Skip to content
AI & automation

AI & automation

goose logo

goose

Logo © Agentic AI Foundation / Block. Used for identification only.

goose (github.com/aaif-goose/goose, formerly block/goose) is Block’s general-purpose AI agent—Rust-built, MCP-native, and model-agnostic. It runs locally as a desktop app or CLI, connects to 70+ extensions, and pairs naturally with Buzz as an in-channel agent.

Visit vendor site →

IR Scorecard

4.5 / 5

Tier A · Budget pricing · Local agent

Overall

87

Quality87/100
Value88/100
Sentiment86/100

Field sentiment

MCP extensions mean one agent can hit Sheets, APIs, and files—if you configure it right.

GitHub / AAIF community

Scores are Installer Report editorial assessments synthesizing public reviews and field chatter — not paid rankings. Weighted: quality 40%, sentiment 35%, value 25%.

Best for

Owners and ops leads who want a self-hosted agent on their machine for SOP drafts, spreadsheet cleanup, and MCP integrations without waiting on SaaS roadmaps.

What it actually is

An extensible agent framework: install, execute, edit files, call APIs via MCP servers, and orchestrate multi-step tasks from terminal or GUI. Supports Anthropic, OpenAI, Google, Ollama, OpenRouter, and subscription-backed providers via ACP.

Where it fits

Shops building internal automation—price sheet normalization, lead CSV cleanup, draft training docs—without shipping data to yet another closed SaaS. Technical enough for an owner-operator or part-time ops hire.

Watch-outs

Not a front-desk chatbot. Requires guardrails on file access and API keys. Verify film specs and customer-facing copy before anything leaves the machine.

Verdict

Best open-source agent stack for shops already comfortable with CLI tools—pair with Buzz for team-visible agent work.

More in AI & automation

Editorial profile — independent of vendor advertising. Logos are used for identification only. See something wrong? Contact editorial.