Orchestrate AI coding agents in parallel on your machine.
Local-first. No cloud. No compromises.
SQLite database, git worktrees, and agent processes — all on your machine. Your code never leaves your environment.
Run as many AI agents as your machine can handle — simultaneously. Each in its own isolated git worktree.
Every work item gets its own branch and worktree. Agents can't interfere with each other or your working directory.
Approve, reject with feedback, or refine. Rejected items carry full history context into the next attempt.
Describe what you want in plain English. The orchestrator breaks it into dependency-aware work items automatically.
Watch every agent work in real-time. Streaming terminal output with tool-call visualization powered by xterm.js.
Tell Mozzie what you want to build in natural language. It uses an LLM orchestrator to decompose your request into discrete, dependency-aware work items.
Mozzie spawns AI agents in parallel — each in its own git worktree and branch. Claude Code, Gemini CLI, Codex, or your own custom scripts.
Approve completed work or reject with feedback. The context carries forward — agents learn from each iteration. Merge when you're satisfied.