ADR
Named by 4 kept items in this workspace.
Search every workspace for ADR. It matches the name as written, so it can find other things that share it.
The items that name it 4
Agent security system open sourced by Uber in August (paper at MLSys) that detects credential exposure and malicious/anomalous behavior in AI agent sessions by capturing the full chain of prompt, reasoning, tool call and outcome via an MCP based sensor inside Cursor, Claude Code and Codex, triaging
enterprise security system for AI agents; deployed in production at Uber; paper accepted to MLSys 2026
Uber's production security system for enterprise AI agents (uber/ADR): an observability sensor captures prompts, MCP activity, reasoning traces, tool calls and execution context, and a two-tier detector flags credential exposure, prompt injection, data exfiltration and policy-violating tool use.
Combines cross-client agent telemetry, ADR-Bench security scenarios, dual-agent detector for suspicious intent, tool use, execution traces
All technologies in this workspace.