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

Agent Governance Toolkit

Agent Governance Toolkit is an MIT-licensed, public Python framework that applies policy enforcement, zero-trust identity, execution sandboxing and reliability engineering controls to autonomous AI agents.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Agent Governance Toolkit — 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 it when an autonomous agent can call tools, send messages or delegate work and those actions need policy evaluation, logging or sandboxing outside the model prompt.

Who it is for

  • Developers building Python-based autonomous-agent systems
  • Teams adding governance to existing agent applications
  • Operators needing policy checks and audit records around agent actions
  • Security and compliance teams working with agentic systems

Documented uses

  • Enforce policies on agent tool calls and other actions
  • Create identity controls for agents and their actions
  • Run agent execution inside a sandbox
  • Add reliability and governance controls to an agent system
  • Use application-level controls around an existing Python agent
Before you choose

Requirements & limits

The repository is public and MIT-licensed. The documented quick start requires Python 3.11 or newer and uses the full package extra. The project is labelled Public Preview and may change before GA.

  • The quick-start integration uses the agentmesh.governance govern wrapper and YAML policies.
  • The toolkit documents policy evaluation, audit logging and GovernanceDenied errors for blocked calls.
  • The base package and the full extra have different roles: the base wheel provides the compliance CLI, while the governance modules are supplied through the consolidated core distribution or full extra.
  • The repository identifies Linux, Windows, macOS, CLI and self-hosted contexts.
  • Current migration notes include a deprecated agent-os-kernel distribution, ACS SDK use for policy-engine host code and an agt-policies v4-to-v5 migration command.

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

Add policy enforcement to a Python agent tool

Install the toolkit and wrap an existing tool with a YAML governance policy.

Before you start

  • Python 3.11 or newer
  • The agent-governance-toolkit package with the full extra
  • An existing tool function
  • A YAML policy

Steps

  1. Install the full package with pip.

  2. Import govern from agentmesh.governance.

  3. Wrap the tool with govern and point it to the YAML policy.

  4. Use the governed tool so each call is evaluated, logged and denied when the policy blocks it.

Expected outcome

The tool is connected to the documented governance wrapper, which evaluates policy, records the decision and raises GovernanceDenied for blocked actions.

Guide limitations

  • The repository labels the project Public Preview and warns of possible breaking changes before GA.
  • The base wheel installs the compliance CLI only; the governance modules require the consolidated core distribution or the full extra.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
2 October 2026
Original website title
microsoft/agent-governance-toolkit · Official homepage title · retained in its source language
How we review sourcesSuggest a correction