Provider health
| Provider | Requests | Errors | Fallbacks | Models |
|---|---|---|---|---|
| Нет AI usage events за выбранный период | ||||
AI quality / evals / observability
LLMOps dashboard
Контроль качества AI: provider/model/privacy metadata, prompt templates, operator feedback, sanitized eval runs, schema success, errors, fallback rate and token usage. Raw prompts, raw outputs and secrets не сохраняются.
Requests0
Schema success0%
Error rate0%
Fallback rate0%
Filters
Dashboard reads internal usage/eval/feedback logs only.
Total tokens00 in · 0 out
Feedback wrong0operator negative feedback
Eval runs0recent internal evals
Safety—sanitized evals only
Safety policy
- Raw prompts stored—
- Raw outputs stored—
- Secrets stored—
- Marketplace writes—
Prompt templates
Versioned prompts with privacy level and expected JSON schema.
| Name | Task | Privacy | Status | Schema |
|---|
Submit feedback
Useful/wrong/needs_fix for future eval candidates. Secret-like content is blocked.
Create sanitized eval run
Creates offline metrics from synthetic/sanitized examples only.
Feedback
| ID | Rating | Task | Reason |
|---|
Eval runs
| ID | Name | Status | Metrics |
|---|