Toggle status bar: Ctrl+SPACE
Cycle backlight brightness: Ctrl+L
Create new file in current project: Ctrl+N
Patrick Knight, Staff Product Manager for Repository Governance. I focus on rulesets and customer properties to let you keep you code safe at scale. 2. What did you build? One sentence + a screen recording (Loom link is fine).
A Chrome extension that turns our extensive emoji collection (57k+) into mosaics.
[video]
Who are you and what do you do at GitHub? I'm Ashley, Senior Director of Developer Relations at GitHub. I lead a team focused on developer community, content strategy, and helping developers get the most out of GitHub's tools
What did you build? I built Pocket Monster; an AI childcare assistant that helps my older daughter Kylie take care of my 5-year-old Marlowe while my partner and I are traveling in Japan. It's powered by Claude via GitHub's Copilot SDK and has all the specific care instructions, routines, and house quirks that would normally require dozens of text messages back and forth.
How long did v1 take to make? About five hours to get to a functional prototype that actually worked. Then I spent three more days tweaking it obsessively - honestly past the point of usefulness - because I couldn't leave it alone. Classic builder problem.
Hey! This is a quick little "crash course" on how to use your terminal on your computer.
First of all, what is the Terminal?
It's not a hardcore hacker setup like you see in the movies (although you can make it look like one). The terminal is where you talk to your computer using text instead of clicking. It's like having a conversation with your operating system.
Now, to open the terminal, it might look different depending on your computer. On a Mac, you want to open an application called Terminal. On Windows, you have a few options, either Command Prompt, PowerShell, or Windows Terminal. All of these apps come built in with your computer! You can install other ones, but we'll stick with the basics for now.
Memory... all alone in the moonlight...
GitHub Copilot has Memory now, y'all.
We know it can be annoying to have to start from scratch with every session with an AI tool.
Cross-agent memory allows agents to remember and learn from working with you, without relying on explicit user instructions. Each interaction teaches Copilot more about your codebase and conventions, making it better and better over time. Kind of like a cast iron skillet.
Anyway.
Have you seen GitHub Agentic Workflows? It's pretty sick.
They are automated, intent-driven repository workflows that run in GitHub Actions, and they're authored in Markdown, and executed with coding agents.
In a bit more plain English: you describe the outcomes you want in plain Markdown, add this as an automated workflow to your repository, and it executes using a coding agent in GitHub Actions.
This brings the power of coding agents into the heart of repo automation. Agentic workflows run as standard GitHub Actions workflows, with added guardrails for sandboxing, permissions, control, and review.
In even plainer English: your repositories can automate themselves, securely.
0:00 This is a quick demo of GitHub Actions Debugging. We start with VS Code open in our repository. We will use the GitHub Actions extension to check on our latest workflow run. 0:12 We see that this step failed, so we'll use the GitHub Actions Debugger to fix the issue. First we'll go to the failed job, and click on rerun and debug job. 0:23 This will take us to an active debugging session in our workflow file, with the debugger paused before the first step. 0:29 We'll set a breakpoint in the task that's failing and hit continue to run the first two steps and hit the breakpoint. 0:36 Here I have access to the variables, call stack, and watch Windows as I would expect in any normal debugging session. 0:44 I can see that the step simply checks if a file exists. The actions extension allows me to navigate the remote file system to see if this file exists and I quickly notice that it's not where it's supposed to be. 0:55 I now click Click on Step over to run this step and see how its state changes. I can also c
Memory usage is opt-in while in public preview, so go to your Copilot settings to enable it.
Here we see a pull request making changes to an API and bumping the version accordingly. The Copilot code review agent is reviewing the changes for correctness, completeness, and quality. Let’s take a look at the session logs.
The code review agent noticed that the API version must remain synchronized across three parts of the codebase to prevent integration failures, and stored this learning as a memory.
A few days later a developer assigned the Copilot coding agent this issue which requires changes to the API.