https://github.com/tnilabs/governed-agent-autonomy

Enterprise Agentic Maturity Model: AMM levels, controls, patterns, and reference implementations for governed AI agent autonomy.
https://github.com/tnilabs/governed-agent-autonomy

agent agent-orchestration agentic-engineering agentic-framework agentic-skills agentic-workflow agents ai ai-agents ai-tools autonomous autonomous-agents autonomous-systems enterprise-ai governance llm-agents llms

Last synced: 22 days ago
JSON representation

Enterprise Agentic Maturity Model: AMM levels, controls, patterns, and reference implementations for governed AI agent autonomy.

• Host: GitHub
• URL: https://github.com/tnilabs/governed-agent-autonomy
• Owner: tnilabs
• License: other
• Created: 2026-04-26T05:27:06.000Z (28 days ago)
• Default Branch: main
• Last Pushed: 2026-05-01T01:07:14.000Z (23 days ago)
• Last Synced: 2026-05-01T03:09:35.742Z (23 days ago)
• Topics: agent, agent-orchestration, agentic-engineering, agentic-framework, agentic-skills, agentic-workflow, agents, ai, ai-agents, ai-tools, autonomous, autonomous-agents, autonomous-systems, enterprise-ai, governance, llm-agents, llms
• Language: Python
• Homepage:
• Size: 3.15 MB
• Stars: 3
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Changelog: CHANGELOG.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE
  • Agents: AGENTS.md

Awesome Lists containing this project

README

          
# governed-agent-autonomy

AI agents are easy to demo and hard to trust in production. The hard part is

not making a model call a tool. The hard part is knowing what the agent is

allowed to do, what data it can use, who approved risky actions, what evidence

the run leaves behind, what happens when it fails, and how you prove the system

got safer instead of just faster.

The **Governed Agent Autonomy Model (GAAM)** gives teams a practical way to

answer those questions. "Governed" here means the agent's authority is bounded

by explicit rules, controls, evidence, and recovery paths. It is not paperwork

for paperwork's sake. It is how you keep an agent from quietly turning a useful

workflow into an expensive incident generator.

Use GAAM to:

- decide how much authority an agent should have right now;

- see what must be true before moving from drafting, to reading, to acting, to

  autonomous work;

- design approval, access, telemetry, rollback, and incident boundaries before

  they become postmortem material;

- assess whether a system's maturity claim is backed by evidence or just vibes

  wearing a lanyard;

- give developers, security, compliance, and operations the same map for agent

  production readiness.

## Start Here

- [GAAM](gaam/00-introduction-and-recurring-examples.md) — the full model,

  levels, controls, assessment method, security analysis, and standards

  mapping.

- [GAAM Developer Guide](gaam-developer-guide/README.md) — the same model in a shorter,

  developer-friendly form.

- [Examples](https://github.com/tnilabs/governed-agent-autonomy/tree/main/examples) —

  a runnable support-ops implementation that walks the same workload through

  GAAM Levels 1-10 across multiple agent frameworks.

- [Skills](https://github.com/tnilabs/governed-agent-autonomy-skills) —

  companion coding-agent skills for assessing, designing, implementing, and

  reviewing governed agent systems.

## License

Code, executable examples, tests, build scripts, package metadata, schemas, and

machine-readable contracts are licensed under Apache License 2.0. Documentation

and GAAM framework material are licensed under CC BY 4.0. See [LICENSE](LICENSE).