An Implementation of the Microsoft Agent Governance Toolkit for Safe AI Agent Tool Use with Policies, Approvals, Audit Logs, and Risk Controls

A Colab notebook shows how to use Microsoft’s Agent Governance Toolkit to place a governance layer in front of agent tool calls.

The policy system evaluates identity, trust score, risk tier, tool type, sensitivity, and rules before allowing execution.

YAML rules block destructive database actions, require approval for external email, sandbox shell commands, restrict sensitive data access, and control financial transfers.

The notebook also adds audit logs, tests, a kill switch, summaries, and a relationship graph to make agent actions traceable and safer.