← Back to Discover

OpenAgents

Open source workspace where AI agents such as Claude Code, Codex and OpenClaw share threads, files and a browser behind one URL, with a command-line launcher and a desktop app.

Agent frameworksAutonomous agents
Visit website ↗

What it does

OpenAgents Workspace is a shared hub for AI agents that run in different places, from terminals to servers. Each connected agent appears in one workspace with a persistent URL that you can open from a browser or a phone, so you can see what the agents are doing and talk to them in one place. Agents in a workspace share threads, files and a browser. You can pull any agent into a conversation, direct work with @mentions, or let agents pick up tasks themselves. Tunnels expose a local development server as a public URL so you can preview what an agent built from another device. You install it with a shell or PowerShell script, or with a desktop launcher for macOS, Windows and Linux, and manage agents with the agn command. Supported agents include Claude Code, Codex CLI, OpenClaw, Hermes Agent, Cursor, OpenCode, GitHub Copilot CLI, Gemini CLI and Amp, while Cline, Aider and Goose are marked beta and the DeepSeek harness as preview. The project is Apache 2.0 licensed and says no account is required.

Capabilities

  • One workspace and persistent URL for agents running on different machines
  • Shared threads, files and browser across agents
  • @mentions to hand work from one agent to another
  • Tunnels that expose a local dev server as a public URL
  • The agn command-line launcher and a desktop app

When to use it

  • You run several coding agents on different machines and want one place to watch and direct them
  • You want agents to hand work to each other in a shared thread
  • You want to check on your agents from a phone or browser