Kim is fresh out of the academy and has spent only 18 months learning development. When I asked her how she felt about Goose she had mixed reactions. While she found it cool that it could do so much for her, she wasn’t actually sure of what it was doing for her half the time, and also why. When she asked Goose to fix a broken build, or chase a bug, It would complete the task and claim ‘Success!’. Which is great, however she felt she wasn’t actually learning as much as when she was in the academy. Add on that sometimes she didn’t even know what to ask Goose to do sometimes.

That afternoon I started to see if I could get Goose to be more than just a ‘magic box’ for my Junior Devs. What if Goose could instead act as a mentor and also teach as well as speeding up development?

Working with AI tools can feel like working with a flaky, chaotic, but overconfident coworker. You know, the kind who forgets tasks, lies unprovoked, starts new projects without telling you, then quits halfway through. It's enough to make you say: "Forget it. I'll do it myself." But before we write off AI entirely, it's worth understanding what's actually happening under the hood so we can avoid common pitfalls and make AI tools worth using.

I taught a vibe coding workshop at UC Berkeley and informed the students that we'd be spinning up 7 subagents. Someone quickly raised their hand and asked "what is a subagent?". At that moment, I realized we're just throwing out terms like agent, multi agent, and subagent, and not really taking the time to explain what these are. So, here goes... a 101 breaking down these various coordination patterns and when to use them.

A website to me is a corner of the internet where I can be who I am, share my work, and ultimately a place I can do whatever I want with. For it to be anything but my personality personified, especially as an ex-nerdy blog designer (in my middle school and high school days), felt so sad! Until suddenly, what started out as a harmless "404 Day" challenge quickly turned into making that website in basically no time.

"The best AI agent workflows go beyond demos. They deliver real productivity."

The DevRel team at Block is a huge fan of PulseMCP. Their weekly newsletter has been an amazing way for us to discover trending MCP servers and stay in the loop with any changes within the ecosystem. When the PulseMCP creators, Mike and Tadas, shared their goals of using Goose to help automate the boring parts of their newsletter workflow, we were excited to see what they'd build.

Their implementation showcased exactly why we built Goose's feature set the way we did, and they documented the entire journey to help others learn from their experience.

When building Model Context Protocol (MCP) servers, most development focuses on tool functionality, ensuring tools execute and return expected results. But just as critical is the quality of tool metadata: descriptions, tooltips, and input schemas. These elements form the "interface language" between tools and AI agents like Goose.

Yet metadata often goes untested. This can break tool discovery and silently degrade agent behavior. In this post, we’ll show how to automate metadata validation using composable Goose recipes, turning manual QA into modular, repeatable workflows that:

• Validate tool discoverability and parameter accuracy
• Detect regressions early
• Safely reduce token usage

All while maintaining the quality that AI agents depend on.

Ever wondered what happens when you let two AI models work together like a tag team? That’s exactly what we tested in our latest livestream—putting Goose’s Lead/Worker model to work on a real project. Spoiler: it’s actually pretty great.

The Lead/Worker model is one of those features that sounds simple on paper but delivers some amazing benefits in practice. Think of it like having a project manager and a developer working in perfect harmony - one does the strategic thinking, the other gets their hands dirty with the actual implementation.

Goose recently released support for MCP-UI which allows MCP servers to suggest and contribute user interface elements back to the agent.

MCP-UI sits on top of the protocol, but instead of text/markdown being the result, servers can return content that the client can render richly (including interactive GUI content).

What if I told you that you could build a complete, working web application in under an hour using nothing but AI agents? Not just a simple "Hello World" app, but a full-stack application with a backend API, responsive frontend, unit tests, and documentation?

That's exactly what we accomplished during the Vibe Coding workshop at Berkeley's Agentic AI Summit, where I demonstrated how to use Goose's subagent orchestration to spin up an entire development team of AI agents. Each agent took on a specific role - from product planning to QA testing - and worked together to build "AI BriefMe", a web app that generates executive style briefings on any topic.

Machine learning and data science workflows are inherently iterative. You load data, explore patterns, build models, and refine your approach based on results. But traditional AI assistants lose context between interactions, forcing you to reload data and re-establish context repeatedly—making data-heavy development slow and expensive.

The MCP Jupyter Server solves this by enabling AI agents like Goose to work directly with your Jupyter notebooks, maintaining persistent memory and state across interactions while letting the AI interact with your data through code execution rather than raw data transfer.