Superset coordinates multiple CLI-based coding agents in parallel by isolating each in its own git worktree with dedicated terminal and environment. It provides desktop GUI, CLI, and SDK surfaces for monitoring, reviewing, and managing agent-driven code changes.
Project overview
Isolating concurrent coding agents in separate git worktrees addresses task interference and context switching directly, which are common operational friction points when running multiple agents simultaneously.
Project type
AI Agent · AI Coding
Use cases
Coding & Development
Deployment
Refer to project documentation
License
License pending
Best for
Developers who use CLI-based coding agents like Claude Code or Codex and need to run multiple concurrent tasks in isolated environments.
Key capabilities
Run 10+ coding agents at once, each isolated in its own git worktree with its own branch, terminal, and environment.
Track every agent from the sidebar, with working indicators, completion chimes, and dock badges.
Inspect, comment on, and edit agent changes without leaving the app, then commit and push.
Preview running dev servers in a browser pane, with ports detected per workspace.
Run agent sessions on a schedule, such as triaging issues overnight or drafting changelogs.
Connect another machine and reach its workspaces from anywhere via desktop app, CLI, or phone.
A single `superset` binary to manage workspaces, agents, terminals, and hosts from any shell.
Drive Superset programmatically via `@superset_sh/sdk` from Node, Bun, or Deno.
Limitations and risks
The system requires an external CLI-based coding agent, such as Claude Code or Codex, to function.
The documented setup path and first success method apply to macOS via the desktop application.
Because external CLI-based coding agents are required, local orchestration does not provide a local-only data boundary for inference.
Getting started
Download the macOS release of Superset, install the application, and launch it to begin managing workspaces.
Alternatives and comparisons
An open-source agent harness that orchestrates sub-agents, memory, and sandboxes to handle long-horizon tasks.
An autonomous work engine that plans and executes multi-step work using real tools across a computer.
A platform to build AI agents, internal apps, and automations that handle operational workflows.