What does Secret Watch do?
Keep secrets out of prompts, outputs and agent memory. Agents read documents, write code and call tools, and secrets leak through all of them. Secret Watch detects credentials in AI inputs, outputs and tool calls, redacts them, and rotates anything exposed.
- Stage: Concept.
- Solieum anchors: Hash of the exposure event (never the secret).
- Solieum anchors: Revocation proof for rotated credentials.
- Solieum anchors: Incident evidence commitment.
Frequently asked questions
Is this demonstration connected to real systems?
No. Demonstrations use synthetic data and never trade, transfer, sign or command anything real.