Agent readiness, measured from the outside
Scan your site
Enter your domain to see what an agent finds there. You get a score and a fix for every failed check, or the reason the site could not be measured. One scan asks your site for at most 136 URLs.
The scanner runs 61 checks from public rubric v0.7 with a plain HTTP client. Each check lists a shell command that repeats its measurement. Read the methodology
What happens after you scan
-
A fix for every failed check
Each failed check comes with the change to make and a shell command that repeats its measurement. A check that does not apply to your kind of site is out of scope, not a failure.
-
A re-scan after each fix
Ship a fix, then press Re-scan now on your report. When both scans have a score under the same rubric version, the new report shows how far it moved. Entering your domain again within 6 hours opens the stored report instead.
-
No irrelevant homework
The scanner decides what kind of site this is before it scores anything. Checks that only apply to other kinds of site are reported as not scored, with the reason — never as failures you are asked to fix.
-
A record you can verify
The report lists every request the scan logged. A SHA-256 fingerprint covers the check outcomes, and /verify recomputes the score from a record on the current rubric. A check that reads the clock, or a response slower than 12 s, can change the result with no change to the site.
Every scan is addressable as a path — /aws/<domain> — so a result can be bookmarked, shared with a coding agent, or fetched by a machine that has no form to fill in.
From a terminal, the same scan returns JSON. No key is needed, and the limit is 10 scans per client per minute.
curl 'https://agenticwebscore.com/api/scan?target=example.com'