goose
@opensource.block.xyz
local open source AI agent that automates complex developer tasks, is LLM-agnostic, integrates with MCP, and is available in both Desktop and CLI
Atomic Chat provider with local/private LLMs now working with goose: https://github.com/AtomicBot-ai/Atomic-Chat (new speculative decoding speedup with MTP)
GitHub - AtomicBot-ai/Atomic-Chat: Atomic-Chat is an open source alternative to ChatGPT that runs 100% offline on your computer.
Atomic-Chat is an open source alternative to ChatGPT that runs 100% offline on your computer. - AtomicBot-ai/Atomic-Chat
github.com
Wondering what WebMCP is? ❌ It's not an MCP server ❌ It's not the same thing as Playwright MCP ✅ Instead of automating the UI, WebMCP turns your website into callable tools for AI agents. Read our blog post to learn more! https://block.github.io/goose/blog/2026/03/17/webmcp-for-beginners/
WebMCP for Beginners
WebMCP lets websites expose structured actions that AI agents can call directly. This guide explains how it works, how it differs from MCP and browser automation, and how to build your own WebMCP-enab...
block.github.io
1.26.0 is here, which means: - local inference just got an upgrade - chat with goose via telegram - and even more providers
What if your AI agent could order your lunch. Well it can now with the Neighborhood MCP server. Check out our post showing you how it works and make sure you give it a try yourself. https://block.github.io/goose/blog/2026/02/25/order-lunch-with-goose
Order Lunch Without Leaving Your AI Agent
Use the Neighborhood extension in goose to discover nearby restaurants, browse interactive menus, and place a takeout order, all from a simple chat prompt.
block.github.io
Heard of WebMCP? It's a new standard from Chrome that bridges AI agents and websites. Instead of AI guessing how to click buttons, sites can declare what actions they support. Join Rizel today as she explore how we can use it with goose. 📆 Feb 26 at 1pm ET 📺 https://youtube.com/live/4LfBsSWEitE
Exploring WebMCP | Vibe Code with goose
YouTube video by goose OSS
youtube.com
What if your AI agent could order lunch? 🥑 We connected Goose to Square's Neighborhood MCP App. Now it can find nearby restaurants, browse real menus, and build your order. Avocado toast with feta? It's on the way! Try it out ⬇️
Just ordered lunch without leaving my AI agent 🤯 With the Neighborhood extension in goose, I found nearby restaurants, browsed menus with images right in the chat, and added my orders to the cart, all from a simple prompt. The future of AI-powered food ordering is here 🍜🥗 youtu.be/DG1HUFsekyc
Order Food Directly From Your AI Agent 🍽️ Goose + Neighborhood Extension and MCP Apps
YouTube video by goose OSS
youtu.be
Bradley Axen, Creator of goose, joined Open Source Security to talk about: ✅ Why Block donated goose to the Agentic AI Foundation ✅ AI-generated PRs and open source quality ✅ What comes after chat-based agents https://opensourcesecurity.io/2026/2026-02-goose-aaif-brad-axen/
AGENTS.md
AGENTS.md is a simple, open format for guiding coding agents. Think of it as a README for agents.
agents.md
What's new in goose v1.25.0: 🛡️ macOS sandboxing for every command ⚡ Unified Summon — one extension, two tools 🎨 Rich interactive UI in chat via MCP Apps 📋 Model & provider recipe config right from the GUI 🔏 SLSA build provenance on every binary Watch the full breakdown 👇
Still drawing architecture diagrams by hand and screenshotting whiteboards? The @excalidraw.com MCP App turns visual workflows into a tool call. Prompt goose to generate, edit, and export Excalidraw files.
Yerkebulan Tulibergenov@yerke.bsky.social · 6mo ago🧵 We had a fantastic turnout yesterday at the first East Bay Rust meetup event at Block's Oakland office. We had over 70 people in attendance in total. We had to turn people away since we ran over our initial planned capacity for the event. #rustlang #meetup
Still running one agent at a time like it's 2025? We took the vision of Gas Town and built Goosetown: a lean multi-agent orchestration layer for goose. Goosetown orchestrates subagents across worktrees so you can run flocks, not babysit a single thread. block.github.io/goose/blog/2...
Gas Town Explained: How to Use Goosetown for Parallel Agentic Engineering
Learn how Gas Town and Goosetown lead the industrial coding revolution by teaching AI agents to work together in a team. This beginner guide explains the infrastructure we're using to move from talkin...
block.github.io
With the Playwright CLI skill you can ask goose in natural language to do browser automation and even generate tests. Join Debbie as she dives into skills and how to use the Playwright CLI skill 📆 Feb 19 at 1 pm ET/10am PT 📺 https://linktr.ee/goose_oss
Check out our tutorial on how to use the Playwright CLI skill to automate your tests using natural language. https://block.github.io/goose/docs/tutorials/playwright-skill
Agentic Testing with Playwright CLI Skill | goose
Use goose with the Playwright CLI to generate automated tests using natural language
block.github.io
Come join us as we explore agentic testing and browser automation using the newly released Playwright CLI skill: https://linktr.ee/goose_oss
@goose_oss | Linktree
local open source AI agent that automates complex developer tasks, is LLM-agnostic, integrates with MCP, and is available in both Desktop and CLI
linktr.ee
In this video, we show how to set up @playwright.dev Skills to browse the web and record traces and video, and turn agent actions into real tests. Agentic testing in action. youtu.be/_MpbmD_unnU?...
How to Use Playwright Skills for Agentic Testing
YouTube video by goose OSS
youtu.be
goose made this video about itself making videos. Yes, it's as meta as it sounds. Try goose today with Remotion Agent Skills.
goose is going home... literally. 🏡 Meet our newest goose grant project by Jerry Ochieng: "Goose In A Pond", a local-first AI home assistant running entirely on edge hardware. ✦ Offline voice interaction ✦ Smart device control ✦ Privacy-first, no cloud required ✦ Powered by goose on NVIDIA Jetson
Welcome to goosetown! Population: your distributed team of async subagents. Inspired by Steve Yegge’s Gas Town, a multi-agent workspace manager. Join Rizel as she dives into goosetown and stress-tests it in real time. 📆 Feb 12 at 12 pm ET/9am PT 📺 https://linktr.ee/goose_oss
One shot prompting died the moment we tried to build something real. A clever prompt can’t compensate for broken context. Check out the blog to learn the workflow that replaced it, so your agent stops forgetting and your work stops resetting. block.github.io/goose/blog/2...
One Shot Prompting is Dead
Practical steps and mental models for building context-engineered workflows instead of clever prompts.
block.github.io
Code Mode was trending for weeks. Most devs don't know what it is. What people think: ❌ An MCP replacement ❌ Faster execution What it actually is: ✅ A pattern built on MCP ✅ Batched tool calls ✅ Fewer tokens https://block.github.io/goose/blog/2026/02/06/8-things-you-didnt-know-about-code-mode
8 Things You Didn't Know About Code Mode
Discover how Code mode reduces context rot and token usage in AI agents making them more efficient for long running sessions.
block.github.io
Everyone's buying Mac Minis to run OpenClaw. @blackgirlbytes.bsky.social built her own version using goose as the backend instead. She used the RPI method to stop the agent from hallucinating non-existent APIs and wrote about it: block.github.io/goose/blog/2026/02/06/rpi-openclaw-alternative
How I Used RPI to Build an OpenClaw Alternative
Learn how I built a minimal, personal AI agent using goose and the RPI method.
block.github.io
There are 4 quadrants of AI usage: • Learning Zone • Thinking Ahead • Underutilizing • Frontier rp-why is a goose skill that shows where you are and how to level up. block.github.io/goose/blog/2026/02/06/rp-why-skill
Level Up Your AI Game with rp-why
A goose skill that measures the cognitive complexity of your AI collaboration using the Gas Town × DOK framework.
block.github.io
Happy Friday! It's time to see what's up next in goose programming.
Last time we turned our MCP UI server into an MCP app. Today we’re plugging that same app into ChatGPT. Join us at 1PM ET | 10 AM PT 📺: https://linktr.ee/goose_oss
From suggesting new features to helping troubleshoot issues, it's time to see who are the top goose community stars for January 2026!
We teamed up with @MCPJams to share everything we learned building MCP Apps. Here's some tips that would've saved us hours: → Adapt to host themes → Control what the model sees → Handle loading states properly → Keep the model in sync → Restrict who can trigger tools Link to blog in the thread 👇
