Goose
Block's Apache-2.0 on-machine AI developer agent — runs tasks end-to-end in your terminal, any LLM backend.
Goose
Goose is an open-source, on-machine AI developer agent from Block (formerly Square), released under Apache 2.0. It runs autonomously in your terminal, executing multi-step coding and DevOps tasks: writing code, running tests, debugging errors, managing files, and calling external APIs — all powered by the LLM of your choice.
Key features
- On-machine execution: reads/writes files, runs shell commands, interacts with the local dev environment
- LLM-agnostic: works with Anthropic Claude, OpenAI GPT-4, Ollama, and other backends
- MCP-compatible — extend with any MCP server (databases, APIs, cloud tools)
- Persistent sessions with memory across tasks
- Safe by design: shows planned actions before execution, supports sandboxing
- Apache-2.0 license
Quick start
# Install via pip
pip install goose-ai
# Or via Homebrew (macOS)
brew install block/tap/goose
# Start a session
goose session start
# > write a FastAPI CRUD app for a todos list, with SQLite and pytest tests
# Goose plans, writes code, runs tests, fixes errors — autonomously
Add MCP tools
# Add a GitHub MCP server to Goose's toolbox
goose toolkit add github
Install via ai-supply
npx ai-supply add goose-dev-agent
Curated mirror of the open-source Goose (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/goose-dev-agent/check). Click a policy:
Consume Goose programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/goose-dev-agent
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/goose-dev-agent/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add goose-dev-agent
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/goose-dev-agent/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "goose-dev-agent" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.