Definition
A record of a production incident — what happened, when it occurred, how severe it was, and what the root cause turned out to be. In Well Tested, IncidentReports are created manually or linked automatically to schema changes via the Memory Graph. They are the evidence layer that enables recurrence risk detection: without knowing what broke before, there is no way to surface 'this pattern has caused problems 3 times.'
Why it matters
Incidents without root cause analysis are wasted learning. Every incident is an opportunity to build institutional memory — to encode 'this is what happens when we change that column.' Without a documented incident report linked to the schema change that caused it, the same mistake will happen again in 6 months and the team will have to rediscover it from scratch.
How Well Tested handles it
Well Tested's IncidentReport model records title, severity, occurredAt, resolvedAt, and root cause. Each incident can be linked to one or more schema changes via the SchemaChangeIncidentCorrelation model with a correlation type (time_window, manual, root_cause) and confidence score. Correlated incidents feed the Memory Graph, enabling recurrence risk detection.
Related terms
No commitment. 20 minutes. We review your release workflow and show you what's possible.