Circuit Breakers
EvalsTool (not an eval — tooling/reference entry)
Representation-level defense against harmful outputs.
- Category
- Guardrails & safety classifiers (adjacent)
- Insurance exposure
- Regulatory/bias
- Publisher
- Gray Swan AI
- Publisher type
- Company (open source)
- License
- MIT
- Status
- Stale (>12mo)
- Last commit
- 2024-09-24
- Stars
- 270
Wanich Risk Lab, 2026.