Skip to content
Open Pulseagent view

Free agent readiness check — Open Pulse

Paste your URL. We check the roughly twenty paths an AI agent looks for before it can act on a site — an MCP server, OAuth protected-resource metadata, markdown content negotiation, an API catalogue, llms.txt and an explicit decision about AI crawlers — and score them out of 100. Free, no email needed to start.

What this page answers

How is this different from the AI visibility check?

They answer two halves of one question. The visibility check asks whether assistants can *read* and cite you, which is about prose. This asks whether an agent can *act* — whether there is a machine surface here it could call, and a way to authenticate against it. A site often scores well on one and badly on the other.

What does it cost?

Nothing, and no email is needed to start. The score and every individual check, with its evidence, are free. A confirmed email address opens the step-by-step fix instructions and a prompt you can paste into a coding agent.

What does it actually check?

Five categories: Discoverability (Whether an agent can find out what this site offers without guessing.); Content accessibility (Whether the content arrives in a form a model can read cheaply.); Bot access control (Whether the rules for agents are an explicit decision or an accident.); Protocol discovery (Whether there is a machine surface here, and a way to authenticate against it.); Commerce (Whether an agent can transact, where there is anything to transact for.).

Does it penalise me for standards nobody has agreed on?

No, and that is deliberate. A check only costs points when its discovery mechanism is ratified — an RFC, or a convention with a stable published path. Emerging formats earn credit and never cost any. Checks that do not apply to your site, like commerce protocols on a blog, are excluded from the score entirely rather than counted against you.

Does it run JavaScript?

No. It reads what a crawler gets on the first pass, which is also the honest way to measure this: a tool surface that only exists after hydration is one most agents never see. It is why WebMCP cannot be checked from outside at all.

At a glance

Cost
Free, and no email needed to start
You provide
A URL
Checked
About twenty well-known paths, headers and robots rules
Scored
Out of 100, across five weighted categories
Free
The score and every check, with verbatim evidence
Behind an email
The fix instructions and a coding-agent prompt

What the check covers

  • robots.txt, the sitemap, llms.txt and Link response headers.
  • Whether your content is in the served HTML, and whether it is available as markdown under Accept: text/markdown.
  • Whether the retrieval crawlers are explicitly allowed, and whether Content Signals state what may be done after a fetch.
  • OAuth protected-resource metadata (RFC 9728), authorization server metadata (RFC 8414), an API catalogue (RFC 9727), an OpenAPI description and an MCP server.
  • The agentic commerce protocols, where the site has anything to sell.

Next

See also