More
The whole thing is 13 MB.
Loom Conductor is a free, open-source desktop app for macOS, Linux, and Windows. One small download, one CLI sign-in, and a fleet of six Claude Code sessions is yours.
All platforms
Pick your build.
macOS
A native dmg around 13 MB, with Conductor keys held in the keychain. Details on the macOS page.
Linux
AppImage, deb, and rpm builds with keyring-backed secrets. Notes on the Linux page.
Windows
A native installer with WSL available per terminal tab. Specifics on the Windows page.
The one prerequisite
The fleet runs on the CLI you already have.
Loom drives six real Claude Code CLI sessions on your existing Claude login, so the fleet needs no API key at all. The only key you ever bring is for the Conductor, and that is any provider you like or a local model.
# install the Claude Code CLI (macOS, Linux, WSL)
$ curl -fsSL https://claude.ai/install.sh | bash
# Windows PowerShell
PS> irm https://claude.ai/install.ps1 | iex
# sign in once; all six sessions reuse this login
$ claudeWhat is inside the 13 MB
No Electron, no bundled browser. Loom is a Tauri 2 shell with a Rust core and a React 19 front end, and that is how the installer stays this small while shipping a complete workspace:
- A WebGL terminal built on xterm.js, with native PTYs, tabs, splits, search, and true color.
- A CodeMirror 6 editor with inline AI autocomplete and AI diffs you accept or reject hunk by hunk.
- A source control panel with a full git graph, an auto-detected web preview of dev servers, markdown preview, and a file explorer.
- The Conductor itself: planning, dispatch, permission handling, stall and rate-limit recovery, and verification across the mission DAG.
- A theme engine with presets, custom themes, and background images, plus the Design tab, a website builder that designs before it builds.
What you need
Three things. A machine running macOS, Linux, or Windows. The Claude Code CLI signed in to your Claude account, which is what the fleet runs on. And a model for the Conductor: an API key from OpenAI, Anthropic, Google, xAI, Groq, Cerebras, DeepSeek, Mistral, OpenRouter, or any OpenAI-compatible endpoint, or a local model through LM Studio, MLX, or Ollama.
There is no account beyond your Claude login, no telemetry, and no charge for the app itself; you pay your model providers directly. Every release is published with full source under Apache-2.0 at GitHub releases. Once installed, the getting started guide takes you from launch to a running fleet in about a minute, and the changelog shows what each version added.
Hand it the work.
Walk away.
macOS, Linux, and Windows. Around 13 MB. Free and open source.