Uber open-sources ADR: enterprise AI agent security and threat detection framework

uber · github · 2026-08-04

Uber has open-sourced ADR, its internally deployed AI agent security project. The framework targets enterprise AI agents, providing observability, security benchmarking, and threat detection. It specifically defends against LLM security risks like prompt injection and is compatible with mainstream models and protocols like Claude, Codex, Cursor, and MCP.

Related event: Uber Open-Sources ADR Framework for AI Agent Security(3 posts)→

Original post →

More from coding & agent

coding & agent channel →