Automate multi-step development workflows like running tests, deploying code, and managing repositories without manual intervention.
Research and analyze data by having the agent browse the web, download files, and summarize findings into documents.
Set up development environments by having the agent install dependencies, configure tools, and scaffold projects from scratch.
Automate repetitive file management and data processing tasks across your local machine or connected systems.
Requires at least one AI provider API key (Anthropic, OpenAI, Google, or local Ollama) and Rust toolchain to build.
Goose is an open-source, general-purpose AI agent that runs locally on your machine and can do far more than just suggest code. Unlike typical AI coding assistants that sit in your editor and offer completions, goose can autonomously install software, execute shell commands, edit files, run tests, browse the web, conduct research, and automate multi-step workflows. You give it a goal and it figures out the sequence of actions to complete it. The core concept is that goose acts as an agent, it has tools it can call, memory of what it has done, and the ability to loop through tasks until a goal is satisfied. It is not limited to programming: you can use it to write documents, analyze data, manage files, or automate any task your computer can perform. Goose connects to over 15 AI providers, including Anthropic, OpenAI, Google, Ollama, OpenRouter, Azure, and AWS Bedrock, so you can use it with API keys or your existing Claude, ChatGPT, or Gemini subscriptions. It supports the Model Context Protocol (MCP), an open standard that defines how AI agents can communicate with external tools and data sources, giving it access to over 70 extensions for databases, web browsers, code environments, and more. It also supports the Agentic Communication Protocol (ACP) for provider connections. The project is available as a native desktop application for macOS, Linux, and Windows, as a full-featured CLI for terminal workflows, and as an embeddable API for building your own tools on top of it. The core is written in Rust for performance and cross-platform portability. Goose was originally developed by Block (the company behind Square and Cash App) and has since moved to the Agentic AI Foundation under the Linux Foundation. It is licensed under Apache 2.0.
Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.