Arne and Raban have launched Emdash, an open-source Agentic Development Environment that lets developers run multiple coding agents in parallel. The desktop app supports 21 coding agent CLIs including Claude Code, Codex, Gemini, and more, each isolated in its own git worktree. Available for macOS, Windows, and Linux with instant task startup times of 500-1000ms.

Built by developers frustrated with messy multi-agent workflows, Emdash centralizes the terminal and enables parallel development across local and remote machines via SSH. The team solved slow worktree creation by maintaining a reserve of pre-created worktrees, dramatically improving performance. The app is MIT-licensed and provider-agnostic by design, allowing teams to freely switch between different AI coding assistants based on task requirements.

Beyond orchestration, Emdash integrates the full development loop with built-in diff review, commit management, PR creation, and CI/CD check monitoring. Users can pass tickets directly from Linear, GitHub, or Jira to agents and use lifecycle scripts for testing. The team is actively seeking feedback on multi-agent workflows and welcomes contributions through their GitHub repository. Emdash represents a significant step toward streamlining AI-assisted development by addressing the fragmentation that currently plagues multi-agent coding environments.