Agent Network is a multi-agent framework designed for orchestrating AI agents using a single command-line interface (CLI). It supports a wide array of large language models, including offerings from Anthropic, OpenAI, MiniMax, DeepSeek, Zhipu GLM, Kimi, InternLM, and OpenRouter, allowing users to switch between them easily. The framework is built with a local-first philosophy, enabling agents to run on a single machine or across multiple servers by configuring the Hub to bind to a public IP. Communication between agents is facilitated by 17 Mesh Communication Protocol (MCP) tools, enabling automatic discovery and task assignment without explicit orchestration logic from the user. A key feature is its integrated web dashboard, built with Next.js 16, which provides an overview of nodes, tasks, messages, and a real-time node topology map. This dashboard also supports human-agent collaboration through a chat interface. Unlike frameworks such as LangGraph, AutoGen, or CrewAI, Agent Network is an npm package with no Python dependencies, emphasizes local deployment over SaaS, offers multi-vendor LLM support, and integrates human interaction directly into the agent workflow.
Agent Network: A Multi-Agent Orchestration Framework for LLMs
Agent Network is a JavaScript-based multi-agent orchestration framework that allows users to run networks of AI agents locally or across servers. It supports various large language models (LLMs) including Claude, GPT, MiniMax, DeepSeek, GLM, Kimi, InternLM, and OpenRouter, and features a web dashboard for management and collaboration.
Why it matters
Agent Network provides a local-first, npm-installable solution for orchestrating multiple AI agents from different LLM providers. Its focus on local deployment, multi-vendor support, and a built-in web dashboard for human-agent collaboration offers a distinct alternative to Python-based, SaaS-oriented, or OpenAI-centric frameworks, potentially lowering the barrier to entry for developers building complex agent systems.