Introduction
Goose is an open-source, extensible AI agent by Block (the company behind Square and Cash App). It goes beyond code suggestions - Goose can install packages, execute commands, edit files, run tests, and interact with external services via MCP. It is available as both a CLI and a desktop app, and supports fully custom OpenAI-compatible providers via a JSON config file. Goose pairs well with SaladCloud because:- Custom provider support - define any OpenAI-compatible endpoint as a named provider, including custom headers for
Salad-Api-Keyauthentication - MCP integration - connect to GitHub, databases, browsers, and more
- CLI + desktop - works in your terminal or as a standalone app
- Per-hour pricing - no per-token cost during long agentic sessions
Prerequisites
Before getting started, make sure you have:- A SaladCloud account with an active LLM deployment
- Linux, macOS, or Windows (WSL2)
Step-by-Step Setup
Step 1: Deploy an LLM Recipe on SaladCloud
First, deploy an OpenAI-compatible LLM server on SaladCloud.- Go to the SaladCloud portal and create an account if you do not already have one.
- Create an organization or choose an existing one, then click “Deploy a container group”.
- Select an LLM recipe. The Qwen3.5-35B-A3B (llama.cpp) recipe is a strong choice for general-purpose agentic tasks. On the recipe page, provide a name and deploy - the rest is preconfigured with recommended settings.
- Once deployed, your endpoint will be live and serving an OpenAI-compatible API.
- qwen3.5-35B-A3B - A powerful Mixture of Experts model optimized for instruction-following tasks.
- qwen3.5-9b-llama-cpp - Optimized for Qwen3.5 9B model.
- llama.cpp - Supports GGUF models
- sglang - High-performance inference
- vllm - Popular LLM serving framework
- ollama - Simple model management
- tgi - Hugging Face Text Generation Inference server
- API endpoint URL (e.g.,
https://your-endpoint.salad.cloud) - API key (from your SaladCloud organization settings, if you enabled authentication)
Step 2: Install Goose
For Desktop App instructions scroll to the bottom of this page. Linux / macOS / WSL2:Step 3: Configure a Custom Provider
The first time you run Goose it launches an onboarding wizard. When asked “What would you like to configure?”, follow these steps:- Select Custom Provider → Add a custom provider → OpenAI Compatible
- Enter a name for the provider (e.g.,
saladcloud) - Enter the API URL:
https://your-endpoint.salad.cloud/v1/chat/completions - When prompted for an API key, enter any non-empty string (e.g.,
dummy), or indicate that authentication is not required (if it is, you will add the real key in the custom headers step) - Enter the available models separated by commas (e.g.,
qwen3.5-35b-a3b) - Set your streaming and other preferences
- When asked “Does this provider require custom headers?”, select Yes if your deployment requires
authentication and add:
- Header name:
Salad-Api-Key - Header value: your SaladCloud API key
- Header name:
goose configure and under Configure Providers select your new custom provider.
Alternatively, you can create or edit the provider config file directly at
~/.config/goose/custom_providers/saladcloud.json:
Step 4: Start Goose and Test the Connection
Start a Goose session:“Create a hello world Python script and run it.”If Goose creates and executes the file, your setup is complete.
Goose Desktop App
Goose also has a desktop app with the same capabilities as the CLI. See the Quick Start instructions for download links for your OS. Download and run the app, then on the startup page:- Click Connect to a Provider
- Click Add Custom Provider
- Click Configure Manually and fill in:
- Provider Type: OpenAI Compatible
- Display Name: a name for the provider (e.g.,
SaladCloud) - API URL:
https://your-endpoint.salad.cloud/v1/ - API Base Path: leave blank
- Available Models: your model name (e.g.,
qwen3.5-35b-a3b) - Custom Headers: add
Salad-Api-Keywith your SaladCloud API key as the value (only if your deployment has authentication enabled)
- Click Create Provider and start using Goose with your SaladCloud deployment
Model Recommendations
- Qwen 3.5-35B-A3B: Best for complex agentic tasks - install, execute, edit, and test workflows
- Qwen 3.5-9B: Suitable for lighter tasks and faster response times