A clearer picture of your website
Can an agent make sense of your website?
Give AI agents a clearer path through your website. Find unclear controls, inspect your page structure, and see which discovery signals you publish.
Start with your website.
One public URL. Practical insights, right here in your browser.
No account. No newsletter. Your results stay on this page.
What we check
See what's working, what needs attention, and what to improve next.
-
A closer look
Agents
Readable interfaces, clear actions, and optional agent discovery signals.
-
A page machines can understand
A clear title, declared language and main landmark give an agent useful context. We also compare text in fetched and rendered HTML.
-
Actions with a clear purpose
Links, buttons and form controls need usable names. Real browser checks find missing labels and focusable elements hidden from accessibility tools.
-
Discoverable agent interfaces
We inspect declarative WebMCP form names and descriptions, plus the optional llms.txt at your origin root. Tools are never invoked.
-
Access on your terms
Search crawler and training crawler policies are reported separately. Blocking model training is a choice, not a failed check.
-
A public-page inspection with evidence for every finding. No sign-in, form submissions or tool execution.
From URL to a clear next step.
-
Enter a page
Choose the public page you want to understand.
-
Follow the checks
See the actual scan activity as it happens.
-
Find your next step
Review the findings and the most useful fixes.
Good to know
Questions, answered.
Know what to expect, understand your results, and make your next improvement with confidence.
About the check
What does the Agents check actually test?
It renders your public page and checks twelve interface rules with axe-core: document identity, language, main landmark, named links and controls, form labels, and focusable content hidden from accessibility tools. It also inspects HTML text, declarative WebMCP forms, root llms.txt and robots.txt.
Does this guarantee an AI agent can use my site?
No. This is a passive technical baseline, not an agent certification or a Lighthouse Agentic Browsing score. Login, purchases, tool execution, consent and task completion need separate testing. WebMCP and llms.txt are optional; their absence does not lower the score.
Does the free check scan my entire website?
The free check audits the page at the URL you enter, with some domain-level checks. It does not crawl every page. Use All checks for all six categories, or choose a category page for a focused check.
What will I see while the check runs?
The scan studio shows a website preview when available, the active categories, and real scan events as results arrive. Duration depends on the page and server load. The final score appears when the check completes.
How is the score out of 100 calculated?
Each category has its own scoring rules. The overall score combines the available category scores using configured weights. Open the scoring explanation in your results to see the default weights and how accessibility findings affect the score.
Working with your results
What do the different result groups mean?
Passed means the automated rule found no issue. Needs attention highlights a finding. Manual review means a person needs to check it. Not applicable means the elements or conditions for that rule were absent; it is not counted as a pass.
Why can a score of 100 still include findings?
Not every finding reduces the score. For example, accessibility best-practice findings and manual-review items are shown separately from WCAG deductions. A score of 100 does not mean every aspect of the page has been verified.
How do I fix the issues in my results?
Open a finding to read its explanation and recommended next step. Where available, you will also see affected elements, HTML examples, and links to rule documentation. Make the changes in your website, then run a new check to verify them.
Why can results change between scans?
Page content, third-party scripts, network conditions, and server response times can change between runs. Performance measurements are particularly sensitive to these conditions. Compare the same URL and review individual findings as well as the score.
What if my page cannot be checked?
Check that the URL is correct and publicly reachable. Login screens, bot protection, server errors, or timeouts can prevent a complete check. Review the scan activity for details and try again after resolving the access issue.