AgentLock

0 reviews

AgentLock is a security gateway that sits between AI agents and the services they act on. Every outbound call, whether an HTTP request, an MCP tool call, an SSH command or a browser action, is routed through a policy engine instead of going out directly. Low-risk reads can pass automatically, while anything that writes, spends or touches production waits for a human approval, delivered to the web dashboard or as a push notification on the mobile app and granted or denied in seconds. API keys and tokens never reach the agent. They are stored envelope-encrypted in AgentLock's vault and injected only at the moment of execution, so a prompt-injected or misbehaving agent cannot leak a credential it was never given, and secrets are redacted out of every stored request and out of the audit log. Every request, decision and result is written to an immutable audit trail with the payload, the policy that matched, who approved it and when: the evidence teams need for compliance reviews, incident analysis and simply knowing what their agents did overnight. AgentLock ships an MCP server, so Claude Code, Cursor and any MCP-capable agent connect with a few lines of configuration, and TypeScript SDKs are available for custom integrations.