WG - Agent-to-Agent LLM Communication Standardization

Hello everyone,

I'm Samuele Marro, currently doing a PhD on agent communication at the University of Oxford with Microsoft. I've been informally following the activities of the group for some time and officially joined two weeks ago.

I've been talking to members of the major companies and organizations working on LLM-based agent frameworks — LangChain, Camel AI, ANP, Wild-Card, Eigent, Stanford, Oxford, Agora, Alan Turing Institute, and GovAI. They've accepted my proposal to set up an early, lean, practical-focused working group on standardizing certain aspects of agent-to-agent communication for LLM-based agents, such as format selection and capability / service description for LLM-based agents.
The main strength of the committee is that it includes the creators of all the major agent protocols (Agent Protocol, ANP, Agora, and agents.json), as well as some of the companies most active in terms of web agents. This means that we can get early feedback on how well different approaches work and practical insights from people using agents in business contexts.

Given the expertise within the Web Agents CG, we would like to invite you all to take part in the group. The activities of the group are distinct from those of the Web Agents CG, since our main focus is on LLM-specific standards that are not tightly coupled to the Web Architecture. That said, we would like to build them in such a way that they can be used in  W3C standards later down the line.

If you're interested in participating, please reach out to me or Jesse Wright (jesse.wright@cs.ox.ac.uk) via email.

Looking forward to working more closely with you all.

Best,

Samuele Marro
Department Of Engineering Sciences
University of Oxford
agoraprotocol.org<http://agoraprotocol.org>