goatnote-inc/ScribeGOAT2 Safety Evaluation
Multi-turn safety persistence evaluation for healthcare AI models. Gates CI pipelines on safety-critical failure detection.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| model-id | Model ID to evaluate (e.g., gpt-5.2, claude-opus-4-6) | yes | — |
| provider | Provider (openai, anthropic, auto) | no | auto |
| scenarios | Scenario filter (all, escalation, defer, balanced) | no | balanced |
| fail-threshold | Pass rate threshold for CI gate (0.0-1.0) | no | 0.95 |
| n-trials | Number of trials per scenario | no | 1 |
Outputs
| name | description |
|---|---|
| gate-result | PASS or FAIL |
| pass-rate | Observed pass rate (0.0-1.0) |
| json-path | Path to JSON results file |