Security gateway for AI agents with human approval
No updates yet.
Agents that can act need a leash, not a permanent API key. AgentLock holds sensitive actions until a human approves them, runs them with credentials the agent never sees, and logs everything append-only.
