goose

About

goose is an open-source AI agent that its site describes as "your open source AI agent". It is a general-purpose agent for research, writing, automation and data analysis as well as coding, with a native desktop app for macOS, Linux and Windows, a CLI and an API. The site says it is built in Rust. The site says it is part of the Agentic AI Foundation at the Linux Foundation. It is licensed Apache 2.0, and the site makes no pricing statement beyond the model costs of the providers you pick. It belongs with the agent tools on our best AI coding assistant page and our AI coding assistants collection. What goose does • Desktop, CLI and API: a native desktop app for macOS, Linux and Windows, a full CLI for terminal use, and an API. • Built in Rust: the GitHub description says it is built in Rust for performance and portability, and describes goose as an extensible agent that goes beyond code suggestions: install, execute, edit and test with any LLM. • Providers: the site says it works with 15+ providers, naming Anthropic, OpenAI, Google, Ollama, OpenRouter, Azure and Bedrock. You can use API keys or existing Claude, ChatGPT or Gemini subscriptions through ACP. • MCP extensions: the site cites Model Context Protocol support with 70+ extensions, and MCP Apps that render interactive interfaces inside the desktop app. • ACP: Agent Client Protocol support lets you connect from editors such as Zed, JetBrains and VS Code. • Recipes: workflows captured as portable YAML configs that can run in CI. • Subagents: parallel task handling that keeps the main conversation clean. • goose Development Kit: for building agents across native apps, servers and browsers, with resumable saved state. • Security controls: the site lists prompt injection detection, tool permission controls, sandbox mode and an adversary reviewer. • Install: a download link for the desktop app, or a curl script for the CLI from the project's GitHub releases. Pricing (as of November 2026) • goose: no listed price. The site shows it as open source under Apache 2.0 and makes no explicit pricing statement. • Model costs: these depend on the provider you choose, whether through API keys or existing subscriptions, or Ollama for local models. Who it suits • Developers who want an open-source agent they can run from a desktop app or a terminal. • People who want to switch between model providers or use subscriptions they already pay for. • Teams who want repeatable workflows written as recipes and run in CI. • Users who want an agent for tasks beyond writing code. Limitations to know • goose is a general-purpose agent, so it is broader than a coding-only tool. • It is not a no-code app builder, and it expects some technical comfort. • Model costs are yours to manage. • The pages we read give no release dates or version numbers. FAQs What is goose? An open-source, extensible AI agent with a desktop app, CLI and API, hosted under the Agentic AI Foundation at the Linux Foundation, per its site. Is goose free? The site states no price and the license is Apache 2.0. You pay your model provider for usage. Which platforms does it support? The site lists a native desktop app for macOS, Linux and Windows, plus a CLI. Which models can I use? The site says 15+ providers, including Anthropic, OpenAI, Google, Ollama, OpenRouter, Azure and Bedrock. Does it support MCP? Yes, the site cites MCP with 70+ extensions. Can I use it in my editor? The site says ACP lets you connect from editors such as Zed, JetBrains and VS Code. What we could not verify • The latest release number and date, because the GitHub Releases section was empty on the page we read. • When goose moved to the Agentic AI Foundation, which the site states without a date. • The "15+ providers" and "70+ extensions" counts beyond the site's own statements. • How the security features behave in practice. How this listing was put together This listing is based on goose's own documentation site and the README of its GitHub repository. Facts verified on 14 November 2026. Sources • goose homepage • goose on GitHub