Goose - AI Code Assistants Tool

Overview

Goose is an open‑source, on‑machine AI agent for automating software development workflows end‑to‑end. It runs locally as both a desktop application and a command‑line client, and it orchestrates plans that can install dependencies, run and debug code, edit files, and execute test suites — all driven by your choice of LLM provider. Goose is built to be model‑agnostic (OpenAI, Anthropic, Google Gemini, Ollama/local models and others) and supports multi‑model configurations and provider routers to trade off latency, capability, and cost. ([github.com](https://github.com/block/goose?utm_source=openai)) The platform is extensible via MCP (Model Context Protocol) servers and an extensions system that gives agents access to tools such as Git integrations, browser/control automations, and external APIs. Goose provides recipes and sub‑recipes to break complex agent workflows into reusable pieces, supports session persistence for long conversations, and exposes both GUI flows and a rich CLI for scripting and automation. Recent docs and blog posts highlight new capabilities such as MCP‑UI (server‑suggested GUI content) and continuous releases with packaged desktop installers and CLI binaries. ([block.github.io](https://block.github.io/goose/docs/getting-started/installation/?utm_source=openai))

GitHub Statistics

• Stars: 25,657
• Forks: 2,326
• Contributors: 328
• License: Apache-2.0
• Primary Language: Rust
• Last Updated: 2026-01-09T14:41:28Z
• Latest Release: v1.19.1

According to the official repository, Goose is a high‑activity open source project with a permissive Apache‑2.0 license. The project shows substantial community interest (the repository reports ~25,657 stars, ~2,326 forks, and hundreds of contributors) and an active release cadence (releases and CLI artifacts are published on GitHub). The repo and docs are updated regularly and the project publishes packaged release assets (desktop binaries, platform tarballs, and an install script) for cross‑platform installation. Note: codebase metrics and release details are visible in the project repo and releases page. ([github.com](https://github.com/block/goose?utm_source=openai))

Installation

Install via brew:

brew install --cask block-goose  # installs Goose Desktop on macOS via Homebrew. See Goose docs for macOS binaries. ([block.github.io](https://block.github.io/goose/docs/getting-started/installation/?utm_source=openai))

curl -fsSL https://github.com/block/goose/releases/download/stable/download_cli.sh | bash  # download and install the Goose CLI (Linux/macOS/WSL/Windows via Git Bash). ([github.com](https://github.com/block/goose/releases?utm_source=openai))

curl -fsSL https://github.com/block/goose/releases/download/stable/download_cli.sh | CONFIGURE=false bash  # non-interactive CLI install. (Windows PowerShell versions exist in releases.) ([github.com](https://github.com/block/goose/releases?utm_source=openai))

goose update  # update CLI to the latest release (after installing). See updating guide in docs. ([block.github.io](https://block.github.io/goose/docs/guides/updating-goose/?utm_source=openai))

Key Features

• Runs locally as both a Desktop app and a command‑line client for interactive or scripted use. ([block.github.io](https://block.github.io/goose/docs/getting-started/installation/?utm_source=openai))
• Model‑agnostic LLM support (OpenAI, Anthropic, Google Gemini, Ollama/local models, OpenRouter, Tetrate). ([block.github.io](https://block.github.io/goose/docs/getting-started/installation/?utm_source=openai))
• MCP server and extension ecosystem for integrating tools like Git, browser control, and custom APIs. ([block.github.io](https://block.github.io/goose/blog/2025/08/11/mcp-ui-post-browser-world/?utm_source=openai))
• Tool calling: execute shell commands, install packages, run tests, and manage files from agent plans. ([block.github.io](https://block.github.io/goose/docs/troubleshooting/known-issues?utm_source=openai))
• Recipes and sub‑recipes let you compose complex agent workflows and reusable subagents. ([pulsemcp.com](https://www.pulsemcp.com/building-agents-with-goose/part-4-configure-your-agent-with-goose-recipes?utm_source=openai))
• MCP‑UI support: MCP servers can return rich GUI elements for the client to render. ([block.github.io](https://block.github.io/goose/blog/2025/08/11/mcp-ui-post-browser-world/?utm_source=openai))
• Multi‑model configuration and router integrations (Tetrate Agent Router, OpenRouter) for balancing performance and cost. ([block.github.io](https://block.github.io/goose/docs/getting-started/installation/?utm_source=openai))
• Cross‑platform packaged releases and an install script for quick CLI installation on Linux, macOS, Windows/WSL. ([github.com](https://github.com/block/goose/releases?utm_source=openai))

Community

Goose has an active, visible community and documentation presence: an Apache‑2.0 GitHub repo with tens of thousands of stars (~25,657), thousands of forks (~2,326) and hundreds of contributors (~328), maintained docs, tutorials, and a project blog. The team publishes regular releases and packaged assets (desktop installers and CLI installers). Community feedback is visible in site testimonials, third‑party tutorials, and issue threads; users highlight productive flows (scaffolding projects, automations) while the Known Issues and troubleshooting docs reflect responsive maintenance for platform quirks like permissions, rate limits, and provider setups. For code/contribution activity and latest release artifacts, see the repository releases and the official docs. ([github.com](https://github.com/block/goose?utm_source=openai))

Key Information

• Category: Code Assistants
• Type: AI Code Assistants Tool