Agent readiness — losen.meta-machine.org

“

No blocking issues found, but coverage was partial.

This is the message an agent forms about the site — and, in effect, relays to the shopper who asked.

100/100

Readiness score

Blocking issues

Failing checks

Passing checks

By category

Access100

Discoverability100

Search & filternot assessed

Deep-link entry100

Product datanot assessed

Servicesnot assessed

Transaction100

End-to-end100

What breaks 0 issues

Worth a look 1

Minor

This scan could identify the site's product pages

Product data

No product page could be confirmed on this site, so the eight product-data checks below could not run. Likely cause: no listing page was identified either.

Fix If this site does sell something, the product pages are not reachable from its sitemap or its homepage links in a form this scan could confirm — which is itself worth knowing, because an agent discovers products the same way. If it does not sell anything, this is expected and can be ignored.

pages discovered16

listing pages identified0

sitemap files read2 of 2

Where to look

https://losen.meta-machine.org/the entry page this scan started from

25 checks passed

✓ Real AI assistants are permitted to read your products

✓ The site serves pages to an honest agent client

✓ The response body is the resource that was requested

✓ Agent persona is not treated worse than baseline

✓ No bot challenge served to the agent persona

✓ robots.txt does not blanket-block AI agents

✓ Content is present without JavaScript execution

✓ Product pages are not marked noindex

✓ An agent can operate the site's actions and get a result

✓ Nothing suppresses the snippet AI answers are built from

✓ Deep-linked URLs from the index still resolve

✓ Checkout does not depend solely on an app handoff

✓ Declaring agent intent does not worsen treatment

✓ The page an agent receives is the site, not a gate in front of it

✓ Pages answer inside an assistant's patience

✓ Pages have distinctive titles

✓ Sitemap exists and lists content URLs

✓ The homepage exposes a traversable link structure

✓ Landing on a URL delivers the page, not a client-side hop

✓ Page structure is machine-navigable

✓ Sitemap advertises when content changed

✓ Rate limiting is polite and signposted

✓ Pages have a meta description

✓ Unknown URLs return 404, not 200

✓ Bot-management layer identified

Conditions this report was produced under

Identity — Measured under our own honest user-agent. We are not a named assistant, so a site that special-cases assistant tokens does not recognise us — what it serves us may differ from what it serves them. User-Agent: Mozilla/5.0 (compatible; AgentScan/0.1; ai-agent; autonomous-shopper; +https://github.com/agent-scan).
Vantage — Scanned from a single network location. Sites that route by IP may serve assistants a different currency, language or interstitial.
Coverage — 12 of 12 requested page(s) were served, and 26 of 74 checks had something to measure. This is a sample, not an audit.
Freshness — measured on 2026-08-09. This describes the site as it answered on that day; a later deploy, or a change to a bot rule, can change any finding here.

What this report covers

Measured: whether an AI assistant that reaches your site can navigate and use it — read the product, express a customer's constraints as a request, get usable results, and establish that the item can be bought. Findings reflect the raw HTML an assistant receives; the major assistants do not execute JavaScript, so this is what they actually see.

Not measured: whether assistants reach you at all. Enrolment in a vendor product feed bypasses your website entirely and cannot be detected by scanning. Search-index membership is only partly observable — we check the things that block it (noindex directives, missing titles, stale sitemaps, agent robots policy), not your actual index coverage. Findings taken under our own identity say "refused for this scanner", never that a named assistant is refused; where this report tests a named identity it says what your site served a request claiming that name, which is not the same as what the vendor's own fetcher receives.

We check that data is readable, not that it is correct. If your markup says a product costs 1 299 kr and is in stock, we verify an agent can read that — not that it is true. Stale or wrong structured data will pass these checks and still mislead a customer, so keep it in sync with your source of truth.

Point in time, single location. This reflects the site as it was on 2026-08-09, fetched from one network location. A later deploy can change any finding, and sites that route by IP may serve assistants a different currency, language or interstitial than we saw.

A good score means an assistant that arrives can use this site. It does not by itself mean you are winning at AI-driven sales.

Generated by agent-scan — a declared agent that identifies itself, respects robots.txt, and caps its own request budget. Its default identity is its own. Where a named agent's token was sent, it is labelled as such and every result is worded as what your site serves a request claiming that name.