OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Security & Sandboxing

Adr

Adr is an open-source enterprise AI agent security system comprising tools for endpoint discovery, agent telemetry collection, security benchmarking, and suspicious-session detection.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Adr — OpenViST project profile in Agent Security & Sandboxing.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Adr when an organization needs a technical workflow for discovering agent-related tools, gathering activity telemetry, benchmarking defenses, or detecting suspicious agent sessions.

Who it is for

  • Organizations inventorying AI applications, agents, IDE extensions, model runtimes, or MCP servers on employee endpoints.
  • Teams monitoring activity across employee-facing coding agents or customer-facing AI agents.
  • Security practitioners evaluating agent defenses or analyzing suspicious agent sessions.

Documented uses

  • Inventory AI tools on employee endpoints and flag unknown surfaces for review.
  • Collect and normalize agent telemetry, including intent, tool use, and execution traces.
  • Evaluate defenses with ADR-Bench security tasks.
  • Detect risky agent behavior using a two-tier detector.
Before you choose

Requirements & limits

The repository is public and licensed under Apache-2.0. The Detection quick start uses `uv` and Anthropic and OpenAI API keys for the default detector; a keyless smoke-test option is documented for `llamafirewall`.

  • The repository separates discovery, telemetry, benchmarking, and detection into distinct components.
  • The README identifies Sensor support for macOS, Linux, and Windows and names multiple coding-agent integrations.
  • The open-source release does not include ADR Prevention or the offline ADR Explorer engine.
  • The Detection quick start requires two provider API keys for its default detector.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up ADR Detection

Prepare the repository's ADR Detection environment and configure credentials for its default detector.

Before you start

  • Git
  • uv
  • Anthropic API key
  • OpenAI API key

Steps

  1. Clone https://github.com/uber/ADR.

  2. Change to the ADR/Detection directory.

  3. Run the command below.

    Documented command
    uv sync

  4. Set the ANTHROPIC_API_KEY and OPENAI_API_KEY environment variables.

Expected outcome

The Detection environment is set up with the API credentials used by the default ADR dual-agent detector.

Guide limitations

  • The README's keyless smoke-test option uses the llamafirewall detector rather than the default detector.
  • The quick start documents setup; consult the repository's reproducibility workflow for benchmark execution.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
9 October 2026
Original website title
uber/ADR · Official homepage title · retained in its source language
How we review sourcesSuggest a correction