- Open source
- Experimental
- MIT licence
Actovault
Verification and runtime controls for AI agents: checking an application's critical rules before deployment, and enforcing them when an agent performs a supported action.
Works today
- An experimental implementation of the shared Core policy layer
- Written specifications for the Verify and Runtime modules
- Public architecture, contracts, threat model and roadmap documents
Limits
- There is no verification engine or production executor yet
- Names and interfaces may change
- Not suitable as a production security control



