The Threat Surface Probe detects CVEs that have a signature in its library. It runs on demand, and in scheduled reports at most once a day. It runs only against sites whose ownership your account has verified, because it sends test requests to your site rather than only reading public pages.
How it tests
- The probe sends non-destructive test requests to your site that check whether a known flaw responds. It does not only read pages: many signatures send crafted requests that test for the flaw itself rather than reading a version number. This is why it runs only on verified sites.
- Every scan runs four categories only: CVE detection, exposed panels, misconfigurations and default credentials. More intrusive categories are never run.
- Before the signatures run, every scan runs two checks of our own on your site's response headers; one of them tests whether the software versions your site announces are current.
- Your host name is resolved first; a host that resolves only to non-public addresses ends the run with that error. Every request, redirect hops and raw requests included, is then sent through a proxy that refuses any non-public address.
- A CVE counts as covered when the library contains a signature for its CVE id in the CVE category.
- You can scan only sites whose ownership your account has verified, and their subdomains.
What you see
- A finding per matched signature, with its name and severity, streamed as the scan runs.
- A scan with no signatures available is an error, never a clean result.
- A timeout is reported as a timeout, and a scan failure with no output as an error, not as "nothing found".
Limits
- The signatures are third-party. Many send crafted requests that test for the flaw rather than reading a version; this is why the probe requires ownership and runs in reports only for verified sites.
- The signature library is refreshed daily.
- A signature's quality is the library's; a CVE with a weak signature is covered only as well as that signature.