ai-agent-governance

pip install ai-agent-governance[full]

⭐ If this project helps you, please star it! It helps others discover the agent governance stack.

🔗 Part of the Agent Governance Ecosystem — Installs Agent OS · AgentMesh · Agent Hypervisor · Agent SRE

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      agent-governance                           │
│                  pip install ai-agent-governance[full]           │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│   ┌───────────────────┐      ┌───────────────────────────┐     │
│   │   Agent OS Kernel │◄────►│   AgentMesh Platform      │     │
│   │                   │      │                           │     │
│   │  Policy Engine    │      │  Zero-Trust Identity      │     │
│   │  Capability Model │      │  Mutual TLS for Agents    │     │
│   │  Audit Logging    │      │  Encrypted Channels       │     │
│   │  Syscall Layer    │      │  Trust Scoring             │     │
│   └────────┬──────────┘      └─────────────┬─────────────┘     │
│            │                               │                   │
│            ▼                               ▼                   │
│   ┌───────────────────┐      ┌───────────────────────────┐     │
│   │ Agent Hypervisor  │      │   Agent SRE               │     │
│   │                   │      │                           │     │
│   │  Execution Rings  │      │  Health Monitoring        │     │
│   │  Resource Limits  │      │  SLO Enforcement          │     │
│   │  Runtime Sandboxing│     │  Incident Response        │     │
│   │  Kill Switch      │      │  Chaos Engineering        │     │
│   └───────────────────┘      └───────────────────────────┘     │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Quick Start

import asyncio
from agent_os import StatelessKernel, ExecutionContext
from agentmesh import AgentIdentity

# Boot the governance kernel
kernel = StatelessKernel()
ctx = ExecutionContext(agent_id="my-agent", policies=["read_only"])

# Establish zero-trust agent identity
identity = AgentIdentity.create(
    name="my-agent",
    sponsor="alice@company.com",
    capabilities=["read:data", "write:reports"],
)

# Execute a governed action
async def main():
    result = await kernel.execute(
        action="database_query",
        params={"query": "SELECT * FROM users"},
        context=ctx,
    )
    print(f"Success: {result.success}, Data: {result.data}")

asyncio.run(main())

Install only what you need:

# Core: kernel + trust mesh
pip install ai-agent-governance

# Full stack: adds hypervisor + SRE
pip install ai-agent-governance[full]

# À la carte
pip install ai-agent-governance[hypervisor]
pip install ai-agent-governance[sre]

Components

Star the ecosystem

Why a Unified Governance Stack?

Running AI agents in production without governance is like deploying microservices without TLS, RBAC, or monitoring. Each layer solves a different problem:

One install. Four layers of protection.

The meta-package ensures all components are version-compatible and properly integrated. No dependency conflicts, no version mismatches — just a single pip install to go from zero to production-grade agent governance.

The Agent Governance Ecosystem

agent-governance ─── The meta-package (you are here)
├── agent-os-kernel ─── Governance kernel
├── agentmesh-platform ─── Zero-trust mesh
├── agent-hypervisor ─── Runtime supervisor (optional)
└── agent-sre ─── Reliability engineering (optional)

Each component works standalone, but they're designed to work together. The kernel enforces policy, the mesh secures communication, the hypervisor controls execution, and SRE keeps everything running.

Examples

See the examples/ directory for runnable demos:

# Quick start — boot the governance stack in 30 lines
python examples/quickstart.py

# Full stack — all 4 layers working together
python examples/governed_agent.py

Framework Integration

# LangChain
pip install langchain ai-agent-governance

# CrewAI
pip install crewai ai-agent-governance

# AutoGen
pip install pyautogen ai-agent-governance

🗺️ Roadmap

Contributing

We welcome contributions! See our Contributing Guide for details.

For component-specific contributions, see:

• Agent OS
• AgentMesh
• Agent Hypervisor
• Agent SRE

License

MIT — see LICENSE for details.