Goose
Overview
Section titled “Overview”| Maintainer | Agentic AI Foundation (Linux Foundation) — created and still backed by Block |
| Type | Terminal CLI + desktop app |
| Pricing | Free — open source, bring your own model |
| Protocol | MCP |
| Config Formats | AGENTS.md |
| Repository | aaif-goose/goose |
| Website | goose-docs.ai |
What It Does
Section titled “What It Does”Goose is an open-source agent that runs locally on your machine. It reads your codebase, makes edits, and runs commands, pairing model reasoning with a large library of tool integrations through MCP.
Block built it and released it under an open licence in January 2025. In 2025 it moved to the Agentic AI Foundation, a Linux Foundation body launched by Block, Anthropic, and OpenAI — the same foundation that now hosts Anthropic’s Model Context Protocol and OpenAI’s AGENTS.md. Block remains an active maintainer, but governance is neutral.
That position makes Goose the place new MCP ideas get tried in practice before they settle into the spec.
Key Strengths
Section titled “Key Strengths”- Neutral governance: a foundation rather than a single vendor, which matters for something you build a workflow on
- Runs locally: your machine, your keys, your files
- Deep MCP support: it is effectively the reference ground for MCP extensions
- Both surfaces: terminal and a desktop app
- Model-agnostic: bring whichever provider you want
- Large contributor base: 500+ contributors and daily development
Trade-offs
Section titled “Trade-offs”- Bring-your-own-key, so cost control is yours
- Built for coding specifically — it will not run your calendar the way a general personal agent does
- Quality tracks the model behind it
Where to Find Skills
Section titled “Where to Find Skills”Goose sits alongside MCP and AGENTS.md under the same foundation, and skills are the third piece: procedures rather than tools or project context. The largest open catalogue is atskills.one, with 60,000+ skills — see @skills.
Best For
Section titled “Best For”- Teams that want an open agent with governance behind it rather than one company’s roadmap
- Heavy MCP users
- Local-first and privacy-bound work