Skip to main content

Goose Developer Guide

What is this about?

Goose is a native open-source AI agent with a desktop app, CLI, and API. This guide explains how Goose is structured, how to install and configure it, how providers and extensions fit together, and where to start when you want to contribute.

Checked against primary sources

This guide is based on the official Goose docs and the official aaif-goose/goose repository, checked on June 26, 2026.

In this section​

1. What Goose is​

The shortest accurate description is:

Goose is a general-purpose local AI agent, built in Rust, that you can use as a desktop app, a terminal tool, or an embedded API surface.

That matters because Goose is broader than a coding assistant. The upstream README explicitly positions it for code, research, writing, automation, data analysis, and other knowledge work.

2. The mental model​

Think of Goose as five cooperating parts:

LayerWhat it does
CLI and desktop UIUser-facing interaction surfaces
Rust agent coreMain runtime and execution loop
provider layerConnects to 15+ model backends
MCP extensionsAdds external capabilities
API and recipesLets teams embed or productize workflows

3. Why developers would choose it​

  • It is local-first and runs on your machine.
  • It is multi-provider, including local-model paths such as Ollama and LM Studio.
  • It supports both MCP and ACP, which makes it more extensible than a closed vendor CLI.
  • It has both a desktop app and a full CLI, so teams can support technical and less technical users with the same core runtime.

4. The project posture​

Goose is no longer just a personal tool repo. It has moved into the Agentic AI Foundation (AAIF) at the Linux Foundation, which is a strong signal that the project expects broader ecosystem participation and longer-lived governance.

  1. this overview,
  2. installation and quickstart,
  3. provider configuration,
  4. extension and MCP docs,
  5. contribution docs once you are ready to work in the repo.

Sources​