ClawTeam-OpenClaw

What ClawTeam-OpenClaw does

ClawTeam-OpenClaw is a fork of HKUDS/ClawTeam that makes OpenClaw the default agent and adds per-agent session isolation, automatic exec approval configuration and hardened spawn backends. You set a goal and a leader agent calls clawteam spawn to create workers, each with its own git worktree, spawn backend session and identity. Workers check inboxes, update tasks and report back through clawteam CLI commands that are injected into their prompts.

It works with OpenClaw, Claude Code, Codex, Hermes Agent, nanobot, Cursor and any CLI agent. Progress can be watched in a tiled tmux view or through a web board started with clawteam board serve. Linux and macOS use a tmux-first workflow, and Windows 10 and 11 fall back to a subprocess backend, with WSL available for the original tmux flow. The README is available in eleven languages.

Key features

• Leader agent spawns workers with clawteam spawn, each in its own git worktree
• Inbox and task commands injected into each worker's prompt
• Web board with clawteam board serve, or a tiled tmux view with board attach
• Works with OpenClaw, Claude Code, Codex, Hermes Agent, nanobot, Cursor and any CLI agent
• OpenClaw defaults: per-agent session isolation and exec approval auto-configuration
• Windows 10/11 support through a subprocess backend

When to use it

• Split a build task such as an auth module across several coding agents
• Review the progress of a swarm on a local web board
• Run a mixed team of OpenClaw and Hermes Agent workers

Who it is for: Developers who want coding agents to organise themselves into teams, mainly on OpenClaw.

How it fits with Hermes Agent

Hermes Agent is listed among the agents ClawTeam works with, but OpenClaw is the default agent in this fork.

Requirements: tmux for clawteam board attach on Linux, macOS or WSL; Windows 10/11 uses a subprocess backend instead

Note: This repository is a fork of HKUDS/ClawTeam adapted for OpenClaw.

FAQ