🔓Shopify Open-Sources Overexposed, Cutting Inconclusive Ingress Verdicts By 80%
Your public Ingresses might be open and you wouldn't know
TL;DR
Shopify built Overexposed, an internal scanner that finds Kubernetes Ingresses and Cloud Run services reachable from the internet without authentication. The rebuild cut inconclusive verdicts by around 80%. If you run on GKE, check your exposure before someone else does.
Shopify's security team rebuilt its internal exposure scanner, and the new version finds internet-facing endpoints the old one missed. The tool is called Overexposed. It hunts for Kubernetes Ingresses on GKE and Cloud Run services that anyone can reach without logging in. If you own a GKE cluster or deploy to Cloud Run, this is the kind of tool you want pointed at your own infrastructure. The old version checked the infrastructure repo and probed public hostnames, which meant false alarms and blind spots. The fix was to stop trusting config files and start from what GCP actually has running. Overexposed now queries Cloud Asset Inventory, which includes Ingresses GKE accepts after admission, so it sees what really exists. The numbers matter here. Overexposed sends a HEAD request first, falls back to GET, and follows at most three redirects, rechecking each hop. It returns six verdicts: exposed, expected, protected, not found, unreachable, and inconclusive. Uncertain evidence lands in inconclusive, never protected. Shopify says the rebuild cut inconclusive Ingress verdicts by around 80%. The team also built the thing with AI coding agents and AI code review, then ran mutation testing against its security guards.

Key Points
Overexposed queries GCP Cloud Asset Inventory, which includes Ingresses GKE accepts after admission, plus Cloud Run services and load balancers, instead of trusting infrastructure repo files.
Keep reading — it's free
Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.
Already a member? Sign in
Comments
Be the first to comment