Runs · Audit log
Every score on the book is a re-runnable public-evidence run.
One row per Claude scoring pass. Search, filter, expand for the full evidence trace, or pull the whole filtered set as JSON / CSV.
Runs shown
68 / 68
Evidence pulled
209
Avg duration
4.5 s
Last run
Jun 19, 11:38 PM
Search
Segment
Model
Confidence
Account
Segment
Model
Evidence
Latency
When
Inspect