2026-05-24
Link: https://github.com/orgs/community/discussions/51573
HN Discussion: 1 points, 0 comments
This is the kind of quiet, infrastructure-level story that should be raising alarms across every security team relying on GitHub's built-in tooling — but it's sitting at one point with zero comments. The linked thread is a GitHub Community discussion (#51573) reporting that Dependabot and code scanning security alerts are silently failing to deliver notifications. Not malformed. Not delayed. Just… not arriving.
If you're a maintainer or a security engineer, the implications are uncomfortable. The entire premise of GitHub's security suite — Dependabot alerts, secret scanning, CodeQL findings — rests on the assumption that when something bad is detected, the right humans get pinged. Email, web notifications, mobile push: if those channels are dropping messages without any visible error, then your CVE response window isn't what your dashboard says it is. You might be sitting on a critical vulnerability with the alert technically generated but never seen.
What makes this worth a closer look:
The thread itself is community-driven, which means it's the canonical place where affected users compare notes, post repro steps, and (eventually) extract a response from GitHub staff. Even if you're not currently affected, the discussion is worth scanning to understand the failure mode — whether it's tied to org-level notification settings, specific alert types, or a broader delivery pipeline regression.
The broader lesson, beyond GitHub specifically: any security control whose effectiveness depends on a notification channel needs an independent heartbeat. A weekly synthetic alert that you confirm receipt of. A scheduled job that diffs the Security tab against what your team has actually triaged. Otherwise you're trusting the pipe, and pipes leak quietly.
