This repository is the public home for ADR: An Agentic Detection System for Enterprise Agentic AI Security.

The code and documentation are coming soon.

In the meantime, you can read the paper here:

ADR: An Agentic Detection System for Enterprise Agentic AI Security

ADR provides observability and detection for agentic AI systems. The ADR Sensor captures high-fidelity AI agent telemetry, including prompts, MCP activity, reasoning traces, tool calls, and execution context, while ADR Detection analyzes that data to identify unsafe behavior such as credential exposure, prompt injection, data exfiltration, and malicious or policy-violating tool use.

We are preparing the initial open-source release and will publish the project code here once review is complete.