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

Adrian

Adrian is an open-source runtime security monitoring and control tool for AI agents. It analyzes agent activity and reasoning traces to detect potentially malicious or out-of-remit behavior, and can audit or intervene before actions proceed.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Adrian — 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 Adrian when agent security needs to be assessed during execution, particularly when monitoring reasoning as well as activity logs matters or when risky tool calls should be blocked or held for approval.

Who it is for

  • Developers adding runtime monitoring and intervention to AI-agent applications.
  • Teams building agents with Python, LangChain or LangGraph.
  • Claude Code users who want tool-call classification and approval or blocking in the terminal.

Documented uses

  • Review agent tool calls and actions against the agent’s intended remit.
  • Watch reasoning and behavior for prompt-injection indicators and policy drift.
  • Audit activity or block/hold actions for intervention.
  • Apply runtime controls to Claude Code tool calls.
Before you choose

Requirements & limits

The repository identifies Adrian as open source under the Apache-2.0 license. Its README documents a managed dashboard requiring sign-up and an API key for the LangChain SDK setup, as well as a self-hosting option.

  • Choose between the SDK, TypeScript integration, Claude Code plugin, or self-hosting route based on where the agent runs.
  • Configure audit versus blocking behavior and the agent’s remit and accepted behaviors.
  • The LangChain SDK setup relies on an Adrian API key; the Claude Code plugin has a separate setup guide.

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 Adrian with a LangChain agent

Configure the SDK to monitor a LangChain or LangGraph agent through Adrian’s managed dashboard.

Before you start

  • An account at Adrian’s managed dashboard and an API key.
  • A Python LangChain or LangGraph agent and the provider package for its model.

Steps

  1. Sign up at the dashboard and generate an API key.

  2. Configure the agent’s remit, audit or block mode, alerting preferences, and accepted behaviors or known risks.

  3. Install the Adrian SDK with the command below.

    Documented command
    pip install adrian-sdk

  4. Install LangChain and the appropriate model provider package.

  5. Initialize Adrian before the agent code and shut it down after the code, following the README example.

Expected outcome

A LangChain or LangGraph agent is configured to send activity for Adrian’s runtime monitoring and control.

Guide limitations

  • The documented SDK setup requires an Adrian API key.
  • The README lists supported LangChain-related version ranges; check the current documentation before selecting dependencies.
Documentation 1 ↗
Evidence and transparency

Sources & review

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