when we started running claude code and codex on the same repo, the thing that kept biting us was two agents quietly editing the same file and one of them finding out at merge time. anthropic s own docs basically tell you to split file ownership by hand for agent teams, which works fine until a task touches the shared types or the router and you re back to babysitting
we went the other way and let each agent see what the others are changing so they can work it out between themselves. in our benchmarks they caught 98.6% of overlapping work before touching the file, compared to 55% without agentmash, but i m pretty sure there are setups where splitting by hand is still the saner call
Agentmash is an open-source multiplayer AI for teams' coding agents, a shared context layer that tells every agent which files a teammate's agent is changing, so the agents talk to each other and work on their own without overwriting anyone's code. You can switch between tasks running on different models, @mention a teammate, your own agent or a teammate's agent like in Slack, and keep skills, chats and automations in sync. Connect Claude Code, Codex, Antigravity or your own open-source models
