agent-language

What is agent-language and what is it for?

agent-language is an AD4M Language — a plug-in module that an AD4M instance loads — that stores and looks up the public profile data attached to a decentralized identifier, or DID. It exists so that agents in an AD4M network have one shared, agreed place to publish who they are, instead of each app keeping a private copy.

agent-language is a small module built for AD4M, a host that runs plug-in modules it calls Languages. It pairs a TypeScript layer that AD4M talks to with a Holochain application — a coordinator zome and an integrity zome — so the data behind it lives on a shared distributed hash table rather than on one operator's server. Its repository describes it as an AD4M language for storing Agent data associated with DIDs.

The idea in human terms is a directory of people that nobody owns. In AD4M a person is identified by a DID, and this language is what turns a DID into something readable: a shared profile made of statements its owner publishes, plus an optional address at which that person can receive direct messages. When one agent wants to know who another agent is, this is the layer that answers.

AD4M's own documentation lists the Agent Language among its bootstrap languages — the ones a fresh instance needs before anything else can work — and describes it as resolving DIDs to agent information. Read the AD4M language documentation at docs.ad4m.dev/languages and the repository's README for the project's own wording. As a historical note with a date: the project's catalog census records no activity after 27 April 2023 and marks its liveness stalled, so the sources above describe a design from that period rather than a currently maintained install.

From 3 public sources, checked 23 September 2026.