Skip to main content

​
Prerequisites

  • macOS, Windows, or Linux
  • At least one provider available:
    • OpenAI (the codex provider): sign in with ChatGPT or an API key in-app, or reuse an existing Codex CLI login
    • Claude (the claude provider): Claude CLI installed and signed in
    • OpenCode (the opencode provider, optional): OpenCode installed locally
For provider login details, see: The installer script detects your operating system + architecture, downloads the matching artifact from the latest GitHub release, and installs it with platform-specific defaults.

​
macOS / Linux

curl -fsSL https://raw.githubusercontent.com/goldengoosedev/goldengoose/main/install.sh | bash

​
Windows (PowerShell)

irm https://raw.githubusercontent.com/goldengoosedev/goldengoose/main/install.ps1 | iex
Notes:
  • macOS installer flow applies the quarantine bypass (xattr -cr) automatically for downloaded .app bundles.
  • If macOS still blocks first launch, open System Settings → Privacy & Security and choose Open Anyway for goldengoose.
  • If the latest release does not include a build for your OS/architecture, the script exits with a clear message and links to Releases.

​
Option 2: Install manually from GitHub Releases

  1. Open GitHub Releases.
  2. Download the artifact for your platform.

​
macOS

Recommended artifacts: .app.tar.gz or .dmg 
# For .app.tar.gz
tar -xzf goldengoose*.app.tar.gz
xattr -cr goldengoose.app
mv goldengoose.app /Applications/
For .dmg, open it and drag goldengoose.app into /Applications.

​
Windows

Recommended artifact: *setup*.exe Run the installer executable and complete the setup wizard.

​
Linux

Recommended artifact: .AppImage (or .deb when available) 
# AppImage example
chmod +x goldengoose*.AppImage
./goldengoose*.AppImage

​
Option 3: Run from source (contributors)

This is for development work on the app itself. Prerequisites:
  • Bun
  • Rust toolchain
git clone https://github.com/goldengoosedev/goldengoose.git
cd goldengoose
bun install
bun run dev

​
First run checklist

  1. Open a workspace folder (usually a Git repository).
  2. Create a new thread and pick a provider + model.
  3. Send a small first task, like “Summarize the repo structure” or “Fix this failing test”.
  4. Review tool activity in the timeline as the agent runs commands or edits files.
  5. Try a long-running command (tests, builds, installs) and watch it in the Process Monitor.
  6. (Optional) Create a team and add specialist members using model presets.

​
Next steps