Free tools / Agent readiness grader
Agent readiness grader: can an AI agent read and use this page?
Our agent readiness grader checks one page against ten published access and usability checks for AI agents, the ones that fetch or use a page when a person asks, and grades it A to F. It covers robots.txt rules for ChatGPT-User, Claude-User and Perplexity-User, the answer an agent user agent gets, challenge pages, raw HTML text, and names on links, buttons and form fields.
We read robots.txt before anything else and follow it, then read your page within eight seconds: up to three redirects, each checked again and each to a public HTTPS address, and at most 1.5 MB. HTTPS websites are supported; other addresses are refused. We request the page twice: once with a browser user agent and once with a user agent that names the agent you choose and Broadcastwell.
No email is needed; runs are limited per network and per day, and by the site's daily allowance.
We do not save your website, company name, page content or result. We keep daily totals and hashed network counters to enforce limits. A shared result stays in its link, not on our server.
What we check and why
Access (55 points)
robots.txt allows ChatGPT-User, Claude-User and Perplexity-User at the path (15); a user agent naming the agent gets the same answer as a browser (15); the main text is in the raw HTML (15); and neither answer is a challenge or CAPTCHA page (10).
Usability (45 points)
Links have text or a label (10); form fields have a label (10); buttons have a name (5); images have an alt attribute (5); the page sets its language and title (5); and its JSON-LD parses (10).
Grades
A at 90 points or more, B at 75 points or more, C at 60 points or more, D at 40 points or more, and F below 40. A passing check earns its full weight and a partial result earns half.
Names an agent can read
An agent that fills a form or follows a link reads the same names a screen reader does. We read them from the HTML, without running scripts, so names a script adds later are not counted.
Our request is not the real agent
The second request names the agent and Broadcastwell. A site that checks addresses sees it is not the real agent, so treat a different answer as a lead to check in your firewall settings.
Questions
What is an AI agent, as opposed to a crawler?
A crawler collects pages ahead of time. An agent fetches or uses a page when a person asks an assistant to, as ChatGPT-User, Claude-User and Perplexity-User do; OpenAI, Anthropic and Perplexity document these names on their crawler pages.
Why do labels and button names matter to an agent?
An agent that fills a form or follows a link reads the same names a screen reader does. A field with no label or an icon button with no name gives it nothing to read, which makes the step harder to complete correctly.
Does an A grade mean agents will complete tasks on our site?
No. The grade covers what the page serves and how it is labeled. Whether an assistant finds the page, names your company or completes a task is a separate observation to measure.
Want the fix made for you?
Answer Repair completes one agreed source-backed page or fix. Choose Live or Handoff before payment. Due 48 elapsed hours after Ready. Full refund on request within 30 days of delivery.
When you want us to do the work, every offer and its published price is on our offers page. More free checks: the ChatGPT Business Hub readiness check, the AI access checker, the crawler identity inspector and the website audit tool, or see all free tools. Read next: What is agent discovery optimization? and What can an AI assistant learn from a software website?