Skip to content

Quick start

  • macOS 12+ (Apple Silicon or Intel), Windows 10+, or Linux with a graphical environment
  • Git installed and configured
  • At least one AI agent CLI installed (e.g. Claude Code, Codex CLI, or Pi)
  • A GitHub personal access token with repo and PR permissions

Download the latest release for your platform from GitHub Releases.

On first launch, Control Center opens the onboarding screen.

Provide your GitHub personal access token. Control Center validates it and stores it securely in your system keychain (macOS Keychain, Windows Credential Store, or Linux libsecret).

If you have the gh CLI installed and authenticated, Control Center detects it automatically.

A workspace is your top-level container — it groups agents, repos, channels, and memory. Give it a name and optionally a logo.

The first workspace also seeds a CEO agent that can hire and coordinate other agents.

Add a Git repository to your workspace. Control Center inspects the repo and detects its GitHub remote. The repo becomes available for worktree isolation — each agent conversation gets its own copy-on-write branch.

Open a channel with an agent (or mention @agent in any channel). Type your instructions. The agent:

  1. Gets its own isolated worktree
  2. Receives a prompt assembled from its role, persona, skills, and conversation context
  3. Executes in a sandbox with capability-gated access to your repos and APIs
  4. Streams its thinking and output back to the channel in real time

When the agent finishes, it opens a pull request. Open the PR in Control Center’s diff viewer, review the changes, and merge — all without leaving the app.