[Full Workshop] Vibe Coding at Scale: Customizing AI Assistants for Enterprise Environments

Okay, uh, thank you. We have, um, "Vibe Coding at Scale." It's a talk I'm going to give at the expo, in a very short version, without all the hands-on stuff. So this is whatever how long this will take.

It's all about vibe coding. So we're going to give fully in to the vibes, embrace exponential, and forget that code exists.

It is about focusing on the output and not actually on the code. And that's where people disagree. Like, if I don't look at my code, as an engineer, what am I? Um, and then if what does even embracing exponentials mean in this case?

So, and that's we're going to get into that. And, but we're definitely going to forget that code exists. And I think, uh, the at the Anthropic conference two weeks ago was a great chart of how exponentially agents run longer and longer and generate more and more code.

So slowly forgetting that code exists, and you want to review every piece of code, but building trust and adding guardrails to the AI is what this talk is about. And this Vibe Coding journey starts kind of initially what pe most people saw of just like, I build an app in just one day, and I put it online, and I made money, and I then things happened to the app, and things got leaked, and things were no longer fun.

And that's that fun chaos state of Vibe Coding. And we're trying to move to professionals then. And that initial state is what I'm now terming YOLO vibes. It's the unofficial wording, but it's all about creativity and speed. And it's a good place to be.

It's being it's there is instant gratification. It's getting things up. It's about learning. It's not about shipping products, but and so we need to get there. Second stage is structured vibes. It's all about balance and sustainability. Like, how do you bring maintainability, more readable code, like things you might actually somebody in the end might want to read that code, and you'll have a handover to somebody else.

And you want to have some quality control that what you built is maintainable and not just a throwaway project. And lastly, we get to spectrum vibes. And if you have done anything on Reddit or on blogs in the in the past, then you past few weeks, you've probably seen people sharing their kind of best practices.

This is how I finally got good value out of AI. And those are those best practices are emerging, but they bring you that scale, reliability, and velocity that comes with that, while still hopefully giving you some speed and gratification along the way, but reducing that chaos, while maybe keeping the fun.

So Vibe Coding, initially, where we see this outcome first, as I said, it's all about not you don't even want to look at the code. Like, if you're in an editor, and I see actually people framing their experience in AI editors of a low-code mode.

So they just look at the chat panel and look at whatever comes out of it, and that's the outcome first. It's all about natural language. It's all about just staying in the flow and doing working with the AI.

And it's about auto-accepting changes. So until it maybe no longer works, we might want to undo, but otherwise, we just keep talking to the AI. No, you did this wrong. Try again. Um, fix it or go to jail is a very popular one.

So you can try all kind of, but stay in natural language. Don't don't be too specific. So there is a use case for that, though. You want to get a sense for YOLO vibe coding for rapid prototyping, for proof of concepts where you just want to get something out.

And that's where I actually have a ton of conversations with larger companies who want to who start a conversation like, how can we do vibe coding? And for them, it's all about getting people who are non-technical to be able to communicate ideas.

It's about UX people just making a mockup and being able to bring that to a meeting and being able to communicate what they want to do in the mockup. It's all about learning. Like, I we had last, uh, two weeks ago, I built we had one hour of Vibe Coding on stage live, and people built games.

And I used Three.js, which I have tried many years ago, but I haven't used Three.js in a while. But once I got the code running, I could start getting into it, like how to structure it, how does it make shapes.

And I could actually understand technology because I have something working. And that's that's really the power of getting something up and running that that gives you the technology something hands-on to actually try out. And of course, personal projects.

So, um, I'm sure how many of you had sit sit down with somebody who's non-technical and showed them Vibe Coding and just built a your water tracking app, or you built something with your kids. There's all kind of personal projects you can now finally solve over the weekend thanks to YOLO vibe coding.

So let's do some YOLO vibe coding. This is, uh, AI generates. So the images do reflect the vibe. So this is really about voice input, relaxing. I guess coffee is in there as well, but let's get it going.

Um,

okay. So in VS Code for YOLO vibe coding, we're going to start with an empty VS Code. So hit Command-Shift-N. Sorry?

Are we supposed to use Insiders if you want?

I might show off Insider stuff, but all of what I'm showing is also in Stable. If you have Insiders, use Insiders. Insiders is the pre-release version of VS Code that ships on a daily basis, like Firefox nightly, uh, Chrome Canary dev ships nightly.

So on your left side, you will have no folder open. On yourright side, you will have Copilot open. Everybody got this?

Raise your hand. Cool. Awesome. So, um, who has used agent mode? Just checking. So I'm not have to explain everything. Okay. Cool. So agent mode, uh, the probably the default you want to have set and default setting. Um, Claude 4 is great at front-end stuff.

So for me personally, my my favorite. Is that big enough? Yeah. And now what's interesting, so just one one quick tour of how we're going to do Vibe Coding is there's actually this interesting setting here. Is this zooming?

Yeah. Is news workspace in VS Code. And for this first round of Vibe Coding, I want you to go into this tools picker down here and actually disable scaffold new workspace. Because it will help you scaffold your workspace, but it will lower the vibes if you're just trying to do HTML coolness.

How do you get that?

Oh, yes. Uh, tools picker down here. That's the little one. Uh, how to get into this menu, there's a tool this tools picker. Okay.

I don't have the way I heard.

Uh-oh. Yes. It might be in a it was in a different spot before. Um, if you don't see it, check that you're in agent mode. Otherwise, you don't have tools.

Oh.

Yes. That's my yeah. Okay. First step, enter the panel, switch to agent mode down here. It should might be in Ask by default.

Yes, it is.

Cool. And then second step, go into the tools menu, which only appears in agent mode. Soon, agent mode's going to be default, and this whole where is my my tools picker is less of a problem. Okay. So and then what you want to uncheck is this new section.

And we're going to check out new.

FTP server and install section.

Yes. Do you see nothing?

And I see those two items and nothing else.

That's not enough. I I see more more of you. Um, you might be yes. I think that's fine. Let me just.

Check Insiders ahead.

Shoot.

Okay. You got Insiders. Yes. I think this is actually it's very recent, and we're actively working on that. So keep it keep it checked. That's fine. So let's actually use it. Um, let's start with our first Vibe Coding using you because it's so hard to disable.

And the way we're going to do this is, um, create a, uh, let's do React Vite. And that's first lesson for Vibe Coding. Use stacks that are kind of popular in front-end where the AI doesn't have to reason too much and make wild guesses.

So React and Vite are good ways to run a project. Um, website that for what do we do? Um, hydration tracking. Water hydration water consumption app. Simple and big accessible UI following, uh, what do we tell the AI to make it really beautiful?

We tell it, uh, I like to tell it, Apple design principles.

You can infuse it with whatever design sense you have. Mine is just make it look pretty, which always helps somehow adding that extra. So Vite, water tracking, hydration. And that's what I see. So we don't give it any constraints.

We don't tell it how many buttons. Like, is it mobile-friendly? Is it not mobile-friendly? Is it, um, what CSS language to use? Uh, I might actually not do React Vite, but might oh, yeah. Let's do that with material design.

With material

material design. So yeah, now we got a stack. We got a design direction, and we tell told it to make it pretty. Okay.

And then hit run. So we're now in in a no folder state. So what's first happening, it's going to tell us to empty to open an empty folder. Everybody got that in their flow? Okay.

It's going to hit continue. Now it's going to actually ask me to open a folder. You're going to make a new folder. Vibing at NAI Engineering. Put that anywhere where you put your stuff, where you put your code.

And now it's actually opening a new folder, and it's going to continue the setup. And to explain a bit what's happening here, so this is using the new command. A new command is is optimized for creating projects from scratch, which if you look at the internet of how people evaluate AI coding tools is what every second person does.

Can it make me a water tracking app? Can it make me a movie database app? And so we optimize for this flow, but also it makes for this nice Vibe Coding from scratch because everybody struggles with what is theright stack, how do I get started, and this is what this is here for.

Uh, we got the latest. So we got we can now read the commands. And this is maybe where we do the first tweak of our settings. So if you go into settings and search for approve, you will find the auto approve.

And that's the first rule of Vibe Coding. As we said, we we don't want to look at code, and we just want to have the AI do stuff. So what you can do in my my case, I'm going to actually go to over to workspace from user to workspace.

And that means that setting is only set for this workspace, which is the safest way to use this setting. So use with caution. I'm going to auto approve, which means all of these continue buttons won't happen anymore, and we're just going to get results.

So check that box, close it again, and we can hit continue here. This is using Vite. This is fine. Now we're going to stop worrying about the code. Uh, but I can still read the plan. So install material dependencies, create hydration tracking, Apple inspired design, and project structure.

So it's running commands. Um,

why does it say files already in here? Okay. Cool.

Skips. What did it create? Oh.

Where's the music come from? Because I already did something. Cool. Okay. We're on. Yes. Keep going.

In this case, it tried to create something, and then it ran another terminal command. Keep going. That's fine.

Is it doing stuff for you? Do you see things popping up?

Where was the auto approve? I can't find the auto approve.

Oh, the auto approve?

It doesn't look like it's in the.

I'm going to go to Insiders settings. Insiders sorry. I'm not using Insiders. I guess I should be.

Oh, wait. Did we that was in the last version.

I have Insiders. I just haven't switched back there.

Okay. Okay.

I couldn't find it either for some reason, but if I the dropdown.

Okay.

It wasn't in non-Insiders though. Was it?

Let me let me check in VS Code Stable.

Here. Wait. It's not here.

Uh, show Stable Code. Yeah. That's my my problem of not being enough in Stable. I shouldn't run workshops.

There we go.

Wait. Why is that different than what came from Jesus?

So you found it in.

Yeah. Thank you so much. There's two ways to get to the same thing, but it's a different menu, but it looks the same.

Oh, it is.

Anyway. Yeah.

The settings.

There's settings from the bottom and settings from the top, and they're not the same.

Oh. Okay.

I'm happy to go figure that out later.

There's settings up here.

The one the one. I used the gear I used the gear setting from the bottom.

Yeah.

And it didn't work. But anyway.

The settings from the.

He he got the top bar if you hit.

Okay. Where's the other one?

Like the one in the.

Oh, in the other setting. Yeah. Yeah. I have customized my UI too much as well. Cool. Question.

Is there a way, uh, for this to run? I know it keeps telling me to run this main terminal, but like the cursor, it it gives you the capability to actually have it run.

Yes. That's that that's why we I think I should have said auto approve before so it's active for the chat session. But this is the setting I just showed is is the auto approve. So you all the continue buttons are basically gone, and it just auto approves.

Running for you actually said the same?

Yeah. Yeah. And for MCP tools, we actually do have dropdowns to allow always allow for session and always allow for workspace and or always allow. So there's some more fine-grained things. We're rolling out to more tools. But for yeah.

It's still prompting to continue when I click.

Yeah. I think it's the auto approve is not applied to the current session. That's I showed it too late. Should have done it the other way around.

Okay. We got some material design coming in. And this is where you need to get your coffee and just wait. Hopefully.

What's the contributor's?

That's a good idea. Let's do that. Okay. Uh, open your window.

Um, okay. What what prompt did we use? Create a Vite. So okay. Now do in the new window, I'm going to do auto approve first. Um, approve. Auto approve is already on. That's good. Um, this is another window now, and we're going to do the same thing.

And we're actually going to use material design. I guess and is still out there,right? And or fluid fluent. That's the Microsoft one. Let's see what that looks like. Again, it's going to prompt me for a folder. Um, Vibing at AIE 2.

And this is, I think, one of the key takeaways. Like trying out different ways to get to the same result is where Vibe Coding really shines. Is continue. It's just trying I I had really big success of just like what are different signup flows that we can create.

Like create three different versions of this design to explore what this could look like.

Okay. It installed. It's updated in the next in the in the one script. And that's where it gets confusing. If you have multiple open now at the same time, you got to figure out what's running what. And now this flow actually over here runs without any confirmation.

So we we set auto approve in a correct order. And it's now just creating Vite, installing installing fluent.

It noticed that it got the wrong fluent because there's a dependency. Now it's fixing those.

Can you show one more time? I can't do that auto approve setting.

Yes. So the way um, I I do command comma is the quick way. So I don't use menus. I should use menus. Um, so you go up here. I have my settings up here. Most probably have them on the other lower side.

And you go in here and go into settings. And this is what it should look like. Do your settings look like this? And then if you look for approve, sorry, can't type auto approve.

Do you see it?

It might be Insiders.

It's all blur. We're shipping on a monthly basis, and I thought we tweeted about it end of last month. So

so you found it?

Okay.

I guess the future I'm sorry. What else? To me, this is a little slightly scary because auto approve is like I'm just going to run it.

Yeah.

Where I'm not trying to compare products. I know they're different. I'm just used to I really want to use this. I want to stop paying for them. Um, but the challenge is that auto approve is like I'm just going to do this.

Yeah. Yeah.

Would you like me to run this? And then you say yes, and it does that work for you,right? So I'm looking for that in between. Butright now it's like quite binary.

Yeah.

It's going to either run it or it's not.

Yeah. The good thing, the person who owns the AI terminal integration just came back from paternity leave. So we're back on the game to work on it. But the um, there is definitely we've been looking at how to allow specific terminal commands, and that's how most tools do it.

But then if you think about the scary parts of chaining and running multiple commands in one command. So terminals are not as predictive as you would think and how you can easily allow list things. So if you're mostly thinking about the how to do itright,right?

Okay. Now we got two Vibe Coding sessions going on here. Hide this one. This is not. This is for tomorrow. This Vibing is happening. So you see it it creates an app TSX. It creates an app CSS. It also creates Copilot instructions.

Uh, who's been using Copilot instructions before? Yeah. So that's that's one way this out-of-the-box experience just does things for you. It uh, comes with instructions baked in, which is nice. It understands which stack you're using. That's mostly about it.

It already captured my design principles that I eloquently put in. Please make it look like Apple. And now it's actually there's a clean, minimal, intuitive interface, consistent UI, and everything else. Uh, it broke down the technical stack, even things I haven't mentioned like CSS and responsive design.

So it calls out some of the assumptions that AI will fill in if you give it a high-level task.

Okay. This is still working on the index CSS. And the other one is working on the HTML file. Yeah.

How do you create the instruction file?

It will just do it for you. It will usually do it as it creates the project. I haven't typed anything else so far in. We're just we're just vibing.

There should be a file called instructions.

It um, here it created one.

Or a page?

Let me see if it created one both.

I didn't see it.

Okay. There we go. Our first app is done. Hydration tractor. Stay healthy. Stay hydrated. Today's progress. We're on a quick add. 500 ml. And it went with metrics. Isn't that beautiful?

Just how I wanted it. Maybe it got my accent. I don't know. Um, we can do plus minus.

It's interesting how mine looks different.

Yeah, I know. I know. So does does everybody look as nice as mine? Do you already see yours? Okay.

Yeah. Yours better. Yeah. Yeah. That's a this like a very wide open Vibe Coding workshop. Um, I ran these a few times, and this is probably one of the nicer ones. So um, but it's also I like actually running these with different models and getting a sense of how good each model is at design and having design sense without me telling it how to do everything.

And Claude is definitely usually rocking the the icons. It got the colors really nicely. So uh, that's been great. This is a really nice app. So now next on because we're visual, like we haven't even checked the code.

We haven't read the CSS. We haven't looked at the TSX. Like is this doing which is doing functional programming? Like how does it handle state? I I don't I don't care. It works. So now you can actually do a new feature we landed.

You can now say

work visually. So I can now say this header up here. I don't know what it's called. Like whatever progress indicator. Um, let's make this

uh, more animated. Adding particles maybe.

That's good. So Mui pet paper. So this is all is this material design? This is material design. Yeah.

So did you copy the component name?

I didn't. I hit start down here. Uh, if you have the browser preview open. Did it open for y'all? Cool. So this is um, two point out two features. So one is in this flow at some point it basically started the task.

It's did npm run dev. And then the next step it did, it opened simple browser. And simple browser is this VS Code in browser preview we have. And it will just um, what we're injecting here, what just went away, is a little toggle you can now use to select specific elements that we then attach as a visual reference reference and as CSS into the current chat.

So if I scroll back down here, I see what's being attached. I actually click it. So everything you see under message is in the context. The element screenshot somehow didn't didn't make it through. Uh, but this one made it in.

And that's basically the the CSS description and HTML of the element we attached. So I didn't have to describe the element where it sits. It just did it for me. Okay.

Ran into snacks. We don't care about that. Let's check the other one. Ooh. Number two, fluent design. Uh, this is what it came up with. It's a little bit plain. Um, this is sad. Okay. At least it has a goal reach.

That's nice. Um, and it has recent entries. It makes similar assumptions on what we want. So feature-wise, it somehow got to the same conclusion. But design-wise, uh, this is definitely more corporate. Um, yeah. So that's the simplicity of Vibe Coding and using the new tool out of the box.

Uh, if you are an insider and you can disable new tool, it's easier to do like a single file HTML thing because new tool is definitely biased towards using npm and installing packages. So it always ends up a little bit more um, complex.

Do you work with do you work with insiders?

Yes. It's basically the team I work on. So I hi, I'm Harold. I work on VS Code.

Is it really cool to understand when you open up for new insiders, what are the new things that just change different? Like like a like a quick diff somehow. Like someone could show me like I have them both on my machine.

Right.

And I use them both, but then I like fall behind a couple days back. It's like what are the new things and why should we use it insiders or what's going?

Yes. So the the best way to stay on top of what's new is so we we do actuallyright now this week it's testing week and we're writing our release notes. So the release notes are usually capturing everything that's new.

But for insiders, it's hard because it's coming out every day. So it's hard to.

Where would you like an AI agent to summarize that?

That's a great idea. We're going to make an MCP server that summarizes what landed. Yeah. Yes. I like that. I like that. I like that.

Do what's new and then every day.

Let's let's do it for the next next demo.

The next Vibe Coding.

Next demo. Okay. So what do we have in our YOLO Vibing toolbox? We have the agent um, which sometimes is hard to find. Now you're all on the agent, so that's great. It's all about um, actually didn't show that, but I could have shown that.

It's different panel styles. So if I go back here, um, you can actually move this into the editor, which is nice. So some people like that. More space for your chat. Um, you can also

if I go back into my panel. Oh, uh, I moved into into a dropdown here and you can move your chats into here. So you can have multiple chats and they actually have names. So it's easy to go back and forth.

You can put them in parallel to your code. So you can use all the like window management things.

Can you run them in separate windows?

Yes. Wait, how did you know? Um, okay. Where do we go? In new window. There you go. Now you have a chat in its own window. You can put it on your own monitor. So feature success. And you can actually pin it on to be top.

So now if I run this and I can close this uh, oop, let's move that away first so I can accept this. This we all going to keep. Uh, close it and then close the other one. And now we have the output and we can just move our chat across and fully focus on the exponentials that are happening in this window.

Yeah. So that's one way you you can really manage the space how you want it. Uh, new workspace flow we showed. So it's really this optimized CLI first. Your question?

How's the legislation?

Okay. Okay. Yeah. And then voice dictation I haven't shown. Uh, who has tried voice dictation in GitHub Copilot already? Okay. Magic moment of

um,

add a dark mode please and maybe give it a cool name that works with a younger audience who needs to drink more or maybe my kids. Make it for kids. So a little more kids-friendly. Thank you. Bye.

Okay. So um, command I is actually a default shortcut. Um, it's it's a local model which is great for privacy and it's really fast. Uh, it's accurate and there's an option as well if you when you use command uh, voice input that it also reads back the text which is great for accessibility.

And yeah, by using just your voice, you can now finally don't put that coffee down and just keep keep vibing. Uh, there's a hey Copilot as well. I think we did at some point which I haven't used in a while.

Okay. Uh, I said all that. Keyboard shortcuts. There's a keyboard shortcut if you want to customize it to actually hold down while you talk and then let go. Uh, visual context I showed attaching. It's great for wireframes. The in-letter preview gets hot reload.

It just works. And you can attach the elements using that little send button. And then auto accept. I showed the auto proof tool. There's also an auto or tool the auto approve tool setting. Uh, there's also an auto accept after delay.

If you don't have that on, I love auto save. It's a great VS Code feature that's already working um, basically after delay or after focus change. It will just save it for you. And what I haven't showed, let's see if this this works here, is the undo button.

Is this still going? What's what's this one doing? I forgot. Um, oh, this is adding the particles. Cool. It still worked on that. Good Copilot.

Okay. Get stunning animations. That's great. That's what I wanted. Is it doing it now? So let's keep. It does animate. Nice. Particle. No. Look at bubbles. Okay. Um, I don't like it. Undo up here. Uh, those are basically our our checkpoints.

Uh, there's a new checkpoint UX coming, but because I have many people saying telling me, "Oh, you don't have checkpoints. I can't undo stuff." But if you already accepted stuff or if you want to go back to something like this is like these particles, but for for V0.

Oh, that was beautiful. I love it. Uh, we don't need that. Then you can also bring that back. Just need to see that again. That was really nice. Okay. So for people who don't like particles, we can now undo and it's now back to the original version.

So it it has stages for each of the of the work it did. You can easily go back and forth to see the before and after as well. Yeah. But in Vibe Code, you don't want to look at the code because we don't look at the output.

Okay. Um, that's the YOLO toolbox. And I think as I mentioned before, you want to try it out just to get into your of the AI. Like in my case, I mentioned I like getting a sense of how good AI is at design.

Like can I just give it wide task to explore a space and it'll like come up with something interesting or do you need to how detailed do I need to be? When does it make mistakes? If I give it a general task maybe about Java where it's uh, not as good at like what will it do?

Um, next one is known frameworks. We went with Vibe material design, things that are kind of off the shelf and haven't changed in a long time uh, in a large scale. So you want to use something that's popular and have has been consistent.

And lastly, we use it as a whiteboard. Um, we showed just attaching a visual element, change this, add some particles. Um, it's really about not being becoming too attached with whatever you're working on, but being um, able and willing to throw it out and start from scratch if things go wrong.

Structured Vibe Coding is this middle stage. Tries to balance the YOLO, the fun and chaos with a more structured approach. And there it's um, I think it's the biggest impact I see from talking to customers on like this is this is how Vibe Coding can work for us.

This is where we can bring somebody non-technical in, give them a good starter template that has a consistent tech stack that comes with clear instructions for the LM and how to work on it and keeps it in its actual guardrails and already brings in some custom tools that bring in expert domain knowledge or internal knowledge that you would need to work on this code base.

And that's really kind of YOLO on on guardrails. Uh, it's faster and gives you more consistent results. So you don't end up with something, oh, it used material design, but it should have used fluent or it should have added dark mode and should have been responsive.

All of that can already be baked into the instructions. So I see a lot of companies bring that into their bootstrapping for greenfield projects. So we have something and you can oftentimes it's a you go into a meeting and you have a product that looks already finished because you Vibe Coded it with your go-to stack and uses your internal design system.

So it already looks way more polished.

And the last piece, I think out of mainstream workloads is where YOLO by default will always bias towards whatever is like top of the training stack. With this one, you can then customize it further down to internal stacks, internal workloads, internal deployment infrastructure that makes it work better.

So let's do structured Vibe Coding. This is now the image as I explained is now has wireframes and uh, more charts. So that's what makes it more structured.

There we go. Let's open. So what we're going to do now, I think I'm going to push this. Oh, it's just still running.

Let's see if this runs. I did create this Vibe Coding. So I can do it with another one that I can share. Just look at this one and I'm going to push it to GitHub. This one's pretty really fine.

Um, cool. Front end vibes. Perfect. Um,

this is all vibes. So we're going to make this live. Um, this is commit and then allright. Yeah. Oh, yeah. Miss this one. Who has been using this commit button here? So Copilot will write your commit message.

Done. This looks good. And now sync changes and I'll share the repository.

This repository might still be an old name. Let me see.

Would it be a tech for us to pull this repository?

Yeah, probably. Uh, let me just check where it sits because I forgot where it sits.

Oh, this is perfect. Sleep vibes. This is one of my Vibe exercises. Okay.

What's the command for browser window in VS Code?

For browser window.

Like a viewing your.

Which the.

It just opens sometimes.

Oh, yes. Yes. Um, yes. The simple browser. So we go into simple browser show. That's it.

Okay. The repo we're pulling down is whoop. Whoa. This one. Sleep vibes.

It doesn't have a def container, but it's only Node.js. So it shouldn't be too gnarly.

Do we want to get the agent to do some special magic and pull this or should we just.

Should I try? Uh, it's a it's a Git clone. You can ask the agent to Git clone it for you. Yes.

I'm curious if there was a special tool that would be more than that.

Oh, sorry. Did you Daryl? That's me.

Yes.

Okay. And then npm install on it and then yeah.

It would have prepared better. It would have been a code space and a def container and you just click open in code spaces and things work. Um, come to my next show and then we'll get that fixed.

You got it.

Yes.

Let me actually just open the code space to see. Now I'm curious if it just works.

Anybody's been using code spaces on GitHub? Not many. Okay. Occasionally.

The complexity lies in all the different versions of the feature paradigm between insiders, the regular version, all of the plugins and some of them don't work in.

Autobug. Yeah. Yeah.

If everything just worked the same it would be amazing.

Yes. It mostly does,right? But mostlyright.

It's 95% there.

Yeah.

But it's the 5% where when something doesn't work you just go back to the other tool.

Yeah.

So VS Code and other code is offering me to reopen it in a container because of the dev container configuration.

You could try. I haven't I'm actually not running it in container, but if you want to.

That's allright.

The container is just a Node.js one and it should work too. I did add a container. See? I Vibe Coded my container too.

Wait. I can now check. So if you ever wonder what you did on a project, um, so this is where I created my container. This is where I just asked uh, GitHub Copilot to update my def container. Just look at my code base and update my def container.

So I did a good job here. Should have maybe.

In container.

Remembered that I did that as well.

Okay. If if you're ready meanwhile while you clone while you npm install, anybody got it working already? Still. Okay. Cool. I'll give the tour of what we have. So one is we again start with good Copilot instructions and they live in .github/copilot-instructions.md.

It's a Markdown file that's included with all your agent requests, all your chat requests, all your inline chat requests. It just Copilot basically gives a grounding foundation knowledge about your code base. And sometimes they feel a bit repetitive like um, depending on I saw some demos of like use basically it repeats linting rules that you expect the AI to just follow anyways.

But I I like um, my go-to is just a one-liner on what what's your stack. That's that's a good starting point. Just point it to what frameworks, what version. And that's one way to just keep it on rails with what what it uses.

Question?

I was experimenting with this and figuring out what would be the best way to have that be a standard that gets included in, but like let people not not mess with that. Like give people some coding structure.

Yeah.

If you have a whole team of Vibe Coders, but you probably don't want to touch them that.

So this is in your repo. So I think it's a good team exercise to iterate on it. Like it shouldn't be a stale document. You can put this in your user settings as well.

You probably don't want it with each app. You probably want it as a different repo,right? Like a.

You could do that.

A set of a way that you code and then each app codes. I've been thinking about that too. I'm really thinking about it.

Yes. So I've been trying to convince on my peers on a GitHub site this should be an organizational setting that people can set easier on like a organizational level and like something as a as a team you can select which ones you want to use.

So we're working on that discovery sharing.

Is it always one file or can you opt in for like because you have different languages, different settings, different things you're building. Like a system as an API.

Good question. Yeah. So we have this one now. These these are new instructions. So they these are these become rather monolith and large and unwieldy. And now uh, just point out here before I go to new ones, I do also guide which tools to use.

I do have my first MCPs in here and I already tell it for front end Q&A and review use the browser tools that come from Playwright research. I use Perplexity. I have uh, Context 7 in here which has library docs and it keeps using this ID tool to look up IDs, but I just gave it these are the IDs you should use.

Don't don't use the other tool. So there's ways you can already guide it to specific tools you want it to apply when needed. Um, the rest is just syntax formatting, optimizations, um, key conventions. Yeah. And then the other format we have is .github/instructions/name.instructions.md.

And those have this front matter syntax that's becoming more popular for rules of what what files it should apply to. So they start to be scoped with a GLOP pattern. And thenright now they're limited to being applied. You actually have to have the file in context.

So a TypeScript file would only be applied if I actually do have a TypeScript file in here or I do have one open and then I enable this context and then it would be applied. But if I only have thisright now, which means this isn't included, it wouldn't actually apply the rule.

We're we're fixing that and it's going to be more working as expected probably. Um, but that's that'sright now. That's like the biggest question I get. Like it didn't include my rule becauseright now it really wants to see that file.

Um, yeah. So those those are new. Uh, those shipped I think the last version. So they should be also in stable and we're actively working on those. So, and then the the new new thing is plans uh, or prompts.

And then we have the first kind of reusable tasks for as a team, how do you think about engraining like, oh, we now have finally a way to tell GitHub Copilot to write tests and your AI champion in the team handcraft this this perfect prompt which one-shots your tests consistently.

And now everybody shares it in Slack, copies it around. Once you run aright test, you go back to Slack, copy it back. And that's what you want to use prompts for. You can finally put these prompts into a place where they can just be used by everybody.

And how can they be used? So I showed these can actually be be attached. So you can also go in here and attach instructions. So you can you can do it manually too. So that's one way. But I can also now go in here.

I'm in the chat window and hit slash and I can now actually run user prompts that are my own that I create for myself and I can use my plan and spec prompt you see over here on the left.

Can you make them custom?

Yes. These are custom ones. So the ones I have here, these are already custom in the workspace. And then the other ones I don't have I'm not showing. So I think we do. Wait, there's a new menu.

A ridiculous name like Dance or something.

Yes. Yeah. Yeah. You can name them free. So let's make it here. Um, so these this one actually just landed yesterday because insiders we can now finally have an entry point because everybody kept asking how do I create prompts and I have to tell them which command to find it in.

So this is the new prompt configuration file and I have some already here. So as you mentioned, like this is one that's interesting for um, if I open this one, this is like defining how I want to write custom instructions.

So whenever I'm in a new project that doesn't have custom instructions yet, I do run this prompt to to bootstrap them for me. And yes, there should be a prompt sharing website where you can find these amazing prompts that I create.

And next week we're going to.

So each prompt is like a separate it's like separate from each other.

Yes. So that's the main difference between instructions. Instructions are you can have multiple. If you work on for example, if you have one for TypeScript and one for your front end folder, they do combine because there's multiple instructions that hopefully don't conflict with each other.

Um, but they they allow you to be attaching multiple instructions and they're really more about code. Whereas prompts are basically uh, easy ways to inject something in this prompt field and they stay in the conversation, but they um, they're mostly around a task and maybe giving the AI something like specific to do.

Instructions you wouldn't necessarily give it like what to do, but more how to do it.

What about if you wanted to, for example, teach it to um, always do TDD when it's writing code? That where was that?

Yeah. Yeah. So TDD um, good point. Uh, that would be a good way to use custom modes. So if we go into here custom uh, sorry. So this is only insiders because it just landed. Sorry. You can't follow along or if you're not insiders.

So custom modes will show up um, in the dropdown. So this is like going to now it just went into the menu, created a custom mode, and now I can pick where it shows up. So it got GitHub/chat modes which put it into the repository or I just want to keep it for myself.

So good pattern if you just want to experiment, put it in your user folder. If you want to make everybody's life better in your team and you have high confidence that your mode does that, then you put it into the project.

And then uh, we're going to do name this one TDD

and then we're going to

ask AI to fill it in,right? Uh, prompt

um, expert expert prompt uh, for wait, I'm just typing.

Hi there, AI. We need a prompt that enforces test-driven development uh, for GitHub Copilot. So it should probably first make sure it understands the problem, then write tests first and only after tests are done maybe get confirmation from the user to then write implementation and then keep running the tests against implementation.

Cool. Thanks. It's important.

I did not I didn't say I'm not worried because I didn't actually activate my file as context. Let's see if it

I think it should have a tool to just create modes for you if I ask it. So it's going to going to make that an MCP server next. Okay. Oh, it got it. Okay. Wonderful. So we have a test-driven development assistance.

This is Vibe Code. We don't want to read the code. Test development assistant core principles understand

red. Write failing tests first. Beautiful. I wouldn't have not green. Wow. It does follow refactor improve code quality strict rules no tests. Beautiful. Beautiful. So this is our new TDD mode. Um.

Sonnet 4.

We have Sonnet 4. Looks pretty good. It has emojis example. Should we try it out?

Yeah.

Okay. We have.

Framework or no? Did it say didn't it say there was a framework in that?

Um.

If you scroll back up there was something that said try to say framework.

These are framework. It's framework independence looks like.

Was it? I thought there was something.

Oh, just yeah. I think it does some. Yeah.

Oh, okay. So it has itright there.

It does make some stuff up. Yeah. Wouldn't need to do that. I can text this out. I already have my so

um, cool. Test drive the design. Wonderful. Okay, let's do it. So we have this project which doesn't do anything if you just run this. Um, npm run dev probably doesn't I think I already did this before. So this looks just a basically plain landing page.

So let's do the um, what feature do we have? I want a dashboard for GitHub users. Just use mock data because I don't want to wire it up to GitHub. So I want to have maybe some interesting contribution metrics and but first actually let's make a plan.

Meanwhile, while I type this in, let's stop it. Um, TDD.

So dashboard for GitHub. Um, don't worry up. So now again we give it a very broad task, but we can now put it into TDD mode

which is our new amazing test and development mode that follows all the best practices. I assume that the AI knows about TDD. And let's see.

So you wait, that's a you created a new mode there.

Yeah. Because so we went previously went into configure chat modes. We created a new mode. This mode is now enforcing the technique. We can actually in a mode you can say which tools it's supposed to use. Sorry.

Is the mode that this kind of confuses me is the mode an agent mode but based on another Markdown file? So it's a it's a custom agent mode.

Yes. Yeah.

When is that going to be live?

Uh, it's an insider now. It's because it's going to ship next week on on the 11th. Yeah.

How often does insider ship?

Insider ships daily, but it only releases monthly. Yeah. And we're one one week late uh, because there was a short short week. So we adjusted our schedule.

Does it need to have to go to the test code?

Yes. Um, so for most of you this menu will just have these entries and the in insiders only if you look for modes in the command palette. So command palette you can also click up here show run commands and then modes and that's that's the place.

During development, I ask you to deploy the application.

Yeah.

There's errors in the console. There's got to be a way for instead of me copying and pasting that error into.

Yeah. Are they in in the output or in the problems view?

It's in the output. It's in the console,right? So just think about it.

Okay. Yeah. Yeah.

Like this one. I I purposely did something wrong and I wanted to see like and the error code is in the chat in the in the terminal.

Yeah.

In order for and it goes tell me if you need anything else.

Yeah.

Well, I want you to look at my I want you to look at my terminal when there's an error,right? And andright now there's only copy and paste.

Yeah.

Is there a mode where it just constantly is looking at the terminal?

It if it runs the commands itself, it will start looking at the terminal. So the easiest way if you if you run the deployment and the scripts itself through Copilot itself. But otherwise, there's also context actually. If you look here, we have um the.

My terminal I'll open it.

There is actually terminal last there's terminal last command which includes the output as well in terms selection. Now, if you ask me why they're not in the add context, I I couldn't tell youright now. Yeah.

But yeah, that's working. I think I did this thing wrong though. TDD. Let's just see.

Oh, the tools. It configured it made up tools. So it did that's that's the part it made up. This actually those are not theright tools. Um, this is why it didn't do anything was when I it just basically acted like chat and gave me the code because all the tools it tried were

uh, it didn't have anyright access. So let's try this again.

So that's probably a good good good point to good thing to point out. So in now prompts as well. Let me open plan prompt. So this one can actually now set tools and if you just make a tools entry here to tools you can now actually click here and say which tools.

In this case, this is a planning prompt. So mostly you probably want it to look at perplexity to come up with anything it needs to find on the internet. Uh, I can select that. So that's the way you can now have tools constrained for a specific prompt

which which always helps with kind of higher quality because if you have many tools which as you install more MCP servers you always have this tooling explosion and they might solve all different problems you're having throughout the day but now you can configure it more specifically for domain.

And also insiders only we have tool

groups uh, tool sets actually we call them. So tool sets I can how do I get here? Um, down here in the tool dropdown configure tool sets and add more tools. This I think sends you to add server.

But configure tool sets open this opens this one here and that's only.

MCP servers or anything?

Anything. Both built-in and MCP. Actually, a lot of the tools you see here, we cleaned this list up. If you use insiders, you see them. Then these are actually tool sets already. So we use tool sets internally because edit files has multiple ways to edit files.

We give the AI a few ways. Uh, code base search has grap has file search has different searches as well depending on what you're looking for. So all of these actually are tool sets in our own backend and we expose this now as something you can create yourself.

So my research tool, for example, has the perplexity tool to ask deep research questions and it also has fetch.

You already dumped 100 MCP SSE services?

No, I didn't. No. We can show.

Will you dump 100?

Yes, I can. Wouldn't be a talk without MCP. It's just

also there's a talk tomorrow about MCP. A whole talk track where I'll be talking about MCP if I can finish my slides. Um, okay. There we go. So let's talk about MCP. This is doing

something. Let's see. Not understanding requirement. It created mock data. Red face writing tests. It wrote tests. It oh, it found that there's a no package library. That's that's sad. Um, and it created the test utility and then it tried to run the tests.

Um, and then it asked to proceed. So that's cool. So it did the first stage of that mode. Um, but only go too deep. But yeah, that's modes. TDD will ask now for because we ask it to ask actually for confirmation.

So that's why it's now pausing. It wrote tests. Uh, they're all red. So that's good. Okay. We accept it and let's go into MCP. So MCP servers. Who has already MCP servers set up in their VS Code?

Two. Okay. Um, one way to get MCP servers is editing JSON and that's really um, there's a few other ways, but let me show you another way. Um, Playwright MCP. Who's been using Playwright MCP? It's probably one of the coolest ones.

Um, so Playwright MCP is a browser testing framework and allows people to access the browser locally and just take screenshots, run websites, get accessibility audits, a whole bunch of utility in there. And how to get it for VS Code, there's a JSON blob that you can all ignore and just hit install server.

So install server is just a VS Code protocol that we use to just wire things up into VS Code. You see the same if you go to the extensions marketplace for VS Code, you can hit install extension that powers the same process.

So now I can inst um, don't hit show confirmation. We need to move this down. But install server actually puts this now into my user settings. And as you can see, you can have MCP servers both for yourself and I have the one for GitHub and for Jestpad, which is a cool one.

I can recommend this one. And then yeah, playright now. So you can already see how many um, tools are provided. You can see if anything fails, you can get to the output. If there would be configuration, I can show that here.

There's actually Jestpad needs a GitHub token. So it's a local MCP. And what have you ever seen that one? So yay. No tokens in my configuration. So you can use inputs in VS Code configuration both in the MCP.json and your settings.

And inputs you might have already seen those in task.json is how you configure your tests and your build steps in VS Code. It's the same system. Um, they're defined up here. So inputs are just an ID, a type description, a default value, and the password true means it's encrypted at rest after you put it in.

So it doesn't ask me more.

That's going to prompt you then?

It would if it but this this shows basically that it already has a token. But if you enter it the first time and we can actually try it if I.

Where does it actually keep?

Uh, in VS Code's key storage. So if I on Mac it actually uses keychain.

Oh, okay.

Yeah. So.

What other MCP servers do you use?

Yeah. So I can show um, Jestpad is fun. It's done by actually um, it's on your GitHub. So jespad.dev jespad.mcp. That's mostly I'm going to show it off tomorrow as well my talk. Um, but it's a fun one that uses uh, JIS as a knowledge base and also for prompts.

So we like this one. Uh, it adopts a ton of like recent MC MCP stuff, but I think the main ones we usually see is GitHub MCP server.

Jestpad. Just.

From.

Lost in tangent.

Okay. Yeah.

Yeah. So this uh, if you just want to play around with it with a really well-done MCP server, then that's that's one. Um, not saying the GitHub one isn't as good, but it's um,

there's a lot more coming here as well. So they're all in like really active development to figure out what the best way for MCP is. Yeah.

That tool just a couple steps from deploying to 3,000. So this is using NPX. If we're using SSE like let's say Python.

Yeah.

So so how how first where did the you got the MCP JSON?

Yeah.

Okay. Where how did you connect to the let's say I have a Python custom Python MCP server on SSE port 1234.

Yeah.

Which is exactly what I do. I don't understand how you just did that.

Okay.

How do I open up.

To SSE?

Yeah.

Yeah. Okay.

Right now it's using NPX.

Okay. And.

Example.

Right. And in your case that server is already running. So how how to hook up.

Running already.

Okay. Cool. Yeah. Let's do that. So SSE um, same way basically. So what just finding my sorry I mixed it up. So go back to my MCP.json and down here.

Tell me what this what is this per what space.

Yes. So to click clarification MCP.json sits in .vs coderight now and it's per workspace and that's shared across everybody. So hopefully it puts stuff either you work on it alone and it's just for you or everybody is happy to have those MCP servers.

Oh, that's awesome. That was a little confusing.

Yeah.

Per workspace and now NPX is using the SSE for.

Yeah. Yeah. Okay. And then the what's now if you hit add server you will find what you're looking for.

And then from add server you can hit down on HTP. So we actually do support both SSE which is actually deprecated and streamable HTP which is uh, the the new fangled easier to scale, better for your cloud.

So is it deprecated already?

It's no longer in the spec. Yes. And we do fall back to it on a client side. But it's it's the SSE is really hard on on hosting,right? Because they have these long running connections.

Long pole.

Long pole. Yeah. Yeah. So

yeah. So that's that's where you put in your MCP SSE server. And if you want to do it manually, it's really just um,

you get a nice autocomplete too. So if you if you pick a name,

example, and um, this would be the type would be not studio. Actually, it's HTP. Yeah. You already would would use HTP and then it would already yell at you that you don't have URL. So I'm going to put URL.

So this this is how uh, everything is by default as STDO. Once you have a URL, I think I can take this out. Yeah. So it would be just that entry.

So you want to go back and say are you just starting your chat in S instead of agent and then it's leveraging the MCP server?

Yes. So to get yes. Um, so many demos I see people hit start here as well just to see that it's working. It's a nice configuration that just make sure it's working. We actually do cache the tools once we saw them the first time.

So how MCP works is that on the first initialization from the client to the server, it shares its tools back. And that's what you see here, the one tool. So if you would do itright, you would never know the tools unless you start the server.

We actually cache them. So you don't have to we don't have to start all these servers proactively once you open Copilot just to get the tools.

Okay. I just make sure. So that plus symbol up top chat, you would be just it would it now recognize the MCP server.

Yes. Yeah.

You can actually get out of agent mode,right? You go into.

No. Actually, you want to be in agent mode. Agent mode uh, ask mode will not run MCPs for you. Uh, you can you can go uh, because ask mode is not uh, it's not actually there's there's no function calling inherently,right?

The func ask mode is really this traditional ask ChatGPT question and we'll answer based on its training there or its context. Um.

Yes.

So there is also an ask mode you don't have tools. You would see it. Um, but I can do actually does this still work? Um,

let's do that one quickly. I think we actually do this still. Yes. So we're actually blurring the line a bit now. So if you do. That's not working. Okay.

Yeah. But if you actually reference specific tools in ask mode, it will invoke them for you. But by default, the the way where you want to execute tools is in agent mode.

I am these are models. You're using uh, for.

Yeah.

For models. Um, I'm just going to find configuration as to like how does it know that that 40 model or the the GPT model? What subscription?

Uh, they're all coming through GitHub Copilot. So they're all using your paid uh, you can add your own models. Anybody has tried it managing your model? So um, so I have Gamma 3 through Ollama which runs locally and I have open routers perplexity R1 uh, which is actually a fine-tuned model from of DeepSeek R1 from perplexity.

So if you haven't tried it yet uh, basically go into the model picker and hit manage models and then we we can actually custom configure your own API keys from Anthropic, Azure, Cerebras, Gemini, Grok, all of these um, Ollama is the local one.

So if you have a beefy M4 Pro, I'm still sad how many models I can actually run on this. But but eventually it's going to be small powerful models. Um, that makes sense. So.

Is there a reason when I do that I don't see cloud for Opus and show up and only shows like three five because I don't know what that is.

Uh, it might be because of your Anthropic tier,right? That's um, or is it.

I have Sonnet 4.

You have Sonnet 4? Oh, oh, uh, the other one. Yeah, you might be in agent mode. So we do actually have filter them down. So that's an ongoing improvement we're doing. That's why it's not it'sright now a preview feature only because we're still having to connect correctly wire up which model allows tool calling.

So there's some some every provider has different indicators of how tool calling works and that's that's one of the matching things we're doingright now. So you might not see it because it's not on our our list yet.

The the robustity of when you go into the profiling, let's say you have like four tools, is is it verbose enough to say like what tool is it calling within MCP?

Yes. Yeah. You you you will see that. So if I do just um, pull one example here. Uh, so Jespad. So A, you can disable them. If you already if you want to be faster, you can do command down and just go through.

So context 7 is the one I want to keep. Player I can disableright now. But once you start using them.

Well, how do you do.

This is command up and down. The power user way of navigating those. So these are all built-in MCP servers. And once you start, you can actually now if I want to be very explicit and I know which tools I want, I can use my tool sets or I can mention specific tools that are um, in in my list.

But I can also now just go in and say what what do we want to do here? Um, research uh, GitHub metrics. Let's actually use the research one because we created it. Sounds better when I use research here.

Use it in a sentence

for productivity.

And now what happens now is uh, this one has now is an agent mode. We have the research group set or tool set. So we'll either use perplexity or fetch.

And one of my perplexity keys actually outdated because it failed before. Let's see. Okay. So you see I already actually proved this before. So you see A that it runs the server and actually click the server to see where it comes from.

Um, if it would have not auto auto proved this because auto proof is still on from our previous session, you can actually go in here and edit what it's sending. Um, which now doesn't make sense because it's already sent.

And then it writes up uh, what it found in this case.

Does that underscore ask have anything to do which mode it's in?

Uh, no. That's just the odd name for the perplexity tool.

Okay. It happens to coincide with.

Yes. Yeah. Yeah. It's just um, the verb should be before. So it's their their naming. Yeah.

Yeah. So that's now it ran two. It actually did a follow-up query as well and explained it. And I could put this into a spec as well. It's actually I did this before. So I wrote a spec using for community dashboard.

So I did the research using perplexity and then ask it to write a spec from it using a little query I have here for the spec. Um, so that's one way you can quickly get things done. And just to point out this one, it's uh, pointing it to the spec.

So these are actually resolved by the AI. So if you point it to specific files, we do actually validate those as well. So if you get them wrong, I think they're underlined. Um, you can also click them. So you get all the markdown goodies.

Um, and then you just ask it to write on the spec, do nothing else. Uh, use perplexity to look up stuff. Don't lose details. Keep updating the spec. So that's that's one way to work on specs. That's probably more more tools we'll we're going to add here.

Yeah. So that's MCP. Any other MCP questions?

Uh, per MCP. So the MCP itself doesn't run anything except when you support sampling, which we do on insiders. Sneak preview for tomorrow. So but yeah. But if you use sampling, actually, I guess I have to explain sampling.

So sampling is a way for MCP to reach back out from the server to the client to use the LLM on the client. And it's you can often think the best use cases are to summarize. Use cases are if you want to um, reduce the amount of tokens you send to the back to the client to explain something.

So there's a few ways. But um, overall there's not enough integration of sampling. Um, but so we're the first ones to to get it out there because we already have the LLM exposed. So that's cool.

Just have one

correction. So like it is more like some of these tools just run better on certain LLMs. Is there a way is there a a non-deterministic way.

Yeah. To pick the model kind of?

No. To pick the DARM tool. So you you you clearly have multiple tools here.

Yeah.

And Claude is deciding what tool to pick.

Yeah.

And you can prompt your way to do it, but it's still highly non-deterministic.

Right. So.

Is there any way to to remove that?

Yeah. So what I recommend is A, in in your modes, boil down the tools to what you actually need. So reducing the tools manually, either deterministically already in in your prompt. So this prompt could have tools for like what it should actually do,right?

That would be one way. Then I can configure what I actually want to have here. Like this should be only doing perplexity because it it needs to do research and that's all it should do.

With custom modes, that would be so much better.

Custom mode. Yeah.

Different pockets of like code like this, code like this. And you then you kind of pick your pockets of tools.

Yes. So custom mode is one way. And then the other one is you can actually mention specific tools. So if you go in here into add context and then you can actually point it to specific tools. So you're not doing like the look up things on GitHub and you try to find theright verbiage that it gets theright tool.

You can just actually mention the tool of it should um, for example, resolve library. So so you can just add these here and then it will be handed to the AI of like these are the tools the user wants to use.

Yeah. But you can use select you see like you deterministically select the bad tool.

It's it's still.

Yeah. Application.

Right.

No one no end user is not going to say here, they don't care what tools. The end user is working with the chatbot. What tool it needs to use.

Yeah.

Still non-deterministic.

Yes. Yes. Tool calling is inherently always even in this case, we're telling the AI you should use it, but it might not use it. Yeah.

Yeah. But there's if you do it in the GCP session, it was like yeah, this could be determined. It's not. There's no finite tool there. So I'm just trying to understand if there's.

Yeah.

Something here that can drive up. I like the instructions.

Yeah.

But I think you're closer to that. It's still not.

Yeah. So my timer's down to zero. Maybe just go back to the slides to wrap it up.

You have more.

You have more,right? Um, Vibe Coding. So we showed workspace instructions. We showed dynamic instructions which only apply to parts of the tool set. Uh, we showed custom tools and we'll playwright, deep research. Uh, I haven't showed using web docs.

Like actually one of my favorites, just point it to an existing repo and say, "Read this repo if you have questions." MCP works great for that. When I work in MCP server, I just tell it, "Look in the TypeScript SDK server for model context protocol if you have questions."

Because we have cross repo search, it just works. Um, the agent actually has access to problems and tasks. So if you have tasks set up and you have linting set up, things will just work. So make sure those are set up in your template.

Um, generate commits I showed and then fine-grained review. You can pause it anytime. If it asks you questions, you can always type something in and keep steering it intoright direction. And you can trust read only in specific tools.

And I showed you also editing. So yeah, uh, instructions, keep refining them as it makes mistakes. One of the key ones is commit often. I didn't show commits now, but anytime you have a working state, just make sure you commit it so AI can continue making mistakes and be creative.

And then last one, there's a clear pause button in the lower end. So if AI goes off and you're like, "What is it doing?" Like, is it doing theright thing? Just just pause and review. And that's possible as well.

Uh, I showed a bunch of this for spectrum development already, but it's really about having a spec, having a like and done a plan and doing more custom prompts and tools, which I showed. Um, showed user prompts. Um, there's more MCPs for database access and logging and project tracking like the GitHub MCP.

And there's also access to actually tests and do debugging within the agent as well. So if you ask it to test during development like we did, it will actually start running the tests if they're set up in VS Code correctly.

And then we talked briefly about models as well. So if you want to use O3 for any of the cool stuff, the the deeper thinking, you can do that as well.

Um, spec driven is really about focusing on the spec. And I think a great way to do that is just create the spec from all the conversations you had about the spec. So one way, if you have a transcript from a meeting about the project you want to do, just feed that in and make sure you call out what the final decision is.

It's a great way to have meetings, but it's also a great way to not write the spec yourself in the end.

Are there any tools to determine whether a spec is good or not? Like on the opposite side of that, like it's writing a spec or project requirements document.

Right.

Is like a.

I would ask AI.

On uh, on Twitter, who really has a lot to say about this. He says what he does is he has a generated spec and then he has a critique to spec.

Yeah.

And say what are things that it's missing and how could it be better and stuff like that. And he basically argues with the LLM about the spec until he gets it to a state that he wants.

Yes. Arguing with AI is one great way. So if I I focus on one run prompt um, or prompt is where's one critique idea is one I like of just ask me three questions about my idea,right? Just have AI go into thinking mode.

Like what would you ask somebody for feedback? Um, and have it critically analyze your stuff. So these prompts, like those are basically the next level of prompt crafting where you don't just ask it to code, but pull it in as a thought partner, as a design partner, as somebody who can poke um, holes in your I the along the way.

So yeah. Um, front-loading review steps is.

Yeah.

Yeah. Good.

Is there any similar is there anything with a similar mode?

A plan you you now can create one. So we haven't don't have a plan mode built in, but we also know ask edit agent will not be there forever. So it's a like series of evolutions. RU has way more modes that you can customize.

Right.

So I think we want to allow developers to create their own. Um, because I even see very few demos of people in CLI using plan. They just give it a thing and then it runs,right?

Yeah. Because I find like when I was using CLI, it was so it was fly out, but it was so much better using plan.

Yeah.

I could have assumed it would become a general pattern, but maybe not.

It is. Yeah. I mean planning in in Vibe Coding, you even do planning and then writing the implementation plan. So you went spend way more time on that initial just what and how we're doing things and then you let it implement.

So that would be even plan write the implementation plan or like write the spec, write the plan, and then implement. So you would even have three modes if you do it correctly. And then yeah. So that's already the last one.

So takeaways, you got to experiment. You got to figure out what works for you. Like at what point can you just give it a task and it runs with it. At one point you want to give it a task and write a spec first and then implement.

Um, keep giving it feedback and iterate. So never just accept a bad answer and then really work on your process. Like what what works best for you, what works best for your team, and use modes and prompts and instructions to ingrain that.

Um, there's some bonus mistakes you can screenshot.

I'll make them I'll make them all.

Yeah, please. One more. There's one more. One more screenshot.

You can end up like Emacs where it has major modes and then you can have multiple minor modes.

Minor modes. Yes. That's how you end up with prompts and custom modesright now. So I got to clean this up too. And then yeah, lastly, I think there is there is a sweet spot for how you define your code bases for AI.

So you want to have well-structured self-explaining code. You want to have the instructions set up. You want to have examples in your instructions. I want to keep instructions updated. So that's it.

That was my unplanned workshop. Thank you for coming.

That's really.

It was kind of planned. I learned about Monday.