# Agent Authority Protocol

Classify what an AI agent may do, what it can affect, and whether it may delegate.

Type: Authority classification protocol

Availability: Public source package · working paper available

Evidence: Download · Source code · Sample files

## The work

An agent’s technical access does not describe the authority people have given it.

An Authority–Reach–Delegation model, class definitions, risk scorecard, example dossiers, and a working paper.

## Where it stands

This is an open classification proposal. It complements access controls and human governance; it does not enforce permissions at runtime.

## Take a look

Browse the public repository, documentation and examples.

- [Explore the repository](https://github.com/unit27research/agent-authority-protocol)
- [Read the working paper](https://github.com/unit27research/agent-authority-protocol/blob/main/paper/agent-authority-protocol-working-paper.pdf)
- [Inspect the public examples](https://github.com/unit27research/agent-authority-protocol/tree/main/examples)



[Portfolio project page](https://joshuabloodworth.com/projects/agent-authority-protocol.html)