agent-accountability
Governance layer for AI agents. Every action intercepted, audited, proven. The AI cannot bypass it.
The runtime coat around agent loops.