← News

Agents & MCPSources: arXiv · 19 Nov 2026

OpenKAGE framework has agent groups share knowledge for open-ended lifelong learning

A new arXiv paper describes OpenKAGE, a framework in which groups of autonomous agents accumulate and exchange experience, tested in Minecraft.

In plain words

The authors propose a way for teams of AI agents to keep learning from their own actions and from each other, rather than improving alone on fixed tasks. Agents store memories and reusable skills, and share them through a growing knowledge graph. In Minecraft, this helped agent groups explore longer and pass their knowledge to a fresh group running on a different model.

What was announced

OpenKAGE, introduced in an arXiv paper, is an agent-group evolution framework for open-ended lifelong learning, where agents explore open worlds and acquire new capabilities by accumulating and reusing knowledge from their own interactions and the experience of others. The authors position this against existing research that pursues lifelong learning within predefined tasks, curricula, or objectives, with self-improvement largely confined to individual agents.

The framework rests on two mechanisms. The first is a two-layer knowledge representation: a memory layer for situated memories and an experience layer for transferable experience and skills that retains the applicability conditions needed for reuse. The second is a knowledge-sharing mechanism that semantically consolidates textual experience into a continuously evolving shared knowledge graph, while executable skills are shared verbatim. In Minecraft experiments, agent groups sustained exploration and reached advanced technological milestones, and accumulated knowledge transferred to a newly initialized group on a different backbone model, improving its survival and exploration progress.

What it does not show

  • The evidence is the paper's abstract only; it reports the authors' own claims and does not include independent evaluation, full methodology, or quantitative results.

Sources

Source-support review completed 20 Nov 2026. This story was drafted with AI assistance and checked against retained source passages. Automated checking is not independent verification.

Coverage: arXiv: abstract.

Updates and corrections

No corrections so far.

First published 20 Nov 2026

New to the subject? AI from the Ground Up starts from the beginning.