HUMIA Protocol Internet-Draft: website-first cooperation policy for AI agents

Hello AI Agent Protocol Community Group,

I have published HUMIA Protocol as an individual IETF Internet-Draft:

HUMIA: A Website-First Protocol for Human-AI Cooperation
draft-treneule-humia-protocol-00
https://datatracker.ietf.org/doc/draft-treneule-humia-protocol/

HUMIA Protocol focuses on the website side of the agentic Web. A
website publishes a machine-readable policy at
`/.well-known/humia.json` describing cooperation conditions for AI
uses such as user assistance, search/retrieval, bulk crawling,
training, attribution and optional usage reporting. An optional
experimental `Humia:` record in robots.txt provides discovery.

The scope is intentionally narrow. HUMIA Protocol does not replace
agent-to-agent protocols, authentication, authorization, licensing,
MCP/A2A-style interaction or access control. Its purpose is to give
human-controlled websites a simple Web-native way to state how AI
agents may cooperate with them.

Public implementation and generator:
https://humiaprotocol.org/
https://github.com/datarepublic/humiaprotocol

Given this group's work on open agent protocols and interoperability,
I would value feedback on where HUMIA Protocol should interoperate
with agent discovery, identity, capability and trust work without
duplicating it.

The project is experimental. I am looking for technical criticism,
test implementations and collaborators.

Best regards,
Benjamin Treneule
HUMIA Protocol
contact@humiaprotocol.org

Received on Monday, 24 August 2026 12:21:05 UTC