🔴Live Vibe Coding 3 Web Apps: AzAnything.dev | Beep.Buzz | Eco3D.Shop (Next.js Supabase Claude Code)

本站添加: 2026-05-23

[00:00:00]Hello anyone. This is anything. You can call me AZ. I go live every few days to work on my three uh web app projects.

[00:00:08]This is Eco3D that you're looking at now. I'm making a procedural animation to replicate how lizards walk so that we can animate these 3D models and then you can actually print these dragons so you can get one yourself. That'll be the selling point to this app once it's finished. The animation has been really tricky. So that's what I'll be working on this stream. Um, we also have as anything.dev. This is a vibe code coaching platform. So, I've done a lot of work on this uh since last time.

[00:00:36]We're just going to continue extending it. All of the introduction getting started chat should now be working. So, now we're going to do a scan. So, the user will connect their repository and then they will scan it. And at the start of each chat, the scan will basically reveal the options of what can be added or improved or fixed. and then uh we can sort of guide along there. Hey, no one here. How you doing? Welcome to the chat. Thanks for joining in. Um hope you like the additions. I've added like some little dudes along the bottom of the screen and stuff. Bit of background music. So let me know if uh if you have any feedback, but yeah, same three apps as usual. Um so those two, we've got as anything.dev and eco3d. And we're also doing uh beep.buzz. So I'm just doing uh step by step with beep.buzz. I have the overall design, the game design in mind.

[00:01:29]It's going to be a cooperative game where you need the other player in order to survive. You actually need to give them fruit and receive fruit from them and then that's how you build things and you hire gnomes and uh rabbits come to attack your trees and you grow trees by doing uh by by sending Morse code.

[00:01:48]Eventually, I'll have communication between these two players, but for now, I'm just focused on the core gameplay loop. Uh, the one that I'd like to start off with, I think, is Eco3D. So, let's get started with that. I have recently updated all the documentation for this.

[00:02:05]Um, did you use Stitch MCP? I don't actually use Stitch MCP. I like to build as I um as I do it. So, I'd like to design as I build is what I'm trying to say. um because if I sort of use an MCP to do the initial designing phase, then that means that I need to build it again when I put it into a real react component. So I just I just build it straight to React. Um and sometimes I might do HTML, but usually I just build the component directly because then that prevents uh any potential issues that I have with like compatibility with the theming components that I'm used that I'm using um and the shared components and stuff like that. So, I just I like to prevent double handling just by building as I design. Um, but that's a personal preference. I also think it's because I'm a full stack dev. Um, but if you're specialized in Python, then it could be a little bit different. Okay, so I'm just going to tell C to take a look at the documentation.

[00:03:05]Um, documentation. I've recently recently removed all of the animation from the um animation step in the studio.

[00:03:21]Um I want to start building in want to start incrementally building incrementally building the uh animation as described in the docs in the docs.

[00:03:36]it's described in the docs.

[00:03:39]Take a look uh and give me a concise overview. Take a look at the docs and the relevant files. Relevant files and give me a concise overview of what we need to do of the road map and yeah the road map to incrementally build this thing.

[00:04:04]uh the road map and tell me where to start.

[00:04:09]So, I should have pretty good documentation for this. Now, I was originally using a uh incorrect model, and the model was actually based on a two-legged creature from a game called Triox just cuz it had really good uh explanations for how it worked, but it was just not what we needed. Instead, I'm doing a uh lizard locomotion. So, I'm actually going to open up a different tab in Eco3D and ask for it to provide um just some reference stuff so I can actually show you guys uh look at the animation documentation uh documentation and give me a link to a YouTube video that demonstrates uh demonstrates the kind of biological biological ical animation locomotion animation we're going for just so I can like visualize it. Um, no one here says I use it because for me it's pretty hard to finalize design. I mean I can't choose what goes where. It looks better and will be pretty easy to Yeah, for sure. U I'm on Windows um not on Linux. I keep, you know, thinking I probably should be on Linux, but I really like games. I I like to play games on this PC and um so I just I play them on Steam, but uh I I do see a lot of draw to to use Linux. I probably should, especially if I can still play most of the games that I although I don't play that often, but uh how about yourself? Are you a Linux user or a Windows or Mac? I used to be Mac and so I've had to actually use Windows Switcher. You'll see me swapping between windows like this and then in between windows within the same project uh program like this. That's actually from Mac. So I use a a program called window switcher to emulate how Mac handles that. Um all right. So let's try out this. I want to show you the kind of animation that we're going for. I haven't seen this yet. So ah see that's the original one.

[00:06:21]I'm looking for an example of the uh evolutionary type of locomotion uh that is explained in the docs.

[00:06:37]The one you linked is the other kind of animation uh that we're also using.

[00:06:48]See if it can figure it out. I'll I'll just try a couple of times then I'll move on. Uh you must use Linux uh system because it gives a faster environment for dev. The speed hasn't isn't too much of an issue. In fact, I was on a Mac that was much less powerful than this PC and that did most of the work. The only real issue is u multiple agentic workflows. Um thank you for whoever clicked that they liked, whoever voted that they liked and subscribed. I really appreciate that. Yeah, in terms of speed, I've got enough RAM and that means I can run enough agents in parallel and that's pretty much that's pretty much all I need. Uh, we also got another vote for as anything.dev vibe Code Coaching. That is this one. So, I'll actually this one here. So, I'll actually catch you up on what I've been up to here. So, we're in we're signed in as a new user. So, I'll just refresh.

[00:07:40]See if I get this hydration error again.

[00:07:43]But I've uh I've rewritten all of the introduction stuff as well and I've removed the options here. So instead of selecting what kind of project we're going for, we're just going to connect to the repo and then we're going to scan that project and we'll be able to figure out what kind of project it is. Um so we start off just by this is old hat for anyone who has been watching the streams. We basically just connect to the GitHub repo and then uh we authenticate them via GitHub and then we connect to the GitHub repo. So, I'll just speed through this so we have it set up for the stream and then the fun stuff will be next when we scan the repo. Uh, get an idea of Okay, this is uh it's kind of jumped to the end here. So, it's a bug. We'll have to fix it. Uh, but before I do, I'm actually going to delete this account.

[00:08:31]Um, uh, I'll need to, if I refresh this page, let's actually see what happens on refresh.

[00:08:40]Yeah, see we're skipping a bunch of stuff in here. I kind of want to show you that stuff.

[00:08:46]>> Uh, no, it's actually got it all recorded. It's just skipped to the end because I've already done this in a separate user. So, I will just let it skip so I can show you. And I'll just quickly scroll through. So, we uh link the account so we authenticate and then we select which repository we want to use. I'm just going with a previous project that I use called tambber.lo.

[00:09:04]It's a voice first dating app that I was working on, but I didn't have time to continue. And then it walks them through how to install VS Code, set up Claude, plug Claude into VS Code, and then set up Superbase. And finally, uh, pull in their Git repository into VS Code and connect to Versil. And then that sets up their whole development environment in an independent way using the latest recommended, you know, the industryleading standards and everything. And so this is the last message that they'll see in the getting started basically saying you've got credits to either chat to the AI or chat to me directly. And uh yeah, so now we're going to figure out the next steps. We hit one of these buttons and it will open up a new chat. Currently, we haven't got any of this implemented, but I have this all designed. So I'm going to ask for it to create an open spec for that. So I'll do that now while we're on the topic. as anything to dev say look at the road map and uh check if there's an opensp spec there's an opensp spec for scanning the repo scanning the repo uh if not then give me a summary then uh yeah give me a summary of what you will write yeah in that opensp spec in that open spec to implement the repo scanning.

[00:10:34]And yeah, so I've planned it all out.

[00:10:37]I'd like to plan everything out before all the streams. Uh, okay. So, we got cool two likes and subscribes. I appreciate you guys. Uh, and one vote for as anything.dev. So, that's what I'll work on. And for Eco3D, uh, let's have a look and see if it's come up with a decent example of the kinds of animation that we're going for. Here we go. So, this is a really good example, it looks like. Um, basically, we're trying to make this happen. Just mute the audio. We only need to look at it.

[00:11:07]So, I have a background in animal science. I actually studied zoology. So, this is kind of where my thinking has started from. But it's also completely different when you try and translate that into robotics or animation. But this right here is actually a really good example. So the way that it's stepping like this, this is what we're going to try and emulate. So it comes from the movement of fish when they undulate from side to side.

[00:11:33]Um, that's kind of cool. So that's what we're going for, except we're also going to have the limbs involved. So, we're combining that with this kind of procedural animation here. Um, so that moves a bit more like this. I'll get to a better example like this.

[00:11:52]So, this is basically chaining together a bunch of nodes that are restricted in how far apart they can move one from one another. So, they create kind of like a trail um or like a train kind of thing or a trailer. And then I'm com going to combine it with this kind of undulating movement where the lizard steps from left to right like this. So that's the goal. That's what we're aiming for. And I've documented all that stuff. So we have a good foundation to work from. Um and this is the road map overview that we've got. Okay. So we're we're replacing the old runtime with one constraint solver running each frame. So the solver is basically taking the attractor direction. It's pointing the head in the direction of the attractor and then it's calculating everything from that. So that includes the limb movements, the spine movement um and each of the steps will be calibrated from that. So it will think how far can I move my leg and how far do I want to.

[00:12:57]So the wanting to move the leg comes from how the head is moving, how the attractor is moving. And then the capability comes from the fact that it can't separate its leg from its hip. It can't turn it too far and it can't reach too far up or down or anything like that. And that's how we kind of have I think it's called emergent procedural animation. And that's what we're going for. So, uh, the hips are welded to the spine. The feet are pinned when planted.

[00:13:25]Um, and b and the bone lengths are fixed. Joint angle caps. Okay. So, I think it's all here. Now, we just need to think about what we're going to do first. So, where to start? Step one, we're going to create the um attractor and wire it to the studio scene. So, I had previously removed the attractor to try and simplify things. But since we're starting with the solver, it needs to solve something, which is the attraction location. So, we need an attractor first. So, we're going to add the attractor. Uh, and then the input when step equals 3, render an invisible floor plane.

[00:14:02]All right. And then we draw a small marker. I'm going to ask for a concise, give me a concise plain language overview of the first step in the development pathway.

[00:14:20]All right, while it's thinking about that, we can have a look at as anything.dev. So, we already have the openspec. That means that we can build it out. And I've made the decisions for this already. Um, so want me to look at the task? Uh, I think we just go for we just tell it to build it out. If the if the spec is complete, then build it out end to end. Build it out completely and we'll come back to that. All right.

[00:14:51]So, Eco3D step one, the goal is to give the creature a target to look at. What we're building is a click anywhere on the floor and a small marker appears.

[00:15:03]So, it's the same action. It's not a before we had the attractor based on Lorenzian motion, which is basically just a random movement around the screen. We're still just going to click and it will move towards where we click.

[00:15:15]So, there are three pieces. A place to store the target um that persists across frames, a way to set it, so that'll just be clicking. And a way to see it. All right. Well, that's really simple. So, that's a really small first step. I might have uh made it too granular these steps, but probably better than the opposite. Um, so meanwhile at beep.buzz, we have uh this state right here. So, I've made some progress. So, I'll basically just play the game and then as I run into things I want to change, I'll just keep improving it. So at the moment um when we hit grow like when we hit a Morse code signal um the plant grows up and it should calculate the height of the plant according to the height of the screen.

[00:16:05]It looks like it does. So it happened only when we refreshed. So I'm going to have to change it so that it happens when we resize the screen. But for now it's a quality of life that we don't need to worry about just yet. Um sorry I think I missed your message there no one here. Um you said uh when I use codeex or any AI models uh that time I feel like I'm not learning and when uh it goes down then I am nothing or when I apply for scholarship then I am uh without AI I am nothing. I get what you're saying. You become dependent on it and then you feel like you can't do anything without it.

[00:16:44]I just got to try and get this real quick.

[00:16:48]There we go. I think I killed the plant.

[00:16:50]Yeah, I I agree with what you're saying, but it depends on what you're trying to do. If the job that you're going for requires you to use AI, then you're going to need to use it. You're going to need to be skilled with it. It's kind of like learning how to uh use a tool. You know, you're always expected to have that tool when you're on the job. It's like learning how to use a calculator. And you might not be able to do those equations without the calculator, but that's the nature of the role. If you got to use a calculator, then you'll you'll need it. Um, but I do feel what you're saying. Um, the thing about not learning, I think that's mostly just about how you're using it.

[00:17:32]If you're just saying build it out, and then you're kind of like judging its response, that's a one-way street. But if you instead have a bit of a conversation and you ask for it to explain to you how it did it and then you make changes to the way that it did it then that becomes a bit more of a learning and yeah just treat it as a teacher rather than a employee or it's a bit of both really but uh I see the struggle that you're going that you have there. Okay, I'm going to try and get this right so I can make this grow and we can test this out.

[00:18:08]Okay, cool. Ed, I think I also need to update the way that the uh the message the Morse code is sent. I don't think I want the letters in the speech bubble.

[00:18:19]When the player sends Morse code, um when the player sends Morse code, it shouldn't show shouldn't show the slash to separate the words.

[00:18:39]Um, it should just show one word at a time.

[00:18:45]Uh, if the player starts uh tapping another word, when the first is finished, the first should disappear.

[00:19:01]And the player speech bubbles.

[00:19:04]I always forget how to spell speech.

[00:19:06]Speech bubbles shouldn't contain the the letters. Shouldn't contain the English letters.

[00:19:14]I think that'll be better. At least I'd like to try it out. Uh, no one here says I built step by step like per functions and try to understand the working and also uh did but it kind of sucks to uh to being depends on something. Yeah. To being dependent on something that is not in your full control. Definitely. Yep.

[00:19:36]There is a dependence that comes with using the leverage of AI, but it's also the advantage. And the I think there's like when new developers, at least from my experience when I started, when I was starting out, I wanted to understand everything before I even started building. And I've now looking back I kind of wish that I had a larger um like a wider perspective where I was just thinking about what is it that you're trying to do? What's the best way to do it? How do you verify that it's done? So much more outcome oriented rather than trying to feel a sense of ownership and completion and full knowledge of the project to just have confidence in the project and an understanding of what it's trying to do and a measure of whether or not it's doing it well. Um and then you can kind of have a bit of a blackbox approach to the rest until there's something that you need to pry into further. But yeah, in general I'd say try and have an outcome oriented approach. it'll just make it much easier to to progress. Um, if you have exams, that's a bit of a different thing, but in general, that would be my my approach.

[00:20:51]All right, so we've set up step one.

[00:20:53]Let's give it a go in Eco3D right here. I'll refresh this.

[00:21:01]Uh, also, I know Python and backend dev, but I don't have much practice. If I start to write code with hands uh handwrite code then I fall back in competition of scholarships. Yeah, I think that that's more to do with a bit of a shortfall of the existing you know uh institutions. people are still measuring um program programming skill development capability based on the pre-AI era and it's understandable because it's I I wouldn't know how to update the system to adopt to adapt to AI but it is very different now it's it's a lot more you don't really need to know as much um technical knowledge you only really need to know the systems the consequences of your actions the decisions you don't really need to know syntax for example, syntax is like the very fine grain stuff that isn't isn't so important anymore. And the scholarships, as you mentioned, uh in competitions are still oriented around handwritten code that I think is a little bit uh old school, a little bit uh behind. But in saying that, I've already learned all that stuff. Like I spent a lot of time writing by hand. Um you know, writing full files by hand. I just couldn't do it anymore in terms of speed. I I couldn't um I couldn't stay competitive in the actual industry. Now, if I were to go back to doing that, all right, so the attractor is built.

[00:22:34]What's the next step?

[00:22:39]Okay, and these two are running.

[00:22:42]So, I've actually this is actually a game that we have here, and I got it set up because I figured it would be good to have something to fall back on when all three things are are like building something, but I haven't quite figured out how to play it. I can give them food, I think, and I can uh I don't know. I don't know how to play it, but I've got these characters.

[00:23:06]Let me know which ones I should keep and if they're too busy on the screen. But these guys shouldn't take too long anyway.

[00:23:15]All right, so step two, head gaze with angle cap. So the goal is that the head rotates to face the attractor capped at its angle limit. So just the head. So that's good. I like this step-by-step approach. Do that. So we've got the attractor now. We can tell it where to look. And now it's going to keep its body perfectly still and just turn its head to look at it. And then the next step, I suppose, will be the spine. And we're going to build it in incrementally to get this combined approach. So it's looking like this.

[00:23:49]This thing right here.

[00:23:52]Try and get a moving section.

[00:23:55]It's going to look like that combined with the robot like this. Yeah. So that's kind of the spine. We're going to have it track something. The head will lead all of the nodes. We'll have rules, internal rules about they can how they can interact with one another. And then we're going to have the legs computed from this like into this stepping motion. So rather than just having like a straight line, if this thing was moving along a straight line like the lizard is, its body wouldn't be bent.

[00:24:33]But we want that body bending from the stepping just like a fish would. So, we're combining these two. And it's going to be tricky. It's been like really hard to to build this thing, but uh we're making progress slowly or at least we have a clear road map now. So, we'll see how we go.

[00:24:54]All right. So, what are these guys up to?

[00:24:58]All right. So, this is testing. This is doing a a um type check. So, we can actually try it now. That means it's finished. So, let's do the input. Um, this should be different now.

[00:25:12]That last E wasn't on the screen. Um, in the in the speech bubble, the plant speech bubble, I think it might be width limited. Um, make sure that all of the MOS signal characters are always visible.

[00:25:40]And then just say currently um the last character can sometimes be cut off.

[00:25:51]Cut off.

[00:25:54]Um, and we'll just run that now.

[00:25:57]And then now if we do one more T, that should grow the fruit. And then one more. Still sending still doing a slash in the speech bubble.

[00:26:10]So now that's dropped down, but I noticed that we don't actually have the fruit visible in the other side here. So when the fruit drops to the other side, uh it needs to render on the other players's um ground, like on the other player's ground level.

[00:26:33]And we'll say visibly visibly render on the other player's ground level.

[00:26:42]And the player needs to be able the other player needs to be able to pick it up.

[00:26:53]Um, just see if they can actually increase their count. Maybe it's just invisible.

[00:26:59]Yeah. No, the other player needs to be able to pick it up um and increase their fruit count.

[00:27:09]When they do, um, the fruit should always drop, always land on one of the uh positions.

[00:27:25]Should always land in a location in a location that aligns with a potential landing spot.

[00:27:36]landing spots of uh the character so that they can move to it and land directly on it.

[00:27:49]So, I'll wait for that to finish thinking. Um I think the thanks for voting in the chat, whoever voted. I think we got two votes out of five for as anything. So, I'll switch my attention a bit over to there now. Um well, actually, I'm going to wait till it finishes building. It looks like it's almost done. Um, and then this will be when we can actually start showing what's like the core use case of this web app. So, I'm looking forward to that. For now though, let's look at Eco3D.

[00:28:18]Step two is wired up. So, now it's just looking turning its head in theory and then from that we'll be able to calculate the rest. All right, that's good for now. It's not perfect, but it is limited. We don't need to think about how much is limited just yet. So that's fine. Um, that's good.

[00:28:40]Concisely, uh, give me a concise summary.

[00:28:45]Give me a concise summary of the next step.

[00:28:52]uh and I want to say uh keep in mind I'll give it I'll ask for it to give me the summary and then I'll check to see if it actually makes sense with the larger model that we're building because I don't trust it. It's been really tricky to keep it on track this whole time. Um but we'll see how we go.

[00:29:12]All right, it looks like we're finishing up uh the process the openspec. There we go. So this is at as anything.dev. Uh, we have migrations migrations shipped.

[00:29:24]So, we need to push them. We need to make sure that we're linked um to the right Superbase project. Um, I'm just going to do I don't think it it matters to show you the Superbase projects that I have access to uh because they're just the names and uh yeah, they should be fine. uh super npx superbase link and we just want to connect it to the project is actually still called um as gazola.dev uh but it's just another name for as anything.dev.

[00:29:57]So let's do npx superbase uh db push and that'll push the migrations as long as there are no errors.

[00:30:07]Once we've done that, >> uh, do we need a what's going on? NBX super push connecting remote DBS. Oh, it's up to date. So, I'll actually ask did you push the migrations?

[00:30:22]Um, and then I want to check what actually like how do we test this thing?

[00:30:28]Uh, the two side effects was worth flagging. The old viest.config was failing. That's fine. That's just a test. We're not running tests just yet.

[00:30:36]All existing tests will pass. Um, task 6.5.

[00:30:42]Okay. Subscribe the GitHub app to the push event. It's a manual config change.

[00:30:49]That's fine as well. We're not worrying about that just yet. We're going to do it manually uh at the moment. Say, give me instructions. I'm going to skip this song. Let me know if you guys have preference for music as well. By the way, I know it's quiet, but I don't think I should have it any louder. I don't know.

[00:31:06]uh give me instructions for testing out the new feature. Um I'll actually just say smoke testing.

[00:31:14]Smoke testing the new feature.

[00:31:18]Manually smoke testing.

[00:31:21]Uh smoke test just comes from the idea that you turn on your PC for the first time. If it doesn't smoke, then it's working. Like it's it passes the first test. Smoke test. Um the new feature in the browser.

[00:31:36]Okay.

[00:31:39]Eco3D. This is explaining the next step.

[00:31:42]And I said concise and that's this is its version of concise.

[00:31:47]Um that's not concise.

[00:31:51]Uh use plain language.

[00:31:54]Concise. plain language to explain the next stage, next step in in the development road map. In development uh in the context of the two types of animation that we're modeling this uh we're modeling from.

[00:32:21]Okay. Um, and this is for testing the changes that we've just implemented in as anything.dev.

[00:32:28]>> All right. So, it's already pushed the migrations, which is handy to smoke test. This slice intentionally ships no new UI. Okay. So, that means that it's syncing in the background. So, it wants to to sanity test uh sanity check the scan output. Click the scan results cell in superase. All right. I don't want to do that. So rather than doing that, uh let's just build out the next thing.

[00:32:56]Let's just build out the UI that depends on it on this feature cuz we got the scan in place now. It's going to get an idea of uh whether it's built with Nex.js, whether Superbase is installed, whether Shad CN, Tailwind, uh and some other things are configured. I'm going to expand the scan to include checks to see whether the recommended patterns that I'm using have already been implemented or whether I can give them a prompt to set that up and whether they have like a sidebar set up and how the sidebar is set up. So, I'm going to expand it step by step. And so when people uh use this app, it's going to grow its capability to be more and more um capable to be able to build based on your existing web app. But we're starting from the very basics. It's just checking what kind of app do you have, what's in there already, what can we add, um is it is it brand new, and is it a lovable app as well. All right, so let's build on the UI that depends on this feature.

[00:33:59]look at the road map and understand the overall direction and what we're trying to build.

[00:34:11]Um, I want to start with the simplest implementation.

[00:34:19]I think I think we want to start off each chat, each AI and as chat with a scan and a set of options that um that yeah a set of options that are determined by the results of the scan.

[00:34:48]So for example uh example um if there's no next installed if there's no next installation uh and the repo is empty then we should give them instructions give instructions on how to set up next and clear out the boiler plate.

[00:35:17]Uh, and then I'll just do etc. Um, yeah, I think that's good.

[00:35:24]Okay. Eco3D. I asked it to be a bit more concise and relate it back to the two ways the two animation processes that we're using to construct this animation.

[00:35:35]So, right now, uh, only the head rotates. So, we've got that working.

[00:35:38]Just as a refresher, I'll show you.

[00:35:40]We've got this happening. Very basic.

[00:35:43]Um, and we need this to lead the rest of the body. Needs to bend to try and see it better. And then eventually it's going to step. But I'm just asking what is next? Do we propagate this movement down the spine already or is that getting ahead of ourselves? Is that going to kind of lead towards this but not this? Like what what happens next to make this happen overall?

[00:36:10]So going step by step. So right now only the head rotates.

[00:36:14]Uh click off to the side and the head turns yada yada. Step three fixes that.

[00:36:20]So the spine cascade, it does want to go for the spine cascade first. When the head uh runs out of cap, the next joint behind it rotates to help. If that's still not enough, the joint behind it and so on. So that is describing exactly this. I think uh actually it's not quite because this is just trailing like if we play that section again this is just like trailing behind right so in theory the head could just keep turning and it unless it was actually moving. So this is about movement and we're not talking about movement yet. We're actually talking about just turning the head and then it'll trail the body behind.

[00:37:05]I think I don't know if I'm understanding it fully. Okay. Um, the demand to face the attractor travels back up the spine like a wave. Each point contributing what it can within its own cap. Hips and tail stay frozen, so the cascade dies out before it reaches them. Okay, this is the uh Argonaut chain, plain and simple. So, the Argonaut is what it's referring to when it says uh this is the Argonaut version. That's his name right there.

[00:37:37]Looks like it's snowing for some reason.

[00:37:40]I guess that's the pet game that we have.

[00:37:44]Um I don't know. I don't know about this game. I've only just installed it today.

[00:37:49]I don't think it's in it's snowing.

[00:37:53]It's only snowing on some screens.

[00:37:54]Anyway, uh this is the Argonaut chain. This is also the PBD in miniature. So each joint cap is a constraint what you'll see. So what I want to know is if we follow if we implement the spine propagation.

[00:38:16]Uh I understand that it goes in the direction of Argonaut in the Argonaut direction.

[00:38:26]But does that direction deviate from the what would this be called?

[00:38:40]>> Um machines modeling after animals.

[00:38:45]The I think it's called elongated vertebraate. Vertebrate locomotion direction.

[00:38:56]So if we if we continue continue along the Argonaut path then the body will move forward in a straight line.

[00:39:12]straight line if the attractor is directly in front.

[00:39:20]Whereas the vertebrae locomotion causes is driven by a step is driven by stepping.

[00:39:35]Um how do we combine these?

[00:39:40]And how do we combine these?

[00:39:43]And is the spine propagation as you've described the correct next step in this context?

[00:39:58]Okay. And as anything is building out the UI so we can check out the um so we can ah I didn't send this off.

[00:40:07]Let's send it off. So as anything is building out the UI for the scan results, this will lay the foundation for then giving them options to build out stuff on the uh as anything like so they can get prompts and that to build their UI. They can make fixes. I can identify things that they might need to fix. And so this is the foundation for that. But it'll make a lot more sense when I can show you which it looks like it should be done. There we go.

[00:40:38]Okay, let's give it a go at as anything.

[00:40:42]Scanning the repo. I'll actually just refresh.

[00:40:47]Scanning the repo. Let's see.

[00:40:51]Shouldn't take that long. It's not an AI doing the scan. Uh, it's a programmatic approach. So, maybe I actually need to Well, we've already pushed the migration, so I'm not sure if that's if there's a delay there.

[00:41:05]Um, thanks for voting. By the way, we got another vote. Uh, I think it was for the as anything. So, I'll I'll keep that in mind. I'll focus on that. Uh, I'm really liking that the votes have been for that because I would like this to be the focus. The idea is that people can actually we'll we'll have streams where I'm actually helping people to develop their apps using this app. um that's part of its purpose and so it'll be even more interactive which will be fun. So it looks like we're hanging on scanning your repo.

[00:41:38]Um it's hanging on scanning your repo.

[00:41:46]Your repo.

[00:41:48]There's no feedback feedback or errors in the console.

[00:41:55]Yeah.

[00:41:57]So, it's just hanging.

[00:42:00]The cookie error there is rejected cookie is interesting, but we're signed in, so that shouldn't be the error. Um, yeah, this will probably prompt Claude to add logging, but it might also just sort of indicate, you know, show it to a pretty obvious error that it's not kind of paying attention to just yet. I'm just going to try and get some of these little dudes on the screen a bit more. They run around. I've only got like a short uh small version, small window of the screen that's actually shared on the stream.

[00:42:36]Okay, Eco3D.

[00:42:38]Um, good question. This exa, this is exactly the tension the design resolves, and it's worth being clear. So, the pure Argonaut would drift. In Argonaut, the head's position is the goal. So, dragging the target um to forward pulls the chain behind it like a string. The body translates whether you want it to or not. Our cascade doesn't. The head's direction is the goal, not its position.

[00:43:05]All right. Um, yeah. So, that's what I was thinking. So, step three only rotates joints around their shared nodes. Nothing translates. So, it's rotation only. So, we're turning left and right. And then the movement. So, we're separating these two. I get it now. Um, let me bring this up again. I'm going to bring it up a lot because it's even just useful for me. So, this is all about movement. The only thing that's making this animation make sense is the head moves and then the body moves. But this is about orientation. So, right now, you can see it's it's in motion, but we're not even touching any of this motion logic yet. We're looking at how the head is actually turning. So, if this lizard was standing still and it looked to the left, it would start to shift its body around to allow its head to turn. And that's what we're starting with. And then the locomotion will be uh secondary while the internal congruence that we're building now is maintained.

[00:44:06]So, that sounds good. Um, sounds good.

[00:44:10]Build it.

[00:44:12]And I'm also just going to uh stage this stuff. I haven't done this in a while, but this just means that I like the stopping point that we're at. I don't necessarily feel like we need to commit.

[00:44:23]But, uh, this will separate things so that any new changes will be in the changes section here rather in the rather than in the staged changes. It's just a good way of like saving a soft checkpoint and a a git push would be a hard checkpoint or I should say git commit.

[00:44:41]Okay, so the type check's clean as anything.dev. Let's give it a go. Um, it was solving an error. Let's just see.

[00:44:51]We've got some kind of We got a new pet, I guess. Okay, that's kind of nice. Okay, so let's test out as anything.dev.

[00:45:03]All right, so this is the result of the scan. I'm going to refresh the page and just see what it looks like. All right, so it scans pretty quick. the initial assessment.

[00:45:13]This is in a uh a chat. This is for messaging as but we also have uh new AI chat. I'll I'll make these clearer separated. We have the getting started which we've finished. That's just setting everything up. And then we have uh right now this will be an AI chat. At the top of the chat we have the scan. We can hit rescan to scan the repo again. I'm just curious to see how long it takes.

[00:45:41]took a little longer than I thought because I thought it would be caching.

[00:45:44]Um, but it's okay. So, we can wire the Superbase MCP server or we can uh add a starter claude skill.

[00:45:56]Okay. So, I want to understand what we're actually scanning. Um, so I don't really like the user interface that we have here. I want it to show the current state of the app. So, um the UI so rather I want to say remove the current UI. Uh I want the UI at the top of the chat. Uh not nec I don't want to actually say at the top of the chat because this same UI will be used everywhere uh that we use a scan. So I'll say the UI from the scan to indicate what was scanned, what the results of the scan were, what the results of the scan were and to provide a set of um user like I will call it a grid a grid of actually do we want a grid or do we want it to be I don't think we want a grid.

[00:47:04]Uh a list of user action components that will enable the user to start a conversation from that action.

[00:47:19]So eg if the repo doesn't have a uh doesn't have nex.js JS installed then it should uh include that as an action but I'm also thinking so at the top at the top um okay so I'm thinking more detailed now so I want the UX remove the current UI I want the UI from the scan to indicate what the results of the scan And let's be more specific here.

[00:48:02]So I want to at the top here at the top should be a collapsible element that uh will expand to reveal the full uh results of the scan.

[00:48:24]This should list uh each item each check that was performed with a uh check mark a check mark to indicate if it passed.

[00:48:43]And I also want to say explain to reveal um a scroll element with the full results of the scan. This should list each uh each check that was performed with a check mark slash x to indicate if it passed and below that should be yeah and then just I'll go straight onto this below that should be a list should be a list of user action components that will enable the user to start a conversation from that action.

[00:49:23]Eg if the repo doesn't have Nex.js JS installed then okay um the action components should be concise and don't need to include any details like prompts uh and only need a concise description a title and a description and we'll say no prompts etc. and then clicking on an action should start the conversation to implement that feature.

[00:50:07]Okay, that was a long prompt. I'm sure that the uh the view count dropped there. I don't have a view count to see, but uh I can tell when it's like it's probably not the most interesting to watch. But thank you if you're watching.

[00:50:21]I really appreciate you being here. If no one's here, then that's fine, too.

[00:50:26]But I'll try and be here as often as I can. It's quite late at night for me.

[00:50:31]Uh, it's about 1:00 a.m., but I stream until about 200 a.m. and uh just cuz that's when I have time during the day.

[00:50:38]So, this is the the app that I'm really keen to see progress. Um, eco3d.shop.

[00:50:46]And the next step should start to see some actual interesting locomotion happen or at least the starting of it should propagate soon.

[00:50:58]But yeah, um so now we have this character and we have star return mission.

[00:51:07]So this is the little game that I have running at the bottom. That's what the characters are. So, we need to collect coal, whatever that is. I think it's coal and wood. So that we can build so we can build a uh what's it called?

[00:51:24]Spaceship to launch the star back home, I guess. But I don't know how to get those things.

[00:51:32]I don't know how to get those things.

[00:51:36]Uh yeah, I don't know. I think we can we have this little house here.

[00:51:44]This is where the the creatures live.

[00:51:48]And I think that we can The fridge is empty.

[00:51:53]We can play this game, but only when we have a coin, so I don't know how to get any wood or anything. I need to look this up.

[00:52:03]But I thought it' just be something interesting that I can kind of have on the screen to keep things moving.

[00:52:11]Okay, so eco3d.shop is at the type check. So that means that it should be ready to test. See if it works.

[00:52:23]There we go. So we're propagating that head movement. It's only happening down to the second element, but the neck is now turning. So that's good.

[00:52:36]Um, so what's next then? So the neck first I just want to check if it's correct is now turning.

[00:52:44]Uh the head and neck are now turning the hips behind the neck. I don't know if you call them hips. Shoulders I guess for I don't know if you would call them shoulders for a quadriped. It's got four feet. The shoulders are not turning.

[00:53:06]Um is that expected?

[00:53:10]and then uh concisely in plain language describe the next step.

[00:53:22]Okie do as anything.dev is still building. We're starting to get some artifacts in here.

[00:53:29]I don't know why that happens. It only happens on this user. I think maybe it's because I use such a thin screen for streaming.

[00:53:37]But usually that doesn't happen.

[00:53:40]Beep. Buzz. All right. So, the fruits now are stacking properly. The other thing that was a problem with beep.buzz was um the actual message sending didn't look great.

[00:53:56]>> We get that slash still.

[00:53:59]So the when a player sends a message, the speech bubble shouldn't contain a slash.

[00:54:12]It should uh only show one word. I actually want to see if that works. So if I try like this. Yeah. So it shouldn't contain a slash. Um it should only so shouldn't contain a slash.

[00:54:36]So the user should be able to say one word and if they start typing another word, another word, it should clear the first one.

[00:54:53]the first one from the bubble.

[00:54:58]But we also It looks like the fruit isn't actually um dropping to the other side.

[00:55:08]This I think the E is actually wrong.

[00:55:10]We're not seeing that. Yeah. So, it's not actually dropping to the other side.

[00:55:13]If I refresh, we're going to have to start again and and uh build another tree. We can do two at the same time, but the fruit needs to stay there so that the other player can pick it up and then we can build things and the rabbits should be attacking. So, if I come back over here when There we go. So, I can scare off a rabbit like this. Then they run off the screen. And if I bring the plant off the screen, then the rabbit's going to come back and eat the leaf again. So, we need to also make sure that the configuration is working properly for the ab for the rabbit leaf eating. Um, but let's just try and get a fruit to the other side.

[00:55:58]Okay.

[00:56:01]Okay. And now the message sending looks good.

[00:56:05]So, I like that the rabbits are kind of like hanging at the sidelines and you got to scare them off. I think that's kind of cool. Let's try this again.

[00:56:16]like that.

[00:56:18]And I want to try doing like two words at once just to see if it >> Yeah, nice. It clears it.

[00:56:27]>> Okay, awesome.

[00:56:29]Um, let's try and get a fruit to the other side.

[00:56:35]Okay. And then I'll wait for those leaves to grow. And every time you get one wrong, uh, the leaves will fall off one at a time. It might be a bit too easy, but eventually there'll be full words that we'll be we'll be playing.

[00:56:51]Okay, so I'll wait for that to grow and we might as well take a look at Eco3D.

[00:56:56]So the next step here is I'm actually going to create a new terminal and continue this so that we don't have the formatting issue that I'm seeing.

[00:57:08]All right, so the shoulder staying still is expected. Okay, nice. Um, that's the front hip group and step threees everything from their back. The body bending up to the shoulder line is exactly what the C cascade should look like. All right, so now the hip wants to rotate, but the feet won't let it.

[00:57:32]The change is that the front hip joint stops being frozen. The cascade now reaches to it um asking the shoulders to rotate further around towards the attractor. The catch is that the feet are still planted on the floor. The legs are like ropes tethering the hip to the feet. When the hip tries to rotate, the leg pulls back. So the hip can't actually move. So turns its head. The hip tries to rotate but the foot is keeping it in place.

[00:58:07]Uh like the leg is like a rope and the foot is like an anchor.

[00:58:13]So what we'll see next, uh nothing new visually. The body still looks exactly like step three. Um that's correct.

[00:58:22]What's actually different? Uh every planted foot now has a measurable number called strain. how far it would have it would have to slide if the hips were allowed to rotate freely. Okay. Uh click far around and the number grows. Click back to neutral and it drops. Why this matters? The strain is the signal that drives the stepping. So the visible test debug overlay shows um okay so you add the strain next. So we're adding the strain next.

[00:58:56]Uh, next with the debug uh debug overlay.

[00:59:02]Uh, do that.

[00:59:04]I'm not sure if it's saying that we already have the strain and that it wants the overlay now, but I think that it just needs to add the strain. We're going very incrementally, but I like that I'm actually understanding it as we go because previously I've just kind of like let it get too far ahead and we ended up going down the wrong path and it took me a while to walk back and realize that we actually need to start again. And so now I think I have the concepts down um and I've constrained it because it keeps wanting to jump ahead.

[00:59:32]But I think it's a better approach. Um so most of the votes are for has anything.dev. So I'm going to focus on that. I do want to just kill this uh terminal and create a new clawed instance so that we have the formatting issues resolved that I get sometimes.

[00:59:49]Okay.

[00:59:52]So the redesigned UI. Oh, okay. we get to test this out. So, um let's see what it looks like if we refresh.

[01:00:02]Um so, this is the scan results. We've got the scroll container.

[01:00:06]We've got all this stuff ready, but we don't have any clawed skills, and we don't have the uh MCP for super base.

[01:00:16]So, that's good. Uh we probably want probably want to start with a repo that doesn't have anything.

[01:00:24]So, I might even just create a new repo from my accounts off screen real quick.

[01:00:29]Won't take too long. Uh, let's go repositories and then new. And we'll call this um as anything test and we'll just leave it like that. It'll be public and we'll go create repo.

[01:00:48]So now I'm going to change to this repo.

[01:00:50]We should have access to all of them.

[01:00:53]Uh, let me see. That's test. That's a different one.

[01:00:59]Um, I'm actually wondering, first of all, this isn't scrolling.

[01:01:05]Uh, so we need to make sure that we can see all of the repos in there. The repo list that opens from the sidebar should scroll like the repo list that opens from the getting started chat.

[01:01:24]getting started chats.

[01:01:27]Um, also if I add a repo, if I add a repo and then open the select the repo select, we'll say add a repo externally to GitHub externally and then open the repo select uh element.

[01:01:52]Currently, it won't show the new one, right?

[01:01:56]the new one, right? Um, I think I want to refresh the repo list. I don't think.

[01:02:06]Oh, if so, if so, I want to refresh the repo list every time that it opens so that I can add a repo and access it immediately um immediately from the app.

[01:02:30]Okay, bit of housekeeping just so we can do the testing, but it's something we need to implement eventually.

[01:02:36]Both behaviors are already implemented from earlier. So, no, I think that they're still there. Maybe I need to refresh. Let's try refreshing.

[01:02:48]>> Let me see here.

[01:02:51]Uh, the new word clears the old one.

[01:02:53]Yeah. So, we've Okay, so we have got that going. I've got to re rebuild these plants now. I should also do the top character.

[01:03:03]Do I There we go. And let's see if we can send a fruit in both directions.

[01:03:11]All right. So, we'll come back to that in a minute. Uh, Eco3D is building out the next step. So, this will just be this is logic only the next step of Eco3D. It's strain. So, telling the hips what they would want to do if they could move.

[01:03:29]So, let's uh keep building keep planting these plants.

[01:03:38]>> I think I'm Yeah, I just killed the plant. I need to wait longer, I think.

[01:03:43]>> Let's try it again over here.

[01:03:49]Okay, so that worked. Uh when the player moves the speech bubble should clear and disappear.

[01:04:01]And if uh we should just say should clear and then say if the speech bubble is clear is clear no content uh it should disappear.

[01:04:21]Okay, let's try and type the That's kind of a useful one, right?

[01:04:29]I think I waited too long and it wrote two words there.

[01:04:36]I'm just doing too many letters.

[01:04:39]And it's also not very clear that the leaf fell off. Like I didn't get a lot of feedback then.

[01:04:47]So, I need to It's good just to keep playing it and see, you know, where it's where it's working and where it's not.

[01:04:55]All right, Eco 3D. Let's give it a go.

[01:05:00]Let's try out the animation.

[01:05:02]Let's refresh.

[01:05:08]Where are you at? There we go.

[01:05:10]Okay.

[01:05:12]So now we should have strain the attractor.

[01:05:20]>> So what are we supposed to see here?

[01:05:23]Test in the browser at rest. Both strains equals zero. I'm not seeing any strain measurement.

[01:05:30]All right. Strain like this.

[01:05:32]>> Left and right front. They're both exactly the same.

[01:05:40]Um, and now there's no strain.

[01:05:47]>> They are identical. Um, I want to make sure that's not like abnormal or problematic. Both strain measures are identical.

[01:05:57]Identical.

[01:05:59]Um, because that expected uh concisely describe the next step.

[01:06:06]the next step.

[01:06:11]Okay.

[01:06:13]Beep.

[01:06:15]Um, let's check out as anything. I'd still gone. So, we'll do beep.

[01:06:23]Uh, let's try it out. So, I think we did when we move now, it should clear. Yeah, it should clear the message.

[01:06:33]Okay.

[01:06:37]Um, I think that I think the spacing between the letters is is okay. It's hard to tell.

[01:06:46]Um, the other thing is I I want the characters.

[01:06:51]So um both so either up or down arrow keys should send up or down arrow keys and either um W or S keys should send MOSS.

[01:07:12]Let's be clear about that. Up or down arrow keys should send Moss for buzz and up or down, no W or S should send MOS for beep or beep.

[01:07:32]So that's just a quick note. Okay.

[01:07:35]Identical strain values are expected.

[01:07:37]The formula is such and such. Pure cord length. Both feet are roughly the same distance from the hip back joint. So the magnitude matches. So it's the direction, but I don't think there's a way of I think it's just pure magnitude.

[01:07:54]So one swings back and the other swings forward. So step five is front feet can step. Okay. Uh the change when a foot strain crosses a threshold, that foot lifts off the ground, swings to where the hip wants it, and replants there.

[01:08:12]Okay. Um so do that.

[01:08:18]Give it a go.

[01:08:20]As anything.dev. So this is just syncing it up so that I can connect to the repo, the new one that I just created, which is uh barebones new repo. so that we can see all of these failing and then see what it suggests. Currently, it's pretty um it's pretty overwhelming kind of thing. Like there's a lot of data here.

[01:08:42]If we start this chat, this where the AI walks through the task yet. So, we don't have any chat stuff yet. But now that we've selected something, it should be in the chat history. So, I'm thinking that we should we first of all, we need a chat box down here just like we had for the getting started. So, we go back to the getting started. We have the chat box. We actually haven't used this here.

[01:09:10]And but we want it on all screens. So, I I should add that right now. add the chatbox UI to the um AI chat session UX and the as chat session UX.

[01:09:32]Um when I'll just add that for now and then I'll I'll think about this. So we should have one. It's not scrolling, which is good.

[01:09:46]>> There should be one called test as anything, >> but I don't think it's going to be there. We should also have a search in here.

[01:09:58]>> As anything test. There we go. And that says current. And so before I forget, I'm going to say add a search uh bar to the top of all of the git repo.

[01:10:13]Um, select option elements.

[01:10:18]Um, select elements, select options, elements. I think we need to say that because otherwise it'll put it at the top of the collapsed state. Um, search at the top. Um, it should filter the options by the search query.

[01:10:44]um by the search query substring so that we can you know it doesn't have to be exact it can just contain the string and I'll wait for that to finish okay Eco3D it's breaking up so let's refresh this again resume all right so identical strain values are expected Oh, did I kill it too early?

[01:11:16]>> Maybe it uh I think it might have been still running. I thought that I thought it would finish. That's okay. All right.

[01:11:25]So, beep. Buzz per player sets track >> only plays when the Y. Okay. So, I think we just added uh up or down can now send the M code. We're going to have conflict here as well when the plant gets tall enough. But I'll cross that bridge when I get there.

[01:11:47]>> So, we just lost a leaf. I'm going to have to add animation to make that clearer. And also, when we get it wrong, I want it to reset. So currently each letter in the plant speech box uh speech bubbles the required required letters um are accepted one at a time.

[01:12:22]They need to they need to um be considered need to be considered together.

[01:12:38]So the player needs to say it say the letters in one sequence otherwise it restarts.

[01:12:49]And then also the the words the letters in the speech bubbles should always spell a word.

[01:13:04]So then you know i.e. oneletter words, words like a and twoletter words like it, etc. Okay, Eco3D, it's building its stuff out.

[01:13:27]Um, I always wonder why it sometimes takes so long to even start burning tokens. And you know, if you if you clear it and then reset it, then it often starts a lot faster. So I don't know if it's like I know I've mentioned this mentioned this before. I think it might be like soft limiting.

[01:13:49]All right. As anything dev. So I just wanted to add a search element to the top of the the repo select and we've so we've added the chat boxes.

[01:13:59]I just want to check whether they're there. So now if we go to the new chat here, I don't see them. I think it's when I click on this. Yeah, I would rather have the chat box always present.

[01:14:15]The chat inputs chat box input needs to always be visible.

[01:14:25]Uh visible.

[01:14:27]I just want to see if it works for messaging as Yeah, there we go.

[01:14:33]Okay, nice. Um, uh, I'm going to have to catch up on what this is, like on how this is set up, but basically this is where you send a direct message to me. So, the chat box input needs to always be visible.

[01:14:50]Currently, it only shows after selecting a um task action component.

[01:15:02]Um the user should be able to just send a message directly.

[01:15:12]Yeah.

[01:15:13]in either the Az chat or the AI chat.

[01:15:22]See, this is still thinking. What y'all thinking about, man? Just get doing.

[01:15:28]So, now we're going to use the strain to determine when a threshold passes where it will try to step to move and then it's going to start to look alive. But I do think it's going to look crazy. Like the threshold for both sides are the same. So it's going to be stuck like this, turn its head, and then it's going to suddenly move both at the same time.

[01:15:53]So, we're going to have to add some kind of logic to like syncopate them so that you have one I don't know if syncopates the right word, but to offset the timing so that one moves when the other doesn't and then it'll look more like a creature.

[01:16:12]Okay.

[01:16:14]All right. They're still cruising. Um, speaking of cruising, it's about 1:30 a.m.

[01:16:20]So, I usually wrap up at about 2 Um, I think it's a pretty quiet stream today, so I might wrap up a little early. I don't know. See how I'm going.

[01:16:32]But, uh, yeah, this game wants me to build a spaceship for this star character, but I have no idea how to do that.

[01:16:46]Once you collect all the materials, the rocket parts, you can send it back home.

[01:16:50]This will lose this pet, but maybe something good will happen.

[01:16:55]So, I don't even I didn't even know that we had materials in this game.

[01:17:00]Um, I know that we have like Here we go.

[01:17:04]We've got this, which is we can unlock a new room.

[01:17:10]We can expand the house. Oh, cool. We can make the house kind of interesting.

[01:17:15]Kind of trippy. We've got money, but I don't know where those other resources are.

[01:17:24]Maybe we've got this Pokedex or pet decks.

[01:17:30]So, you can kind of see this is the character names that we've got.

[01:17:37]We have this guy. I like this guy. Leaf otter breeding decks. It's kind of cool. I don't know how you can breed them. Maybe a special kind of room or something.

[01:17:49]But yeah, I don't know how to uh So, we can right click to choose focus level, right?

[01:18:02]Full. I just going with standard.

[01:18:07]Oh, made them hide. Now they're back. Okay, another machine. So, we got a token.

[01:18:19]I don't know if this game is I don't think that anyone's watching anyway, so I'm just going to fill the time in with the game.

[01:18:25]This token gets you new blueprint. New blueprints like All right, so let's do that.

[01:18:32]Cool. I don't think you guys can see the whole thing. I can't move its position, but let's see what we get.

[01:18:41]How do we make it come out?

[01:18:44]Do I just click one of them? There we go.

[01:18:50]I don't think you can see it. It's off the screen. But we got a blueprint. Ah, it's just like for a house. I want to figure out how we can get the stuff we need for this dude to go home. He's uh this little star guy. He needs to go home.

[01:19:07]But I don't know how to outfits. Thought I need an outfit.

[01:19:11]Uh I think it's got to have something to do with this thing. I see that it's finished building the background. Let's just send that off actually while I've got this.

[01:19:21]Uh so we can like we have a shop, but I don't know how to use it.

[01:19:28]So I don't know. Let's put this aside.

[01:19:30]I'll get back to work now.

[01:19:32]Make this a little smaller. Okay.

[01:19:36]All right. So beep. Buzz. So, there's still What's going on here? Let me kill this. And yeah, see, it hasn't even when I hit escape, it returns it to the the prompt input. So, it hasn't even started building it. So, let's start a new chat and resume.

[01:19:58]And I might even compact. But if I compact it.

[01:20:03]>> Yeah, I'll compact and then we'll continue from there. Look how crazy this stuff goes.

[01:20:12]For some reason, this doesn't happen hasn't happened elsewhere. I might have different formatting set up on this uh this user on this computer.

[01:20:25]All right. So, beep.buzz Buzz theoretically should be the first thing to get completed, but I can also just kind of review, see what we're up to here. Um, like I don't like this UI. It looks very different to the UI that we had for the start. And I also don't like that this is I've got to spend some time refining this. But what I want to do first is um make sure that the logic is in place.

[01:20:55]Excuse me.

[01:20:59]And for beep. Buzz. Let's see if we can get it to the other side.

[01:21:05]>> And if the fruit actually persists on the other side.

[01:21:09]I think I like the pacing of this game now because you want to go like abroad to plant more plants, but then if we do that, when we come back, they're being attacked by rabbits and we need to scare them off. And so we can only have two to start with, at least reasonably, unless you have a wider screen. That's actually something I need to think about. So rather than having it off screen, having the calculation come from being off screen, I'll need to actually have it measured by distance from the character.

[01:21:40]So that's something I'll need to revisit later. But for now, we can keep it like this.

[01:21:47]So, this is a case where the eye isn't actually showing cuz the eye is two short dots, but it's not visible.

[01:21:58]I think it's two short dots.

[01:22:04]Yeah, see that worked. Um, sometimes the Oh, we need to go to the beepers.

[01:22:11]Sometimes the Morse characters aren't visible in the plant speech bubble.

[01:22:24]For example, um I just saw one that had uh the letters I and N.

[01:22:35]That should have um did dit for i and dit for n. But it only showed um the dit no dit.

[01:22:55]Okay, I'll send that when it's finished.

[01:22:57]Okay, so uh what is the next step?

[01:23:03]concisely explain what you will do next and how it will work.

[01:23:14]Remember that the strain on both hips is currently exactly equal.

[01:23:28]So I think that means that the first step will look step won't look right because both legs will move at once.

[01:23:43]Um should we resolve this now or after implementing the first step?

[01:23:53]the next step and I'll send this one off to okay at asnthing.dev.

[01:24:08]So now we should have the chat box always visible so they can just start sending something off. What happens if we send off a hello?

[01:24:17]Okay, sweet. So we need to send set the anthropic API key. I'll do that before next stream. I don't think that we'll get to the actual AI generation just yet. Um, but I want to switch to as anything test. It's going to scan the repo. When we switch, that should clear that error.

[01:24:38]But, um, I'm not even going to concern myself with the errors just yet. Just getting the basic configuration down.

[01:24:47]So, zero out of 10 passing. That's what we expect.

[01:24:52]um to the framework. So why are some question marks?

[01:24:58]Some are question marks.

[01:25:01]All right. So I created a new repo and I run a scan.

[01:25:08]It showed. I'm just going to tuck my dog in. Just give me a sec.

[01:25:26]All right. She was looking cold.

[01:25:29]Okay. So, I created a new repo and I run scan. It shows zero out of 10 passing uh new empty repo.

[01:25:43]Uh but when I expand some have uh question marks rather than crosses X's next to them.

[01:25:58]Some have Oh, what's going on here? Hello.

[01:26:06]That's weird. What? What is Where are we at right now? New AI chat. Hello.

[01:26:14]Oh, no. I sent hello. That's from me. I was like cuz I thought that we didn't have the API key. Okay. So, if we go to test this one, right? As anything test. If we refresh, should show the scan at the top.

[01:26:35]New AI chat. So, that's registering as already started.

[01:26:41]If we start a new chat, let's hit new chat. And now we should have this. Okay.

[01:26:46]So, new AI chats.

[01:26:50]We got the scan results. Some things are showing question marks. Um, I want to understand what that means.

[01:26:56]And we also only have a few options here.

[01:27:03]>> Install Next.js.

[01:27:06]>> I want to make this a lot more intuitive.

[01:27:09]But I don't want to spend too much time on stream doing design decisions. I'd rather just uh show implementation.

[01:27:18]So I think what we could do see and this is saying set up Nex.js and superbase. I want to have a corresponding one for each of these failure cases.

[01:27:30]And I think that separating it out into a scan isn't ideal. I think what we want to have is a set because this is also showing pass or fail and the scan won't eventually always do a pass or fail. It might say it's this decision or that decision.

[01:27:55]It's, you know, it has this, you've added this feature, but you haven't added this feature. And there's going to be a whole lot of different checks that we're passing. So, I think I need to iterate on exactly how this is going to work. Um, I think that's a good a good place to continue next time. So, I'll do that design offline. It seems to be a pretty quiet stream today anyway, so I'll probably wrap it up pretty soon.

[01:28:20]Um, what I'd like to work on for now, uh, is Eco3D. So, I want to try and get this going. Um, but first, I might as well follow up on this. just saying that it says zero out of 10 passing some having question marks rather than nex uh rather than access to them. What does that mean?

[01:28:40]Because they don't seem to be interactive. I can't click on them.

[01:28:46]Yeah.

[01:28:49]And what I'll need to do is create uh workflows for each of the missing things. So rather than just having nextjs plus superbase setup, we'll have one at a time. And if it shows that that's not there, then that'll be like the only thing that's actually suggested. It's like before you do anything, we have to use uh next.js.

[01:29:12]Um or it will be like it'll show that as the recommended and then say these are the other things we can do, but I'll need to think about it. Um, but these are the kind of design decisions that I like to do ahead of stream.

[01:29:27]>> Okay, so the root cause was in this. All right, so this was just fixing up how the the messages are displayed here. So now all of these should have full words.

[01:29:38]So this is a full word here. Um, let's try sending it. I think we got three, then two, then that.

[01:29:50]Sweet.

[01:29:52]It is like you do learn it. It is teaching you just very slowly and it's the same word. They should always be different words.

[01:30:02]>> I did that too close. See?

[01:30:09]>> Okay. Nice. Um each word uh for each plants each word should always be different. Um for each segment each segment each word for each segments of each PL plant should always be different uh different once a segment is grown.

[01:30:43]It should keep the same word uh the same word assigns to it but um a different segment should have a different word.

[01:31:03]And um level one segments should have a oneletter word uh level two equals two letters uh etc. And I think it's already got that rule.

[01:31:24]But see oh I that it doesn't. I I thought that it would already have that rule but uh it doesn't. So, let's do I.

[01:31:33]And now we should grow a fruit.

[01:31:40]So, the fruit still isn't dropping over to the other side.

[01:31:45]So when I grow a plant and then grow a fruit and it drops to the other side to the other side, the fruit disappears.

[01:32:06]Disappears.

[01:32:08]It should persist on the ground of the other players of the other players's um world.

[01:32:21]And the other player should be able to pick it up.

[01:32:26]Pick it up um by stepping on it, by landing on that location.

[01:32:38]Let me just There we go. Okay. So, I'm going to copy this even though I don't think we need to because I can just resend it. But I'm going to copy it, clear it, and then say continue. And then hopefully Okay, so I did need to copy it. I'm glad I did.

[01:32:55]Uh, and I'll copy it using control shiftv. Ah, it's actually on my my other user has that shortcut set up. the other user on this computer or the other sign in whatever you refer to it as.

[01:33:09]Okay, so um the question icon means intermediate.

[01:33:15]The scanner couldn't decide. The intent was for that to only fire on transient read failures. 5xx on GitHub a timeout for your brand new empty repo. There's just no package JSON at all. The fix is okay. So, it's basically just allowing for absent files to fail and rather than showing as like a complex error. So, let's refresh.

[01:33:44]And now it looks exactly the same.

[01:33:50]Um after a rescan of your empty repo, you should see vite installed because the check is not on vit is a pass.

[01:34:03]No. Um so we've got like still heaps of question marks.

[01:34:09]This is specific to the scanning logic though, which I guess is good to to work out. And it's not a design decision just yet. So, I'm actually thinking maybe we should start by stripping back to the very basics here. Um, and just bench most of these scans and just start from scratch and build up from there. So, let's go. Uh, let's remove most of the build.

[01:34:42]Let's remove most of the scan stuff.

[01:34:47]Uh, and the let's remove all of the scan checks except for the Next.js build, I think.

[01:35:07]>> Yeah, we'll start with Next.js, the NextJS installation because that's always where we begin.

[01:35:13]Yeah, remove all the scan checks except for the next JS installation.

[01:35:19]Um, it's still it's still uh returning the check as a question mark for a new repo.

[01:35:36]It should be X.

[01:35:41]And then I'll say remove all of the existing um user action prompts from the scan.

[01:35:54]And then I'll just say add one user action prompt.

[01:36:00]We don't want to say prompt. user action for installing um installing Nex.js that will show when that check fails.

[01:36:15]Start with the smallest and then we'll build up from there. Um and I'll just keep, you know, making sure each of these checks works properly and that we're we're building the workflow from the ground up. One thing at a time. I can't read that, dude. I don't know why it does this. I think I actually don't know that would be a total guess.

[01:36:36]Doesn't take too long to refresh it though.

[01:36:40]All right. So the next step when a foot strain exceeds that. Okay. So on the equals train question they're the same problem. The the equality is real in symmetric. So magnitude alone.

[01:36:55]Oh no. This is symmetric. So magnitude alone can never decide which footsteps first. The fix isn't a separation task.

[01:37:04]It's it is a tiebreaker baked into step five. Just going to make sure that I'm still playing from this same playlist because I don't want to accidentally play non-copyright music. I can just start playing a new one from here.

[01:37:29]Or I should say copyrighted music. Make sure I'm not playing that. Okay.

[01:37:35]Um Okay. So, the solution, the fix isn't a separate task. It is the tiebreaker baked into step five. Use a signed direction value, not the magnitude. Oh, okay. Concretely, each foot's signed strain is the projection of its hip rotated displacement into the foot's lateral axis. On a left turn, the right outside foot needs to slide forward and the left inside foot needs to slide back. The foot whose signed strain stays slide forward steps first because lifting it lets the hip rotate without dragging outside without dragging the outside foot through the ground.

[01:38:29]So the resolve is part of step five.

[01:38:32]Okay. So the next step is step five. So okay do that. So, it's going to make it step by step, one thing at a time.

[01:38:45]And this is Oh, wait. No, this is a different thing. Okay. So, beep.buzz.

[01:38:51]And this is the next prompt that I had ready to go.

[01:38:56]Um, okay. So, each segment of a plant should be grown by a unique word. Yeah. So, we're just doing incremental improvements. We got these plants here.

[01:39:07]Here, the rabbits are kept at bay.

[01:39:11]They're staying outside.

[01:39:13]We can shoe them away.

[01:39:18]We need to improve how they work, but it's it's kind of cool. I like it. Okay, so we can send a fruit. Uhoh. Oh, it's okay. It's cuz it's building it.

[01:39:31]All right. So, all that's thinking and that's all they're all thinking. I'm going to wait for this to finish before I restart this terminal to fix the formatting issue. I need to figure out why this formatting issue is going on.

[01:39:47]But in the meantime, I'm going to figure out how to play this game.

[01:39:54]Oh, is that how you get is do you just have to I'm not sure, but I'm just like going through the tutorial here.

[01:40:02]Uh, we need materials. Yeah, there we go. To build room, send pets on adventures. So, drag them to the adventure area in the toolbar. Ah, that's what that is. Let's do that. I'll I'll show you the toolbar that I'm doing that I'm using. Drag them into here.

[01:40:19]Now, they're on an adventure. How clever. Okay, so they're on an adventure. I'm going to leave that down the bottom.

[01:40:28]Um, okay. I'm going to kill this terminal, open up a new one, and uh create the new clawed session so it has that problem resolved.

[01:40:44]Okay.

[01:40:48]Okay. while your repo still showed that uh the scan result is cached um action you needed to take hit the rescan icon. So even refreshing the page still used the cache response.

[01:41:03]So if we hit this then it will rescan and now we're getting so no next.js installed um install Next.js set up Nex.js with the See, but this doesn't look like a button to me.

[01:41:17]Um, I also I don't like how we've got this chat title right up here. I don't think we need that at all.

[01:41:27]So, I'll think about the UI later. Let's start thinking about just what happens when we click this button. It looks like it changes to a different Okay, so it's using a different coach avatar, which I don't really like, but it does make sense. If we talk to me, like if I go message as, then you get the avatar with me. I've got to get a better photo.

[01:41:55]That's not a very good photo, but um it's the same kind of thing, but I like this better.

[01:42:06]Um I need to think about the UI. If I click this in here, it still goes to the task.

[01:42:16]So, it's all split up. I'm going to have to untangle this um offline, but we are making progress. We're making progress.

[01:42:25]So, I'm going to put that on the back burner for now. I only got 10 minutes left. So, I'll wrap up in about that time. Probably wrap up a little bit earlier.

[01:42:33]So, I'm going to leave this for now. I'm just going to say push a commit with all changes.

[01:42:41]Uh, Eco 3D. I really want that to work, but it is taking so long. I compacted the conversation.

[01:42:48]Um, and when it says thinking, it's like before alt V to paste images. Oh, alt v paste images from the clipboard. That's good to know because I was wondering why it wouldn't paste unless I copied the path.

[01:43:08]I don't know why it takes so long to think this up. I get impatient and then I end up like canceling it and starting it again. But I'll just I'll let it do its thing.

[01:43:20]And they're all thinking.

[01:43:26]Let me check the git status for recent commits to see what's currently in the codebase. Why? Let me verify the current state of the fruit rendering and pickup logic.

[01:43:37]I swear it slows you down. It like if you do too much prompting, it just makes it take longer.

[01:43:49]because there's no context bloat or anything like that. So, I don't know why it would take so long.

[01:44:03]Okay. So, this dude when he comes back, is he going to have stuff for us?

[01:44:10]No. Do I send you off again? Dispatch allowed? Do I send like multiple swap pets? Oh, nice. He's got like a different environment. So, I don't know how to play this game properly yet.

[01:44:26]Although, I don't really know if there's much to really play.

[01:44:32]Okay, so this is committed um and pushed. So, I'm going to wrap up as anything.dev for now. And both of these are just sitting on spinning wheels reading two files. Let me actually reread the most recent state of the relevant code to spot anything unusual.

[01:44:54]Let me see if it's working.

[01:44:59]Uh I need a wave of uh speeding this along doing like a dev dev work dev workflow for this.

[01:45:09]Okay, so that grows.

[01:45:14]All right, I'm going to wrap it up there. If anyone happens to be watching, thank you so much for watching. Um, I really appreciate the votes I got tonight. I got six votes. I think maybe this stream was a little bit more boring, but it's I guess it's a bit of a boring stream until we have a lot of viewers.

[01:45:31]So, I'll just keep up the consistency, keep making progress on these apps. I think also once uh I have as anything going then this will be a place where I can actually help people to code and like show them directly how to vibe code through this and get insight into their projects and then we can have like an onstream chatting thing about what we're actually doing. But yeah, I'll leave that there for today. We made a little bit of progress um in Eco 3D. The head and neck are now turning to look at the attractor. And next, the shoulders going to are going to turn and the the legs are going to step. And I look forward to that. So, thanks for watching. Uh my name is as anything and as