Agent Governance Toolkit

Active
GitHub Python MIT

Description

Microsoft's AI Agent Governance Toolkit providing policy enforcement, zero-trust identity, execution sandboxing, and reliability engineering for autonomous AI agents. Covers 10/10 OWASP Agentic Top 10.

Key Features

  • Deterministic policy enforcement engine: intercepts and evaluates all operations before model output reaches tool calls, with YAML policy files defining allow/deny/approval-required rules — denied actions are structurally impossible
  • Zero-trust identity and audit trail: assigns unique DIDs to each agent, generates tamper-evident audit records for every operation including active policy, request content, and decision rationale
  • Multi-language SDK support: Python, TypeScript, .NET, Rust, and Go SDKs with unified policy evaluation APIs covering major development stacks
  • Sandboxed execution: agent tool calls run in controlled sandboxes with restricted filesystem access, network calls, and database operation permissions
  • OWASP Agentic Top 10 complete coverage: built-in support for OWASP agent security risk classification with pre-built policy templates for accelerated compliance
  • MCP protocol integration: native Model Context Protocol support, operates as an MCP server providing governance capabilities to any MCP-compatible AI client

Use Cases

💡 Enterprise AI agent production deployment: provides production-grade security policy enforcement and operation auditing for autonomous agents, meeting compliance requirements in finance, healthcare, and regulated industries
💡 Permission management in multi-agent systems: precisely controls which tools and operations each agent can access in collaborative multi-agent scenarios
💡 Security red teaming and vulnerability assessment: automated attack probing against AI systems using OWASP framework, quantifying Attack Success Rate (ASR) metrics
💡 Compliance auditing and regulatory reporting: generates operation logs and audit reports meeting regulatory requirements, demonstrating controllable and traceable AI decision-making processes
💡 AI application security lifecycle integration: operates as a security gate in CI/CD pipelines, automatically validating policy compliance before agent deployment

Quick Start

Install: pip install agent-governance-toolkit[full]; govern any tool function in two lines: from agentmesh.governance import govern; safe_tool = govern(my_tool, policy="policy.yaml"); every call is automatically evaluated against the policy, logged, and raises GovernanceDenied if blocked

Related Projects