{"id":236,"date":"2026-11-18T20:05:54","date_gmt":"2026-11-18T20:05:54","guid":{"rendered":"https:\/\/blockoala.com\/index.php\/2026\/11\/18\/shopifys-exposure-scanner-checks-cloud-resources-not-just-code\/"},"modified":"2026-11-18T20:05:54","modified_gmt":"2026-11-18T20:05:54","slug":"shopifys-exposure-scanner-checks-cloud-resources-not-just-code","status":"publish","type":"post","link":"https:\/\/blockoala.com\/index.php\/2026\/11\/18\/shopifys-exposure-scanner-checks-cloud-resources-not-just-code\/","title":{"rendered":"Shopify\u2019s exposure scanner checks cloud resources, not just code"},"content":{"rendered":"<p>In a report published on November 18, 2026, Shopify described how it rebuilt Overexposed, an internal tool for finding GKE Kubernetes Ingresses and Cloud Run services that might be reachable from the internet without authentication. The scanner now combines Google Cloud Asset Inventory data with anonymous HTTP requests, rather than relying only on configuration files in code repositories.<\/p>\n<p>Shopify said its first version could miss resources created by other repositories or platform tools. It could also produce false positives when teams configured proxy authentication directly. Admission controllers posed another complication: they could rewrite settings before a cluster accepted them, and not every cluster ran the same controllers. Repository configuration showed what was intended, but not necessarily what was running.<\/p>\n<p>The rebuilt scanner runs on a schedule. It queries Cloud Asset Inventory for GKE-accepted Ingresses, Cloud Run services and load balancers, and makes a separate query for access policies without needing cluster credentials. Shopify said this revealed Cloud Run services in projects the first version had never covered.<\/p>\n<p>For each candidate endpoint, Overexposed makes an anonymous request. It tries HEAD first, falls back to GET if needed, follows up to three redirects and checks each step of the redirect chain. It also tests path variants such as <code>\/ADMIN<\/code>, because path rules can be case-sensitive.<\/p>\n<p>The tool assigns one of six outcomes: exposed, expected, protected, not found, unreachable or inconclusive. \u201cExpected\u201d means a public endpoint exactly matches a reviewed allowlist entry in code, including its hostname and path. The scanner continues to probe allowlisted endpoints on every run and flags an entry as stale if the endpoint becomes protected or disappears.<\/p>\n<p>Shopify cautioned that an HTTP status code alone does not settle whether a service is protected. A 404 behind an Ingress shows the proxy responded, even if the requested path does not exist. For Cloud Run, Google\u2019s front end might return a 404 before the application runs, so the scanner treats that result as inconclusive. It does not automatically treat a 403 as proof of protection either. For Cloud Run, it examines both ingress settings and IAM: a service is public if <code>allUsers<\/code> has the invoker role or the invoker check is disabled. For services behind load balancers, URL Maps help identify the backend for each path and where authentication such as Identity-Aware Proxy is required.<\/p>\n<p>In one July scan, roughly three-quarters of results were inconclusive. Shopify grouped those results by cause, added definite classifications where evidence supported them and replayed saved scan results to test rule changes before release. After a few weeks of tuning, the share of inconclusive Ingress results fell by about 80%. Where evidence remains insufficient, the scanner keeps the inconclusive label rather than assuming a service is protected. It stores classifications, reasons and responsible teams, but not response bodies or credentials.<\/p>\n<p>Because the scanner sends requests to arbitrary hostnames, Shopify also hardened it against server-side request forgery. It checks every IP address returned by DNS, rejects private and loopback addresses, connects directly to an IP it has checked, and repeats the checks after redirects. The report\u2019s author also recommended restricting the scanner\u2019s outbound network access, including access to cloud metadata endpoints.<\/p>\n<p>Shopify said it used AI coding agents and AI code review while building Overexposed. To catch security tests that pass while checking the wrong behavior, the team uses mutation testing on safeguards and maintains a canary target for each classification outcome.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>In a report published on November 18, 2026, Shopify described how it rebuilt Overexposed, an internal tool for [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-236","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts\/236","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/comments?post=236"}],"version-history":[{"count":0,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts\/236\/revisions"}],"wp:attachment":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/media?parent=236"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/categories?post=236"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/tags?post=236"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}