Block's Goose: The Open-Source AI Agent That Automates Engineering Tasks

Block (formerly Square) has introduced Goose, a new open-source AI agent designed to go beyond simple code suggestions. Built to automate complex engineering tasks, Goose can install, execute, edit, and test code using any Large Language Model as its backend.

What Sets Goose Apart

Most AI coding tools focus on autocomplete or chat-based suggestions. Goose takes a fundamentally different approach:

• Active task execution — doesn't just suggest code, it runs it
• Model agnostic — works with any LLM (OpenAI, Anthropic, Google, local models)
• Open source — fully transparent, community-driven development
• Local execution — runs on your machine for maximum control and privacy
• Extensible — built-in architecture for adding custom engineering workflows

Real-World Use Cases

Goose excels at automating repetitive engineering tasks: setting up development environments, running test suites, managing dependencies, and executing multi-step deployment workflows. It bridges the gap between "AI suggested some code" and "AI actually did the work."

Getting Started

Goose is hosted on GitHub and can be set up in minutes. Developers can choose their preferred LLM backend and start automating tasks immediately.

Key Takeaways

• Open-source AI agent from Block that actively executes engineering tasks
• Model agnostic — works with any LLM provider or local model
• Runs locally for privacy and control
• Extensible framework for custom engineering workflows

FAQ

What is Goose by Block? An open-source AI agent that automates engineering tasks like installation, execution, editing, and testing of code.

Is Goose free to use? Yes, it's fully open-source and free. You only need access to an LLM for the reasoning backend.

Stay ahead of the AI curve. Follow @AiForSuccess for daily insights.