microsoft/agent-governance-toolkit
AI Agent Governance Toolkit — Policy enforcement, zero-trust identity, execution sandboxing, and reliability engineering for autonomous AI agents. Covers 10/10 OWASP Agentic Top 10.
What it solves
它解決了自主 AI 代理執行未授權或破壞性行為的風險。由於提示層面的安全性是機率性的，且容易受到注入攻擊，此工具組提供一個確定性的執行層，會在工具呼叫與訊息執行之前攔截，讓未授權的行為在結構上變成不可能，而不只是可能性低。
How it works
工具組作為 AI 代理與其工具之間的治理核心。它使用政策引擎（支援 YAML、OPA 與 Cedar）來評估每一個請求的動作是否符合規則。若動作被允許則繼續執行；若被拒絕則拋出 GovernanceDenied 錯誤。系統同時整合身分管理（SPIFFE/DID）與防篡改稽核日誌，精確記錄是哪個代理執行了哪項動作以及為何被允許。
Who it’s for
需要將自主代理上線的開發者與 SRE，必須要有嚴格的政策執行、符合法規遵循（例如 OWASP Agentic Top 10）以及在多代理系統中提供詳細稽核能力。
Highlights
- Multi-Language Support: SDKs available for Python, TypeScript, .NET, Rust, and Go.
- Framework Agnostic: Integrates with major frameworks including Semantic Kernel, AutoGen, LangGraph, CrewAI, and OpenAI Agents SDK.
- Deterministic Enforcement: Moves safety from the prompt layer to the application code layer to prevent prompt injection bypasses.
- Comprehensive Tooling: Includes a CLI for OWASP compliance checks, policy linting, and prompt injection auditing.
- Advanced Governance: Features like execution sandboxing (privilege rings), a kill switch, and an MCP Security Gateway for tool poisoning detection.