Goose

What it does

Goose is a general-purpose AI agent that runs entirely on your local machine. It can write, debug, and deploy code autonomously, but its capabilities extend beyond coding to research, writing, automation, and data analysis. Goose offers a desktop app (macOS, Linux, Windows), a CLI, and an API, all built in Rust for performance. It connects to 70+ extensions via the Model Context Protocol (MCP), including databases, APIs, browsers, GitHub, and Google Drive. Users can also capture workflows as portable YAML recipes, spawn independent subagents for parallel tasks, and render interactive UIs inside the desktop app via MCP Apps. Security features include prompt injection detection, tool permission controls, sandbox mode, and an adversary reviewer.

Who it is for

Goose is designed for developers and power users who want a free, local AI agent without subscription fees, cloud dependency, or rate limits. It appeals to those frustrated by the pricing and usage caps of commercial tools like Claude Code (which costs $20–$200/month). The open-source nature and community governance (via the Agentic AI Foundation at the Linux Foundation) also attract users who prioritize data privacy, vendor neutrality, and long-term project sustainability. The tool is suitable for individual developers, teams sharing recipes, and organizations embedding AI agents via the API.

Why it matters

Goose addresses a growing pain point in the AI industry: the high cost and restrictive rate limits of commercial coding agents. By providing a free, local alternative with comparable functionality, it democratizes access to powerful AI-assisted development. The project has gained significant traction, with over 26,100 GitHub stars, 362 contributors, and 102 releases as of January 2026. Its support for 15+ LLM providers (including Anthropic, OpenAI, Google, and local models via Ollama) and adherence to open standards (MCP, ACP) ensure flexibility and interoperability. The move to the Agentic AI Foundation guarantees the project remains community-governed and open-source.

Launch signal

Goose was originally developed by Block (formerly Square) and has since moved to the Agentic AI Foundation (AAIF) under the Linux Foundation. The project has seen rapid adoption, with the latest version (1.20.1) shipped on January 19, 2026. It has been featured in VentureBeat as a free alternative to Claude Code, highlighting its ability to work offline and keep data local. The community is active on Discord and GitHub, with a growing ecosystem of extensions and recipes.

Brand and naming

The name "Goose" is short, memorable, and approachable, evoking a helpful but slightly mischievous assistant. It positions the tool as a friendly, non-corporate alternative to more formal names like "Claude Code" or "GitHub Copilot." The branding emphasizes openness, community, and local control, with a logo featuring a stylized goose. The project's tagline "Your AI agent, your machine" reinforces the core value proposition of privacy and ownership.