Problem solved

It solves the complex problem of managing multiple AI‑powered coding assistants simultaneously by providing isolated, persistent tmux sessions with git worktree isolation. This eliminates workspace collisions, prevents the loss of agent context when disconnecting, and gives developers a single TUI to supervise and switch between various autonomous code‑generation pipelines.

Use cases

• Run parallel code‑generation tasks with different AI models
• Compare outputs of multiple AI coding assistants side‑by‑side
• Manage long‑running agent sessions with full session persistence
• Create isolated workspaces per task using git worktrees
• Define custom CLI agents to integrate in‑house or third‑party tools
• Orchestrate remote AI coding sessions accessible over SSH via tmux

Audience

• Developers
• ML Engineers
• Researchers

Pros

• Session persistence across disconnections and restarts
• Isolated workspaces via git worktrees prevent conflicts
• Supports any CLI‑based coding agent
• Multi‑agent parallel execution with central monitoring
• Lightweight, terminal‑based interface with no heavy GUI

Cons

• Requires familiarity with tmux and git worktree concepts
• Setup and configuration may be complex for beginners
• Limited to terminal environments – no graphical interface
• Relatively new project with evolving documentation
• Does not provide its own AI models – relies on external CLIs