AI Tool

goose Review

Goose is a general-purpose, native open-source AI agent available for macOS, Linux, and Windows, built on open standards like the Model Context Protocol (MCP) and Agent Client Protocol (ACP).

shipped Jul 23, 2026aifree

Why it matters

1Goose is an Apache 2.0 licensed project under the Agentic AI Foundation.

2It supports over 70 integrations through the Model Context Protocol.

3The agent works with more than 15 LLM providers, including Anthropic, OpenAI, and Google.

4Block, the developer, reports 60% of its workforce uses Goose weekly, with 50-75% development time savings.

About goose

Business Model

Open Source

Platforms

macOS, Linux, Windows

Target Audience

Developers and users of AI tools

GitHubOpen Source

Specs

API Docs

View Documentation →

GitHub

View Repository →

API Available

Yes, public API

overview

What is goose?

goose is a general-purpose AI agent tool developed by Block (formerly Square) that enables engineers, developers, researchers, writers, and data analysts to automate tasks across various domains. It functions as an autonomous agent that can build, execute, test, and debug code independently, running directly on a user's machine with desktop, command-line interface (CLI), and API access. The agent emphasizes extensibility, supporting over 70 integrations through the Model Context Protocol (MCP) and Agent Client Protocol (ACP), and is an Apache 2.0 licensed project under the Agentic AI Foundation. It is designed for diverse workflows, including code generation, research, writing, automation, and data analysis, and is compatible with macOS, Linux, and Windows platforms.

features

Key Features of goose

Goose offers a comprehensive set of features designed for versatile AI agent functionality, emphasizing local operation and extensibility.

Native desktop application for macOS, Linux, and Windows.
Command-line interface (CLI) for scriptable automation.
API access for programmatic integration into existing systems.
Extensible architecture supporting over 70 integrations via the Model Context Protocol (MCP).
Compatibility with more than 15 LLM providers, including Anthropic, OpenAI, Google, Ollama, OpenRouter, Azure, and Bedrock.
Support for existing LLM subscriptions (e.g., Claude Code, ChatGPT, Gemini) through the Agent Client Protocol (ACP).
Security features, including prompt injection detection.
Built on open standards: Model Context Protocol (MCP) and Agent Client Protocol (ACP).

use cases

Who Should Use goose?

Goose is designed for a broad target audience requiring automation and AI assistance across various technical and creative domains.

Engineers and Developers: For automating engineering tasks like code migrations, refactoring, deployment, code reviews, and full-stack development.
Researchers and Data Analysts: To assist with research, data analysis, and workflow orchestration across multiple tools.
Writers: For content generation, summarization, and other writing-related automation.
QA Engineers: To automate tasks such as dependency updates, test execution, and code migration.
Users needing Task Automation: For general task automation across diverse domains, including onboarding to new projects or unfamiliar codebases.

how to use

How to Use goose

Goose can be utilized through its desktop application, command-line interface, or API, allowing users to automate tasks by defining objectives and leveraging its extensive integrations.

1Download and install the native desktop application for macOS, Linux, or Windows from goose-docs.ai.
2Configure your preferred LLM providers by adding API keys or linking existing subscriptions (e.g., OpenAI, Anthropic).
3Utilize the command-line interface (CLI) to initiate tasks, such as goose build project_description.
4Define specific objectives for code generation, research, writing, or automation tasks.
5Leverage the Model Context Protocol (MCP) extensions to integrate with databases, APIs, browsers, GitHub, and Google Drive.
6Monitor and debug agent execution, allowing Goose to self-correct errors and refine outputs.

pricing

goose Pricing & Plans

Goose is an open-source project released under the Apache 2.0 license, making it available for free. Users are responsible for any costs associated with third-party LLM API usage.

Goose: Free (Apache 2.0 licensed)

Pros

+Native open-source AI agent available for macOS, Linux, and Windows.
+Supports over 70 integrations through the Model Context Protocol (MCP).
+Compatible with more than 15 LLM providers, including major services and local models.
+Offers desktop, command-line interface (CLI), and API access for flexible usage.
+Designed for diverse workflows, including code, research, writing, and automation.
+Emphasizes data privacy by running locally on the user's machine.

Cons

−Requires initial setup and configuration, which may be a hurdle for new users.
−While robust, the command-line interface could benefit from a more interactive UI for some users.
−Performance is dependent on local machine resources and chosen LLM provider.
−Users are responsible for managing API keys and costs for third-party LLMs.

Similar Tools

goose vs Competitors

Goose differentiates itself in the AI agent landscape through its local-first, open-source nature, and emphasis on broad workflow automation via open standards.

Open InterpreterOn Stork Compare

It provides natural language control over the local machine, executing Python, bash, and browser commands directly through a conversational interface.

Similar to Goose in offering cross-platform local execution and general-purpose capabilities like data analysis and file manipulation. However, Open Interpreter's primary focus is on direct system control via conversational interface and code execution, whereas Goose emphasizes broader workflows (code, research, writing, automation) and its Model Context Protocol for integrations. Both are open-source and free.

AGiXTOn Stork Compare

AGiXT is an enterprise-focused agent framework with an extensible plugin architecture, supporting any LLM provider and including a web UI for agent configuration.

AGiXT functions more as a framework for building agents, offering a web UI and RESTful API, akin to Goose's API access, but it is typically deployed via Docker and targets enterprise workflows. Goose provides a native desktop application and is designed for a broader user base beyond just enterprise. Both are extensible and support various LLMs, and AGiXT is open-source.

OpenHandsOn Stork Compare

This MIT-licensed, open-source platform is specifically designed for building and running AI coding agents, featuring a local-first workspace called Agent Canvas for managing multiple agent sessions.

OpenHands is more specialized for AI coding agents and offers a unique 'Agent Canvas' for managing multiple agent sessions, whereas Goose is a general-purpose agent for diverse tasks including code, research, writing, and automation. Both are open-source, local-first, and emphasize customization, but OpenHands' core focus is narrower on coding tasks.

Agent Zero AI↗

It focuses on operational reliability and deterministic execution for AI agents, offering a guided desktop installation and management of local/remote instances.

Agent Zero AI directly competes with Goose by providing a desktop experience for AI agents and emphasizing reliable, consistent performance through 'agentic context engineering' and 'dynamic tool creation.' Both are open-source, free, and aim for general-purpose utility, with Agent Zero highlighting its foundational architecture for consistent agent behavior.

See every goose alternative, compared→

Visit goose↗

Connect

𝕏

X / Twitterx.com/goose_oss

⌘

GitHubgithub.com/aaif-goose/goose

LinkedInwww.linkedin.com/company/goose-oss

💬

Discorddiscord.gg/goose-oss