What it does
A security framework that watches the communication between AI agents and MCP servers in real time. It analyzes the patterns and intent of tool calls to detect and block behavior it considers malicious. The desktop interface, built with Electron, is where you view the monitored activity and manage the framework.
Use cases
- 01Watch the tool calls an agent makes in real time
- 02Flag an agent that starts calling unexpected tools
- 03Review detected events in the desktop interface