Tag
When Coinbase's ingress gateway failed on 14 July 2026, the path engineers needed to reach their own rollback tooling ran through that same gateway. AWS hit an identical loop nine months earlier.
A July 2 Railway outage report shows the real damage came after the fix: a containment decision that removed the last default route, and a well-known Linux default that turned a fallback path into a silent bottleneck.
A Cursor agent found one unscoped API token and wiped a production database and its backups in nine seconds. The real failure was credential scoping, not the model.
Meta's Instagram breach traced to a basic authentication gap, not a sophisticated attack, after its Trust and Safety team lost half its staff to an AI reassignment. No public postmortem has followed.
A single AWS zone failure turned into an 18-hour Coinbase outage. The postmortem reveals two specific ways 'multi-AZ' architecture quietly wasn't, and how to check your own systems for the same gap.
Most postmortems document everything and change nothing. The failure is in the document design, not the analysis. Here is the structure that works for the three distinct readers every incident review actually has.
Most engineering postmortems are structurally correct and practically useless outside the team that wrote them. The problem is not the template — it is that postmortems are written for the author, not the reader.
Practical guides, product updates, and compliance notes — straight to your inbox. No fluff.
Newsletter is opening soon. We'll switch this on once we've got our first issue ready.