Run a local AI agent that handles code and business automation
You can install an AI on your computer that performs actual tasks, from file editing to running tests, rather than just suggesting code. goose is a general-purpose AI agent available as a desktop app, terminal command, and API. It handles various tasks, including research, writing, and data analysis, with a single tool.
A native app that runs in your local environment
goose runs in the user's local environment without relying solely on the cloud. Written in Rust for performance and portability, it is available as a native desktop app on macOS, Linux, and Windows. It also supports a CLI method for entering commands in the terminal and an API method for embedding in other applications. This allows it to handle complex tasks such as workflow automation and data analysis, in addition to code writing.
Connecting various AI services and extensions
It integrates with more than 15 AI providers, including Anthropic, OpenAI, Google, and Ollama. You can enter API keys directly or use existing Claude, ChatGPT, or Gemini subscriptions via ACP. Additionally, you can connect more than 70 extensions through the Model Context Protocol (MCP), an open standard, allowing you to freely expand the agent's capabilities.
Installation and getting started
To install the desktop app, download and install the installer for macOS, Linux, or Windows. To use it in the terminal, run the provided script to install the CLI. After installation, you can perform your first task by referring to the quick start guide and tutorials. You can also build your own goose distribution, including pre-configured providers and extensions.
Things to check before use
goose is maintained as part of the Agentic AI Foundation (AAIF) under the Linux Foundation. Since it runs locally, you must verify your system environment and access permissions for the AI services you intend to connect. You may need a subscription or API key for specific AI providers, and if you use extensions, you must separately check their requirements. It is recommended to check compatibility in your current environment by referring to the diagnostics and reporting section and the list of known issues in the official documentation.