Which vulnerabilities to patch first
Every published CVE, ranked by what predicts exploitation: CISA's evidence that a flaw is being exploited, then the daily EPSS probability. Rebuilt every day from CISA KEV, FIRST EPSS and the NVD.
Does the ranking work?
The numbers, by budget and by period
Patch first
The top of the ranking today. Known exploitation comes first; the deadline is the one CISA's directive BOD 26-04 sets for an internet-facing system.
Recently added to KEV
Who scores severity now
Our own dependencies
The same ranking applied to the Python dependencies of real systems, transitive ones included, matched against OSV.dev. The last one is an intentionally outdated example.