Appearance
Evaluation Engine
Evaluation Engine is /evaluations/settings. Open it from Configuration → Evaluations (or /evaluations, which redirects here). /settings/evaluation also redirects here. This is a standalone page, not a Configuration tab on an Analytics screen.
Required fields
| Field | Notes |
|---|---|
| Provider | From Manage Models |
| Model | Models for that provider, or a custom model name |
| API Key (Vault) | Select a Vault secret, or Create new |
Save with Save Changes or Create Config.
Auto Evaluation
Toggle Enable auto evaluation and set a cron schedule (for example */15 * * * *). Auto Evaluation runs enabled evaluators against newly ingested LLM traces.
There is no sample-rate control on this page.
Scheduled jobs require ELSAI_CRON_SECRET on the server. See Configuration.
Which types run is controlled on /evaluations/types. Manual Run Evaluation on a trace uses the same judge.
Manual evaluation
The page explains both Manual and Auto paths. Open a trace in Telemetry and run evaluation from its Evaluations panel, or use Go to Traces. See LLM-as-a-Judge.