← All writing
Projects · Vulnerability management 3 min read

Why I'm building a vulnerability prioritization tool

Sorting devices by vulnerability count felt logical, until I saw it put low-risk machines ahead of ones with actively exploited CVEs.


Part of my co-op has been patching devices with open CVEs. Some of those vulnerabilities were on CISA's Known Exploited Vulnerabilities list, which means attackers aren't just able to use them, they already are.

When I started, I did what felt obvious: sort devices by how many vulnerabilities they had and work from the top down. The machine with 40 findings got fixed before the machine with 3.

It didn't take long to see the problem. A device with dozens of low-risk findings could get patched before a device with one vulnerability that's being actively exploited right now. Count tells you how messy a machine is. It doesn't tell you how much danger it's in.

CVSS alone doesn't fix it

The next obvious move is to sort by CVSS severity. That's better, but it still misses a lot. Plenty of "critical" vulnerabilities are never exploited in the real world, and some "medium" ones are. CVSS tells you how bad something could be. It doesn't tell you whether anyone is actually using it.

Two free signals that help

Two free resources fill that gap:

  • CISA's KEV catalog lists vulnerabilities confirmed to be exploited in the wild. Federal agencies are required to patch them by set deadlines, which is a pretty strong hint about where to start.
  • EPSS, from FIRST, gives every CVE a score from 0 to 1 estimating how likely it is to be exploited in the next 30 days.

CVSS tells you how bad it could be. EPSS tells you how likely it is to actually happen.

What I'm building

So I'm building a Python tool that does the sorting for me. It takes an endpoint vulnerability export, checks each CVE against KEV, EPSS, and CVSS, factors in how many devices are affected, and outputs a ranked patch list, plus a CSV I can bring into Power BI.

The scoring idea, roughly:

score  = 40 if cve.in_kev else 0
score += cve.epss * 25          # 0–25
score += (cve.cvss / 10) * 20   # 0–20
score += reach_points(cve)      # 0–15, more weight for servers

Those weights are a starting point, not a final answer, and figuring out whether they hold up is part of the project. It only runs on generated sample data, never anything from work.

Beyond the tool itself, I want to get better at Python, understand vulnerability management more deeply, and learn more about how endpoint security fits together. I'll write about what I learn as it comes together. You can follow along on GitHub.

Related project Vulnerability Prioritization Tool Python · CISA KEV · EPSS · NVD · Power BI View on GitHub