Goose Setup and First Run
1. What installation actually gives you​
Goose offers two primary entry points:
- a desktop app for macOS, Linux, and Windows,
- a CLI for terminal-first workflows.
The current repo README points to this CLI installer:
curl -fsSL https://github.com/aaif-goose/goose/releases/download/stable/download_cli.sh | bash
2. Desktop vs CLI​
| Path | Best when |
|---|---|
| Desktop app | You want the easiest onboarding and a visual setup path |
| CLI | You want terminal workflows, scripting, or server usage |
The good news is that both sit on the same product family, so teams do not have to choose completely different tools for technical and non-technical users.
3. Provider setup is the real first-run step​
The official docs make provider configuration a first-class onboarding topic. Goose supports a long list of backends, including:
- Anthropic
- OpenAI
- Azure
- Bedrock
- Groq
- OpenRouter
- Ollama
- LM Studio
- LiteLLM
That means "installing Goose" is only the first half. The real first run is not complete until you pick and authenticate one provider.
4. The practical first-run flow​
A good first sequence is:
- install the desktop app or CLI,
- configure one provider,
- start with a simple low-risk task,
- add MCP extensions only after the core loop works.
This keeps provider problems separate from extension problems.
5. Why the first run feels broader than a coding CLI​
Because Goose can act as desktop app, CLI, and API, it is best understood as a small local platform. Even if you only want code help today, the initial setup choices can shape later use for automation, research, or embedded workflows.