goose: an open source AI agent that works on your computer
goose is an open source AI agent that runs on your own machine, as a desktop app, a command-line tool or an API, to write code, automate tasks, do research or analyze data. Created by Block, it suits developers and power users alike who want a general-purpose agent with the model of their choice.
Data as of October 22, 2026
Key figures
| Developer | Block (créateur), projet de l'Agentic AI Foundation (Linux Foundation) |
|---|---|
| Type | Low-code |
| Languages | Rust, TypeScript |
| Open source | Yes |
| License | Apache-2.0 |
| First release | 2025 |
| Suggested level | Beginner |
| Use cases | coding, automation, data |
| Self-hostable | Yes |
| Business model | Free |
| Cloud offering | No |
| Multi-agent | Yes |
| MCP support | Yes |
| Bring your own model | Yes |
| Official site | Official site · Documentation · Repository |
GitHub
| Repository | aaif-goose/goose |
|---|---|
| GitHub stars | 55,525 |
| Forks | 6,421 |
| Latest code activity | October 22, 2026 |
| Latest release | v1.55.0 (October 16, 2026) |
GitHub activity checked on October 22, 2026.
A desktop agent, not just a library
Unlike most frameworks in this section, goose doesn’t require you to write code. You install a desktop app (macOS, Linux or Windows) or a command-line tool, pick a model provider, then chat with an AI agent that can read and write files, run commands and use extensions. Written in Rust, goose also exposes an API so it can be embedded elsewhere, and its maintainers document how to build custom distributions.
The project’s history explains its ambitions: Block released it in early 2025, then contributed it in December 2025 to the Linux Foundation’s Agentic AI Foundation, created at the same time to also host the MCP protocol and AGENTS.md.
What it can do
All of goose’s capabilities come through extensions built on the Model Context Protocol: browser automation, access to online services, developer tools. For repetitive jobs, “recipes” save a reusable workflow, and subagents let you split work across several instances running in parallel (orchestration). Uses range from building a small app to migrating code or researching a topic.
Getting started
Download the app from the official documentation or install the CLI, then run the configuration step to choose a provider and enter your key. The quickstart has you ask goose to build a small browser game to get familiar with the tool. After that, enable only the extensions you actually need.
Strengths and limitations
Strengths: free, cross-platform, usable without programming, model-agnostic, highly extensible, neutral governance.
Limitations: an agent acting on your computer carries real risks (deleted files, prompt injection from a web page); the can I trust an agent with this? tool helps you set boundaries. Results depend heavily on the model you choose, and API costs can climb during long sessions.
Alternatives
OpenHands targets development teams, Aider stays focused on code in the terminal, and the Claude Agent SDK is for building your own agent.
Frameworks of the same type
- Botpress · chatbot, RAG, automation
- Microsoft Copilot Studio · chatbot, workflows, automation
- Dify · chatbot, RAG, workflows
- n8n · automation, workflows, RAG
- Relevance AI · multi-agent, automation, workflows
- Langflow · RAG, multi-agent, workflows
Frequently asked questions
Is goose free?
Yes, goose is free and open source under the Apache 2.0 license. You do need to supply a model: an API key from a provider, a local model through Ollama, or, according to the documentation, an existing Claude, ChatGPT or Gemini subscription via the ACP protocol.
Does Block still own goose?
Block created it and released it in early 2025, then handed it in December 2025 to the Agentic AI Foundation, part of the Linux Foundation, alongside the MCP protocol. It has followed open governance since then.
Is goose only for coding?
No. It is popular for code, but its creators describe it as a general-purpose agent: research, writing, automation or data analysis, depending on which extensions you enable.
What is a goose recipe?
A reusable workflow described in a file: instructions, parameters and required extensions. You can share a recipe with your team or combine several, which makes it an easy way to automate a recurring task.
Data sources
- github.com
- goose-docs.ai
- goose-docs.ai
- linuxfoundation.org
- GitHub — Repository metadata: GitHub
Data as of October 22, 2026
Report an error
Does a figure, a date or a fact look wrong? Tell us: every report is checked.