The problem
Modern AI coding assistants are powerful, but they work on one task at a time. You describe a feature, wait for the agent to finish, review the result, then move to the next task. On a busy day, you might have ten independent tasks — bug fixes, refactors, new features — and the sequential workflow becomes a bottleneck.
What Shellarium does
Shellarium is an AI Agent Orchestrator for macOS. It creates isolated git worktrees for each task so multiple agents can write code at the same time without stepping on each other.
- Parallel execution — run dozens of AI agents simultaneously. Each agent gets its own branch and worktree. No merge conflicts between agents.
- Multi-provider support — use Claude, Codex, or any ACP-compatible agent. You are not locked into a single AI provider.
- Built-in code review — diff viewer with syntax highlighting, squash merge and merge commit options, and GitHub integration. Review and merge without leaving the app.
- Permission control — approve or deny every tool call, or enable autonomous "brave mode" when you trust the agent.
- Real-time progress — watch each agent work live in a unified dashboard.
How it works
- Create a workspace — pick a git repository and describe the task. Shellarium creates an isolated worktree and branch automatically.
- Agent writes code — the AI agent works in its sandbox while you create more workspaces or review completed ones.
- Review & merge — inspect the diff, request revisions, or merge directly. Squash merge keeps your git history clean.
Why try the beta now
- Free — no payment required during the beta period.
- Shape the product — beta feedback directly influences the roadmap. Join the Discord community to share ideas and report issues.
- Immediate productivity gain — if you already use AI coding assistants, parallelizing them is a multiplier, not an incremental improvement.
Requirements
- macOS 13 or higher
- GitHub account