100 % agree.
I’ve lived through three Log4j-style fires.
The difference between a 30-minute panic and a 4-day all-hands meltdown?
Having (or not having) an SBOM that’s actually wired into CI.
My current rule on every new platform team I join:
- Every container image must ship an SBOM (Syft / Grype or Trivy)
- SBOM gets attached as an artifact stored in ECR image labels
- Dependabot Grype scan every push → auto-PR if CVE ≥ High
- On-call runbook step #1: “grep <CVE> *.sbom.json” → done in 45 seconds
Result:
Last time a critical vulnerability dropped (Spring4Shell, Kubernetes CVE, etc.), we knew we were clean in <5 minutes while the rest of the industry was still writing bash scripts.
SBOMs aren’t a compliance checkbox.
They’re the fastest way to turn a 3 AM page into “yawn, not us”.
Do it once properly, sleep forever after.
Great take, OP.