X thread mockup · not published
Brandon Gilchrist · Defense at Machine Speed

Part 4 of Defense at Machine Speed: Scan Twice.

How does a vulnerability die in your program? Ticket opened, fix merged, ticket closed.

How do you know it is dead? That is a different question, and most programs cannot answer it.

1/ The attackers verify their work. ARTEX ships a dedicated retest agent: revisit every claimed finding, verdict reproduced, fixed, or inconclusive.

The offense built verification into the loop because unverified findings waste its time. Defense verifies almost nothing.

2/ The standard: a fix counts only when the exploit path fails on a re-check. Same attack, patched system, watch it fail.

A code review is a claim. A quiet scanner is a claim. The re-attack is the proof.

3/ Adopt the standard and you learn your real fix rate. Patches fix the reported instance and leave the class. They close the bug the scanner described and miss the variant the harness finds next pass.

Most programs have never measured fix survival. That is the number to get.

4/ Cadence follows. One scan is an event, and events decay; the estate changes daily. The loop: scan, verify by attempting the exploit, fix, re-attack to prove, feed failures back as data.

Measure outcomes: attack success rate, containment time, fix survival under re-attack, cost per resolved incident. Not findings counts.

5/ One org rule: verification stays independent of fixing. The ticket closes on the retest verdict, not on the patch author's word.

Attackers scan you continuously and retest themselves by architecture. Scan twice is the minimum reply: once to find, once to prove.

Full piece on Substack.