The Agent Governance Toolkit is Microsoft’s open-source toolkit for governing autonomous AI agents, covering policy enforcement, zero-trust identity, execution sandboxing and reliability engineering. Published on GitHub under the MIT license, it has attracted about 6,400 stars and is actively developed, with updates in 2026.
As AI agents move from demos into production, they need the same controls that enterprises apply to any powerful automation: who the agent is, what it is allowed to do, where it can run, and how its failures are contained. This toolkit packages those controls for agent systems. It maps its capabilities to the OWASP Agentic Top 10, claiming coverage of all ten risk categories, and combines a policy engine with identity, sandboxing and reliability patterns so teams can deploy agents with guardrails instead of blind trust.
The Agent Governance Toolkit is free and open source under the MIT license. There is no paid edition or hosted plan; you integrate the toolkit into your own agent infrastructure at no cost.
This toolkit is for platform teams, security engineers and enterprises deploying autonomous AI agents in production who need governance, compliance and safety controls around what those agents can do.