Not every task needs a genius. And not every step should cost a fortune.

That's something we've learned while scaling Goose, our open source AI agent. The same model that's great at unpacking a planning request might totally fumble a basic shell command, or worse - it might burn through your token budget doing it.

So we asked ourselves: what if we could mix and match models in a single session?

Not just switching based on user commands, but building Goose with an actual system for routing tasks between different models, each playing to their strengths.

This is the gap the lead/worker model is designed to fill.

As Goose users, we have two main ways to provide persistent context to our AI assistant: the .goosehints file and the Memory Extension MCP server. Today, I'll share what's in my .goosehints file, why some of it should probably move to the Memory Extension, and how you can make that choice.

Detection engineering stands at the forefront of cybersecurity, yet it’s often a tangled web of complexity. Traditional detection writing involves painstaking manual processes encompassing log format and schema comprehension, intricate query creation, threat modeling, and iterative manual detection testing and refinement, leading to time expenditure and reliance on specialized expertise. This can lead to gaps in threat coverage and an overwhelming number of alerts. At Block, we face the relentless challenge of evolving threats and intricate system complexities. To stay ahead, we've embraced AI-driven solutions, notably Goose, Block’s open-source AI agent, and Panther MCP, to allow the broader organization to contribute high-quality rules that are contextual to their area of expertise. This post delves into how we're transforming complicated detection workflows into streamlined, AI-powered, accessible processes for all stakeholders.

Goose is LLM-agnostic, meaning you can plug in the model of your choice. However, not every LLM is suitable to work with agents. Some may be great at answering things, but not actually doing things. If you're considering which model to use with an agent, these 3 prompts can quickly give you a sense of the model's capabilities.

Localhost Ports Hoarding​

I'm perpetually drowning in open tabs. Yes, I do need Bluesky, ChatGPT, Claude, Goose, Cursor, Discord, Slack, Netflix, and Google Docs all open at the same time. I've learned that tab management isn't my only vice.

"Hi, my name is Rizel, and I'm a localhost ports hoarder. 👋🏿"

I taught Goose how to drive (a rover)​

Goose has no hands, no eyes, and no spatial awareness, but it can drive a rover!

I came across a demo video from Deemkeen, where he used Goose to control a Makeblock mbot2 rover using natural language commands like "drive forward/backward," "beep," and "turn left/right" powered by a Java-based MCP server and MQTT.

Inspired and excited to take it further, I taught the rover to spin, blink colorful lights, and help me take over the world!

A couple of weeks back, Qwen 3 launched with a raft of capabilities and sizes. This model showed promise and even in very compact form, such as 8B parameters and 4bit quantization, was able to do tool calling successfully with goose. Even multi turn tool calling.

I haven't seen this work at such a scaled down model so far, so this is really impressive and bodes well for both this model, but also future open weight models both large and small. I would expect the Qwen3 larger models work quite well on various tasks but even this small one I found useful.

Development teams can operate like sports teams. Each member has a role and a shared playbook helps coordinate efforts. Let's explore how AI-driven "plays" can form a starter "playbook" for your dev team, helping with common technical tasks. You can use recipes with Goose to leverage the Model Context Protocol (MCP) to make you more productive.

Ratatouille isn't just a heartwarming (and slightly unhygienic) film about a rat chef. It's also analogous to a popular tech trend: AI agents and the Model Context Protocol (MCP).

So you’ve heard about Goose. Maybe you saw a livestream, someone on your team mentioned it, or you just stumbled into our corner of the internet while trying to automate your dev setup. Either way—love that for you.

Goose is a local, open source AI agent that can automate tasks, interact with your codebase, and connect to a growing ecosystem of tools. But before you hit install, here are four things you should know to get the most out of it.