SENTINEL AI logo SENTINEL AI TAHAI Web Services
SENTINEL AI

Trust posture

Security

SENTINEL is designed around explicit boundaries, approved access, review gates, and traceable action.

Approved access

Local folders and tools should be explicitly exposed; arbitrary browser filesystem access is not the model.

Policy controls

Input, retrieval, tools, output, code, and media can be governed by risk tiers and approval requirements.

Trace evidence

Important actions should leave route, policy, source, and actor records.

No bundled secrets

Repository artifacts should not include secrets, runtime databases, model weights, caches, or private source material.

Responsible disclosure

Security reports need a clear contact path and expected handling model.

Production evidence

CI, scanners, SBOM, signed artifacts, and runtime proof should gate production claims.

Security is product UXPeople trust what they can inspect and control.
Fail closedUnavailable proof should be labeled unavailable, not papered over.