Primary sources, visible interpretation.
Every case links to an original incident report, investigation, or official account. Summaries condense that record. Failure-pattern tags, diagrams, questions, and lessons are our editorial interpretation, not quotations from investigators.
The collection is selective. Different incidents have different scopes and measures of impact; we do not combine them into a severity ranking.
People operate inside systems.
A mistaken command is a starting point for investigation, not a complete explanation. We look for the conditions that allowed it, the controls that failed to contain it, and the recovery path that followed.
Real outages affect people’s work and lives. The museum’s tone should make the learning memorable without making light of those consequences.
Rehearsals are authored exercises.
The lab uses fictional scenarios with predefined branches. It does not connect to your infrastructure or predict your next incident. Historical cases and fictional exercises are labeled separately.
Disaster Check organizes the answers you provide. It does not independently verify evidence, scan your system, or certify readiness. Live AI is not connected.
Your reading list stays here.
Saved cases and read markers are stored in this browser. They are not synced to an account. Clearing this site’s browser storage removes that list.
System descriptions, evidence notes, and action plans stay in page memory unless you explicitly copy, download, or print them. Refreshing or leaving the page clears those entries. Do not enter secrets or customer information.
The application adds no advertising or analytics trackers. The hosting service handles network requests and access controls under its own policies.
A living collection.
We favor a clear, source-linked record over a dramatic claim we cannot support. Original reports can evolve and sources can move. Read the source notes with each case, especially before using it in training or a decision.
Find a place to start →