Filters
Time window
Impact
Component
At a glance
Stability over time — is Claude getting more or less reliable?
Same metrics across rolling windows ending today. Arrows compare the shorter window's
incident rate to the 12-month average (up = worse than average, down = better).
Filters above do not affect this section — it's always a whole-dataset snapshot.
| Window | Incidents | Per week | Critical+major | Median duration | Mean duration | Total downtime | vs 12-mo avg |
|---|
Charts
Incidents per month, stacked by impact
Counts fall in the month each incident started.
Total downtime per month
Sum of
resolved_at − started_at across all incidents that started in the month.Duration distribution
How long outages typically last, in the filtered window.
Day of week (your local time)
When do outages start? Day of week the incident kicked off.
Hour of day (your local time)
Hour-of-day histogram of incident starts. Helpful for spotting deploy windows.
Incidents by affected component
Counts how often each component went degraded or worse in the filtered window.
Incidents in window
| Started | Duration | Impact | Components | Summary |
|---|