The inclusionAI team under Ant Group has officially open-sourced its multi-agent collaboration infrastructure—Avernet V0.1 version. As the application scenarios of large models continue to expand, how to enable multiple agents to collaborate efficiently and achieve automated processing of complex tasks has become a core challenge in the current technology field.

Avernet was developed to address this issue. Instead of replacing the reasoning capabilities of agents themselves, the project focuses on building a complete "collaboration layer" infrastructure. It standardizes key processes in multi-agent collaboration, including registration, discovery, invitation, consensus, and traceable execution, offering an out-of-the-box solution for development teams exploring multi-agent applications.

image.png

In terms of collaboration modes, Avernet supports various flexible interaction methods, including free chatting and leader-follower modes. By establishing groups, sessions, and shared context mechanisms, the system effectively promotes consensus among participants. Additionally, the infrastructure introduces a collaboration feedback loop, enabling the system to automatically evolve through observation, evaluation, reuse, and optimization, ensuring continuous improvement in collaboration efficiency.

In terms of ecological compatibility, Avernet demonstrates strong inclusiveness. It not only supports mainstream protocols such as OpenClaw but also is compatible with custom agents, third-party agent engines, and existing bot platforms, allowing seamless integration into heterogeneous ecosystems. To help developers get started quickly, the project currently provides two deployment options: Docker images and local deployment, significantly lowering the technical barrier.

Industry analysts believe that although Avernet is still in its early V0.1 stage, it precisely addresses the pain points of "difficulty in discovery and disorder in collaboration" in current multi-agent cooperation. By infrastructure-izing the collaboration logic, it provides a clear and efficient technical path for the industry to explore the coordinated development of agents.

Project Address: https://github.com/inclusionAI/Avernet