inventing support.
Not a claim of PAP support. Spec v0.1 is expected later this month; reference implementation coming. Related surfaces below (UCP, WebMCP, MCP well-known) are existing agent-discovery checks — not PAP.
Probed 2026-10-11 17:44 UTC · source announcement: sierra.ai/blog/introducing-personal-agent-protocol
Browser cart attempts. Score is this lane only.
Not on the curated Index yet. You can still run a free product scan.
Typed decisions over an HTTP product scrape. Not a live personal-agent session.
No Jev facts stored yet for this host.
Quoted mismatches only, from one AI model’s read-back (currently Grok via API). One model, not a consensus.
No real Grok fact check stored for this store yet.
From stored ARC rows only.
Claims never change ARC’s measured scores.
What Grok tells shoppers is a real model call via API, checked against the store's pages, not the consumer Grok app. Whether agents reach checkout is two lanes: a real HTTP check of the store's agent catalog, and five ARC browser cart tests. Those attempts are not a live Muse, Instinct, Grok Bot, OpenAI dots, or OpenClaw session. The score is the browser lane. The catalog check does not change it.