Boundaries
Boundaries
Audit results are useful only when read with the supplied model, configured bound, assumptions, and exclusions.
What the Result Means
UNSAFE
A bad path exists in the supplied model within the configured bound.
SAFE_WITHIN_BOUND
No bad path was found within the supplied model and configured bound. This is not a global safety claim.
What Remains Outside Scope
- It does not cover behavior outside the supplied model.
- It does not cover all real-world system behavior.
- It is not a replacement for security review.
- It does not model probabilistic LLM/tool behavior.
- It does not perform full IAM/cloud analysis.
- It does not establish independent benchmark validation or third-party repeatability.
Formal Research Context (Optional Background)
Separate formal research motivates the replay-to-constraint framing and claim discipline behind this demo. The audit's public claims stand on the supplied model, bound, trace, remediation, and retest artifacts. Formal context does not certify this implementation as production-safe and does not prove agent safety. See the technical appendix for optional background.