35 kLOC is quite a bit. I wonder how straightforward and maintainable this app ended up to be. This would require taking a look at the sources. While good Rails code tends to be very terse, frontend may be quite voluminous.
> I believe within a couple of months, when things like log tailing and automated testing and native version control get implemented
This sounds a bit too optimistic, especially around automated testing, but yes, eventually this all will be there.
> an extremely powerful tool for even non-technical people to write production-quality apps
But why would non-technical people would even think in terms of log tailing and version control, any more than they think about the gauge of wiring in their walls, or the kind of modulation their Wi-Fi device use? For really non-technical audience to make a good use of such tools, it won't just take the AI to be a competent coder. The AI should become a competent architect and a competent senior SWE to translate from the product management language to the software development language, without even surfacing it when not explicitly asked. It's going to be quite a bit of a challenge to make it work, and work about as reliably as with a human team.
Since I wrote the blog post, I actually added automated tests which Windsurf / Claude Code runs before committing any change. It saved me a lot of headaches when the LLM decided to make random logic changes for no reason.
The LLM swaps out business logic without you knowing it? As someone who worries about determinism and doesn’t do much vibe coding, I am curious what’s going on here, thanks!
Absolutely, very often. It typically happens because it's trying to do more than you asked for.
Example scenario: you have a codebase that you iterated on with LLM, and it contains let's say 15 features with various implementation details. You continue tomorrow and want to make a small change to handle an edge-case. While making changes to one of the 3 required files, suddenly it will decide to also rewrite / "improve" other parts of the code that have nothing to do with your request, where pieces of previous logic will no longer exist - since it made 18 changes to the file, and there are 3 such files, good luck spotting it without thorough and detailed change review.
Also, if you made manual changes to generated code and than you ask it to add something to the code (within the same "conversation" / context, it will often replace your changes with how it originally wanted code to look like.
"Let's take a different approach. Instead of trying to fix the existing file incrementally, let's rewrite it completely with a simpler approach that uses the --print-config flag to test the configuration:"
Not 35k lines, but there's one line that's about 48,000 characters long. So either they've intentionally obfuscated it, or chatgpt just churned out one long line
Edit: Running it through a prettifier the code comes about 33k lines
> 35 kLOC is quite a bit. I wonder how straightforward and maintainable this app ended up to be. This would require taking a look at the sources. While good Rails code tends to be very terse, frontend may be quite voluminous.
I think people will have to recalibrate on this. The LOCs do things that you otherwise would not do. Features and details that simply would not happen — because they are too code/time intensive for most projects. It just won't matter anymore.
> But why would non-technical people would even think in terms of log tailing and version control
They won't! They won't have to. The obvious good stuff that everyone thinks the AI tool should be able do, will just work, because the people building the tools, will mostly obviously focus on making them work.
That or we're completely ignoring principles like DRY and producing a head spinning amount of tech debt.
I can't really imagine producing that much code in that short amount of time and holding any amount of it in my head. I'd bet money there's code in there that does the same thing but different, leading to all kinds of little inconsistencies that make this code worthless in any serious context.
I’m an old school coder. I haven’t vibed. I’m guessing the “who is going to maintain this” could only be another tool similar to what built it. I can’t imagine trying to get my head around that much code in a timeframe acceptable to someone demanding product. So are we headed to a place where everything is written and maintained this way or a lucrative future of having to get these things to work properly.
I also wrote a recipe simplifying website, so this is a fun project to see. Incredible work!
Probably the main value engineers have for a maintenance project is context. I wonder what happens when we fully cede context to the machines...
Today, I got a request at work for a feature ("let's offer coupons!") that I thought would take a week. That was until I found out that another engineer wrote most of the code last year, and it'd take him a day to dust off.
I'm totally onboard with, and grateful for, larger-scale experiments like this...thanks for putting the effort in. I wonder how well Cursor (or similar) would handle a situation in which large amounts of code are _almost_ being used. What if 3k LOC accidentally get duplicated? Can our automated systems understand that and fix it? Because if they can't, a human is going to spend a _long_ time trying to figure out what happened.
Over the next 18 months, I expect we'll hear a few stories of the LLM accidentally reimplementing an entire feature in a separate code path. It's a whole new class of bugs! :D
I have similar thoughts and I have come to conclusion that that's the beauty and the curse of this technology. If one relies too much on it - it's gonna be a curse. However, if technology is used with care - it's a beauty. Not only does it keep SWE jobs "secure" - it really helps a lot for those who know what they are doing.
I think in the end AI will be more advanced tool, but a tool nonetheless. Like methodologies and principles, good practises etc. - they only work if you use it with care and added thought and adaptation to your case. DRY it a great principle. But sometimes it's better if you repeat yourself. For one reason or another. And these are the the tradeoffs that human in the loop should be making imho.
> I have similar thoughts and I have come to conclusion that that's the beauty and the curse of this technology.
I agree. When I read these articles on vibe coding I can't help to think that these guys are basking in the glory of the impressive maze they built around themselves. Of course running these things in production and having them reach the state of legacy code is an entirely different thing. Building a maze is one thing, having to run around it is an entirely different challenge.
It's like one of those world expos: everything looks fantastic, but the moment the event ends everything just crumbles.
> using OpenAI's realtime API for voice-to-voice responses
Does it mean it use this expensive open ai audio model in the app? Don't you worry this will make it bankrupt if app goes viral and not monetised?
Can you share what's your strategy here, like topup $2000 open ai account as kind of marketing expenses for users to try for free? Genuine questions since planning to use openai audio API in other case and this kind of expensive price worry me a lot even if switching to new mini-transcribe and mini-tts
I have a paid subscription to Windsurf using Claude. I struggle to find much utility in terms of actually writing code. It spends a lot of time trying / retrying and glitching out. I'm sure in a couple of years it will be amazing (assuming a fundamental breakthrough isn't required) but it isn't quite there yet.
LLMs are super useful but currently, the primary use case is teaching, not doing. For this reason, I think ChatGPT is really just as good as an AI enabled editor (or both if you don't mind paying for two subscriptions).
To correctly vibe code something useful, I find that I need to religiously give faith to the tool and forget all about SWE principles and best practices and instead treat it like a child who makes mistakes and corrections while the adult must not intervene or admonish too much but rather nudge it to right direction.
Also vibe code has a parallel feature, while the code is generating, you are also doing live review and correcting it towards right direction, so depending on your experience, the end product can be a bad mess or wonderful piece of creation and maintenance dream.
The issue with seasoned SWE is that, the moment a mistake(or bad pattern) is made, the baby is thrown with bath water.
For a tiered app like the one presented, 35k LOC is not really that impressive if you think about it. A generic react based front end will easily need a large number of LOC due to modular principle of components, various amounts of hooks and tests(nearly makes us 25-40% of LOC). A business layer will also have many layers of abstractions and numerous impl. to move data between layers.
The vibe code shines, when you let it build one block at a time, limit the scope well and focus. Also, 2-3 weeks is a lot of time to write 35k LOC. at start of any new project, LOC generation rate is very high. But in maintenance phase it significantly falls as smaller changes are more common.
I've had a bit more success on the front end as it is possible to see the results of a change very quickly. In fact I would prefer it to just auto-apply the changes and I would visually inspect. It isn't bad but the workflow is pretty slow. The resultant code is also very verbose - likely 3X more code than an experienced engineer would create (this is one part that I am certain will improve dramatically in time). While I do use this workflow, sometimes I feel like I am just being lazy as opposed to productive.
I'm just being honest. For my use case, I would be much better off if LLMs could just do everything.
Lots of apps are quite repetitive: for building APIs for example you generate one controller and the ask the app to generate more using the first ones as a pattern. For frontend you do the same for forms or lists.
Tests are often quite good, but I think they were already great even back in the first ChatGPT release.
With this strategy and the fact that some patterns are quite verbose (albeit understandable for an AI or a reader), it is quite easy to get to a big LoC while still maintaining consistency.
> LLMs are super useful but currently, the primary use case is teaching, not doing.
For code? Autocomplete on steroids is the killer-app.
The other things the LLMs give me are prone to be over-engineered/overly verbose code or similar.
I went through a lot of "Why are you also doing $FOO then $BAR? Doesn't seem necessary if we skip them and do $BAZ which will make one or both of those redundant" and it responding "You're right! Lets use $BAZ instead".
And giving them code to make a small change to was pointless - they would often, but not always, make an incidental change far from the point where you asked for the change.
But autocomplete? That works just great and because I've already got context of the code I am writing I can check it in (at most) two seconds and move on.
I think you're missing the value of rubber-ducking with the chatbot on a fairly well-defined/well-known problem in a somewhat specific/novel scenario—this is the vast majority of software development being done today.
Depending on the situation this can be invaluable. If you're experienced in the domain you probably know generally what you need to do but you might get a better result by reasoning through the best solution with the constraints and requirements you have. Or maybe you missed something obvious when you write out the full context—which is a required step for getting a good output from the chatbot, and generally isn't a required step if you aren't explaining your approach to someone else and you don't want to be rigorous.
> I think you're missing the value of rubber-ducking with the chatbot on a fairly well-defined/well-known problem in a somewhat specific/novel scenario—this is the vast majority of software development being done today.
I actually do use ChatGPT for rubber-ducking, but in that context there is no (or very little) code. In a coding context, I've resigned myself to purely autocomplete-on-steroids.
The thing is, in the vibe-coding context (having the LLM write the code for you), I've had atrocious results across all of the popular LLMs.
This does seem like something that is worth getting to the bottom of. Why are some people having such good results while others do not. What is the secret sauce?
you need to be clear on what you want the LLM to do
break down the tasks and give byte sized tasks to llm to do specific thing and sometimes I had to tell it not go and change random files because it found the need to refactor them.
Full Disclosure: I'm building a LLM chat app for software developers and domain experts
> I struggle to find much utility in terms of actually writing code.
I personally feel you need to give up some control and just let the LLM do its thing if you want to use it to help you build. It honestly does a lot of things in a more verbose way and I've come to the conclusion that it is an LLM writing code for another LLM. As long as I can debug it, I'm okay with the code, as I can develop at a pace that is truly unreal.
I finished my "Recent" contexts feature in a half a day, today. Without the LLM, this would have taken me a week I think. I would say 98% of my code in the past few months has been AI generated. You can see a real life work flow here:
I truly believe if you provide a LLM with the right context, it can meet your functional specs 90% of the time. Note the emphasis on functional and not necessary style. And if *YOU* architecture your code properly, it should be 100% maintainable.
I do want to make it clear that what I am doing right now is not novel, but I believe most problems are not. If the problem is not well understood, it can be a challenge like my my chat bridge feature. This feature allows you import Git repos for chatting but I will probably need to rewrite 50% of the LLM code since the solution it built is not scalable.
On a more serious note: I've found that for debugging difficult issues, o1 Pro is in a league of it's own.
Claude Code's eagerness to do work will often fix things given enough time, especially for self-contained pieces of software, but I still find myself going to o1 Pro more often than I'd expect.
A coworker and I did a comparison the other day, where we fired up o1 Pro and Claude Code with the same refactor. o1 Pro one-shotted it, while Claude Code took a few iterations.
Interestingly enough, the _thinking_ time of o1 Pro led us to just commit the Claude Code changes, as they were both finished in around the same time (1 min 37s vs. 2+ minutes), however we did end up using some feedback from o1 to fix an issue Claude hadn't caught. YMMV
Is the key feature here just the voice control? I’m wondering what you feel is missing from other popular recipe websites/trackers, and why I would choose to use this over something with more care put into it.
Or, was this mostly just an exercise in engineering/testing AI?
Why would I want to do that when I can go to one of dozen AI chat interfaces and ask for recipes and further customization without having to look at any interface?
Could you 'vibecode' a website that tells us where to get the ingredients? I don't know where around here I'm going to find some werewolf-harvested honey...
While I'm not going to make any food according to the recipes yet, the recipes gave me good chuckle. I'm always amazed by the creativity people have, a feeling that dissipates the moment I know something is AI generated. Comprehensive JavaScript Tutorial https://www.recipeninja.ai/recipe/r_VDcP5QiJjAPVhV/comprehen... is my favourite so far.
I think the recipes on your site are AI-generated? If you're looking to generate a large number of recipes, I highly recommend my tool: https://github.com/Yiling-J/tablepilot. It's specifically designed for this kind of task. Don't forget to check out the examples folder, I use recipes as examples a lot.
I still haven't done the AI vibe coding, but I think I get a similar effect with the Conjure plugin for Neovim with Clojure.
Being able to type out and immediately execute it directly in the window, and even have your code replaced by the output, is kind of life-changing. It fundamentally changes the way you write code, like the REPL isn't just a quick way to test your code, but a direct helper to test the stuff that you right.
I did a project in Clojure recently, heavily using Conjure, and then my next project was in Rust. Rust has nice Neovim plugins as well, but it still kind of felt like a step backwards; I found myself reaching for the "automatically evaluate" keystrokes that don't exist on Rust.
It is so hard to reconcile my view of Karpathy as a world class instructor who does great things like making you work out gradients via the chain rule even though you don't need to anymore with this version of Karpathy.
But I really detest the ever-growing universe of unnecessarily obscure slang. It's not clever and not informative. Good slang is akin to a word that you can figure out from its Latin origins; it makes sense, might be clever, and you can figure it out if you think about it.
It's a term to describe using an agentic LLM based assistant to write the majority of the code but without even taking the bare minimum of time to review the output. I laughed when people naively claimed that it was intended for proof-of-concepts only and would never be used in actual "production".
It's a newish term that I have understood to mean coding by letting the AI do nearly all the work, just telling it what you want done in broad strokes. If the code returned seems to "vibe" with what you're trying to do, then accept it.
Contrasted to AI assisted coding, where you would give much more detailed prompts with technical specifications, and read over every line to make sure you understand it before accepting a response.
In theory, vibe coding can let someone with very limited technical expertise build complete apps, so understandably a lot of people are excited by it.
In practice, it doesn't seem like we're there yet. But each new step in AI development leads to people trying again, and it's hard to deny that the results are getting better. I think we're at the stage of where AI image generators were a few years ago. Very much in the uncanny valley.
<quote:esperent>
In theory, vibe coding can let someone with very limited technical expertise build complete apps, so understandably a lot of people are excited by it.
</quote:esperent>
Actually it is booming. In bsky, X and Linkedin, I see another recipe/todo/budget management/profit tracking/SaaS starter template/landing page/people-to-follow-directory etc. being pumped out every single day. Before GenAI, this would be more like one partial feature per user every month, now post GenAI, entire product in weeks or even hours.
I believe, the indie entrepreneurs are making the maximum bang for the bucks with AI codegen compared to any other groups.
Well this is a good start, I tried searching for “Anti-inflammatory” and the app crashed. Likely that scaling is way out of proportion considering it’s here on HN , but do you have any load balancing or caching setup? Debouncing on the inputs?
Nice work! The voice control is actually really, really cool. I was able to say "show me a chicken tikka masala recipe" which got me 3 recipes, and then I said "show me the third one" and it opened nearly instantly. Pretty awesome. I want this for my entire computer!
Nice work! I also made a recipe app, but after seeing the recipes people generated on yours, I'm choosing not to share it at this time.
Seriously though - vibecoding is great. Even better (or only feasible) as engineers who can dive in when we need to.
My app is iOS and I had never done any Swift. I do have AI generation but that was more of a fun afterthought. The main utility is extracting recipes from the web and having a synced shopping list that I can share with my wife.
I understand that this is just kind of a proof of concept exposition of AI capabilities, but I fucking hate it and it is emblematic of the kind of low effort drivel that is infecting the internet. This is totally unusable in every way possible. While technically impressive (that AI can make a pretty-ish and functional website) it is practically useless.
Also, the photos are some of the most un-appetizing, uncanny valley, shit I've ever seen.
It has some basic features that are 'in-house' like a login, everything else seems to be using third party API's. As per the longer article :
> Eleven Labs API for text-to-speech conversion
> streamio-ffmpeg gem for audio file analysis
> Active Storage integration for audio file management
> OpenAI integration for recipe generation
etc. I'm not saying it's not a fun excersize in 'vibe coding', I'm just very curios the quality of the code that was actually produced. Its my feeling project maintainability is something that will be a big pain point in the future is people are relying on these tools.. 35k LOC for this recipe app smells like that to me at least.
> The app’s security is not perfect, but I’m pretty happy with it for the scale I’m at. If I continue to grow and get more abuse, I’ll implement more robust measures.
Not in position to judge the product, but had me thinking...
Perhaps this is 1998 again, when you could earn big money on creating a visitor-counter service, or a guest book service.
Perhaps, now is the time for a lot of smaller projects with AI, that will, in a few years, all be blasted off the market by big corporations and change in trends.
I assume the insane recipes are generated from users’ prompts, so you might want some content warning or moderation. Kinda cool though that it generates recipes for anything, regardless of if it involves food.
Because the app shows entirely unfiltered "crowdsourced" latest recipes which of course are ridiculous given this is public, I first thought this was some high effort april fools joke on vibe coding, AI slop and the insecurity / thoughtlessness of the results.
But there is some quality in it, I can't argue against that
It does not seem very "viral" or income-generating. I know this is premature at this point, but without charging users for the service, is it reasonable to expect to make money off of this?
Among Heroku and OpenAI bills, the project is apparently not optimized for low-cost operation just yet. Should it become popular, this may become an issue.
Interesting... Thanks for putting in the effort. I'm curious, did your previous knowledge of programming help you accept code changes, or do you just keep everything written by agent?
Quick note: In my opinion, the ai-generated food pictures are deep in the uncanny valley. Unappetizingly so. The hollandaise sauce image is disturbing in a way that's hard to describe, which somehow makes it worse.
I don't think I would want to cook and eat AI generated recipes. I don't like to read AI generated texts and I don't like to listen to AI generated music.
You can call me a snob, but I appreciate some things only if they are the result of work and creativity of humans.
> It now includes a fully interactive voice assistant so you don't need to get your dirty hands over your new iPad when you're cooking.
Always floored about the problems people think need fixing. The problem is not that you get your dirty hands on the iPad. The problem is that you want real recipes. You know, things people have actually cooked and found to be good. With real photos of how the result actually looks (instead of what an AI thinks it might look based on the description).
You might be lucky and find these for free someplace on the web. However, those LLMs that "vibecoded" this Rails app for you are now also used to flood the web with garbage recipes, so finding good recipes on the web will become much, much harder than it already is. I browsed through the recipes and could not find a single one that actually looks real, so at the moment, you are just adding to this problem. This is why people still buy physical cooking books. The good ones are made with sturdy, thick paper so that you can get your dirty hands on them. This is what cooking is all about. Only unused cooking books stay clean.
My argument is: the problem is solved. The solution is called: books containing real recipes.
The "problem" people have is: yeah, I want those, but for free. But creating a cooking book is a ton of work and very expensive. You need to first find good, original recipes. Then you need to actually cook each of them to see if they are any good, at least once, sometimes several times to perfect the recipe. And unless you are a photographer as a side-hustle, you need to do this again and hire a professional to make good-looking photos of the result.
Printing the books is also expensive. Good cooking books need to have very good binding so that they can be flattened to show one page for a long time without disintegrating. The paper must be thick and the printing must be able to withstand stains and moisture without becoming unreadable.
I use a delivery service that packages the food and the recipe in a box. The recipe is 1 page long, on thick paper, and I have used it as temporary food storage when the cutting board gets too full.
Who cares if it solves a problem? Look how fast they made it! Bro it’s so easy! 20 minutes! Who needs to know how to code! Who needs skills? We just vibed it!!
Imo, this is not just a vibe coders problem. It's been a classic builders mindset issue, where many start with building and look to find problem to the solution.
Absolutely agree. Real recipes, tested by actual cooks (home or professional) with genuine photos, matter far more to me than avoiding messy screens. Does anyone know trustworthy sites that consistently offer recipes meeting these standards?
Seriouseats, Epicurious, Bonappetit, or anything with Kenji Lopez-Alt (for always going the distance in methodically and empirically testing all variations of whatever he is evaluating/developing).
If you want something in the short form video era, I do appreciate Andy Hearnden (andycooks) as he is both concise, consistent and always posts the full recipe in the video descriptions (all too rare).
BA might have gone downhill, it's been a few years since the whole unfair pay debacle, and I know a lot of good people left, and I haven't checked in as much lately.
If you want just recipes that were published in the magazine (that this website is the companion to), you can also filter results to author = Good Food team
If you like Indian food I’d say Swasthi’s Recipes is decent. A lot of what I have seen on there is similar to how my aunties in Punjab n aunties of my friends in UP make food/have taught me to make food.
I cook every day and love trying new things. There's no reason to pick just one, but if I had to, Chef John is my go-to. I stumbled across him when I was trying out pretty much every birria recipe on the internet, and his (yes, a white dude) is by far my favorite. That pattern seems to repeat with his recipes.
As somebody who cooks quite a lot (for a diverse set of people), I have the opposite experience of what you describe (also I hate having to touch my screen while cooking).
The cooking book scene has been openly criticized for not actually trying the recipes, even before LLMs were a thing[0]. Regular cooking websites have always been somewhat unusable due to massive ads and fluff text because 1) SEO and 2) recipes are not copyrightable, but the fluff text is.
For quite some time I get my recipes directly from chatgpt, the instructions are very condense, they work quite well, and most importantly: It knows how to substitute ingredients. "My friend is vegan and allergic to heat-resistant soy protein" and it's going to adjust accordingly.
What the hell is the "cooking book scene"? You make it sound like some cabal of cooking book writers sitting in some basement, conspiring to lead people astray. Nobody is saying that all cooking books are great, of course there are loads of shitty cooking books, and of course you are free to criticize that, even openly if you dare. I somehow cope by buying the good ones.
My experience with LLM recipes is like with pretty much everything else these things generate: usually very mediocre, mixed with glaring errors in between. If you are an experienced cook, you'll be able to manage since you'll recognize the errors.
In another comment, you say photos are required. I would claim that the cooking book scene is the publisher interest in the limited shelf space for photo-heavy cookbooks. Celebrity and TV chefs thoroughly represented.
Books without photos (McGee, Hewitt, Potter, Julia Child) are not interesting to publishers.
So you're allegedly annoyed with untested recipes and decided to solve this problem with a tool that only generates untested recipes? You're happier to cook or bake something from a wall of text that wasn't authored by anyone with any amount of culinary experience, that nobody has ever read before, attempted to replicate, or publicly critiqued, why?
I don't buy it, that's going to be a strictly worse experience than trying out the first search result that has photos, is rated 4/5 stars or better and that has a few positive comments. You can always ask an LLM for substitution recommendations separately.
Agreed! That was a problem even before LLMs showed up. I found my subscription to Cooks Illustrated/America's Test Kitchen to be incredibly valuable. Bonus: You don't have to read through 15 paragraphs of SEO fluff to get to the recipe.
I'm convinced that people who complain about this don't use recipe sites very often. There are loads without any preamble, and many of those with narrative provide a 'jump to recipe' button. Even if it is a long article with no jump (rare), it's still not a big deal
1. Some people do want the narrative
2. It's not that much of a hardship to scroll
You're entirely right. I don't use recipe sites very often, and yet it still happens often enough that I have to keep scrolling and reading on different sites to get past the backstory
I've got a shelf full of physical cookbooks -- The New Best Recipe, The Food Lab, an Ottolenghi book, On Food and Cooking, etc. I like them, I've read through all of them, I sometimes use them for inspiration.
It's rare that I actually cook directly from them -- usually, that'd be big and fancy stuff or stuff I'm very unfamiliar with; in both cases I usually take the time to cross reference whatever the cookbook says with additional resources from the internet.
ChatGPT, on the other hand, I frequently use when or before cooking (and I cook virtually every day).
It's great when I only have a vague idea based on stuff in the fridge; five minutes later I've got a checklist I can reference. If it hallucinates something that I flat out don't think will work or, much more likely, comes up with something that I don't want or cannot do for lack of ingredients or time or whatever, I'll tell it to adjust the recipe and it does.
It's also great when I feed it a couple of existing recipes (from real people) to compare and contrast and integrate and reformat in a way that's most useful to me, e.g. a tabular format, or scaled to a different serving size.
With all that said, the AI based recipe sites don't really do it for me, either. If I want to cook purely AI generated recipes, a chat interface works fine -- and probably better. What I really want is an AI tool that helps me curate my own recipe collection. E.g. I want to ask it "I'd like to make Ramen, how did I do it the last time, what were my notes" and when it's done I want to tell it "ok, this was fine, I decided to double the mirin and next time I'd marinate the eggs longer" and have it update the recipe.
Is the project anything more than a (fairly elaborate I must recognize) 1 April joke? I agree with your point even on a more general level, amateur techies thinking to fix stuff which never needed fixing are nothing unusual. We "disrupt" the economy, the finances, housing and more, every day, while being barely legal and sometimes outright abusive to some part of the society, all in the name of the Saviour Technology. Is this just another example of "solutionism"? Or it's just me waking up grumpy and without coffee...
You don't fancy Cyanide Custard? Radioactive Slime Jello made with actual radioactive waste? Sweet Tooth Delight made with real human teeth? Or a few other desserts with NSFW names that would go down a treat for a family dinner. It is hard to please some people /s.
But yes, websites will now be filled with these low-quality recipes, and some might be outright dangerous. Cyanide custard should ring alarm bells, but using the wrong type of mushroom is equally dangerous and much more challenging to spot.
I don't know why everyone is so salty. It's like putting "vibecode" in the headline attracted the worst in people.
For context, I'm not a cooking geek or virtuoso. I enjoy it to some degree, but mostly it's just about having a nice, nutritious experience, in line with whatever my mood might be. I only ever measure things super accurately when I'm baking things in the bread maker (because it doesn't let you make corrections). For most meals, I wing half the measurements and time estimates.
In my experience, most "human recipes" are just random variations on some baseline. I hate looking for recipes on recipe sites, youtube, etc. There are food bloggers that are exceptions, but usually I'll just end up scrolling for a long time with just frustration to show for it. If I sort of know some of the ingredients I want to use, have some sense of the type of eating experience I'm going for, and I want a bunch of recommendations based on that, regular "human recipe" sites are not the answer.
90% of my new recipes come from ChatGPT, and that ratio is increasing. I just marinated some chicken based on a recipe it magicked for me. I asked for insights on mixing mayo and yoghurt in the same marinade, because I had leftovers of both. It gave me 5 or so diverse recipes, and I just picked the one that best fit my pantry and mood. I also asked it to convert the recipe from volume to weight, not to mention scale it for my specific quantity of chicken, which was super handy.
I find that ChatGPT is great at providing common sense instructions and approximations. It's absolutely awesome at clobbering together a meal from ingredients I tell it I have. I can have an actual dialogue about any of it, get all kinds of recommendations and insights. That's been very useful to me. I'd go as far as to say that recipe generation is one of the easiest real problems for an LLM to solve. Or, at least for the kinds of recipes I use.
I've done my share of recipes from Serious Eats, but they weren't particularly good. I was doing Breton galettes the other week, which are notoriously fiddly to get right. Serious Eats had a huge article about it, interesting insights, but their final recipe sucked, and I was trying to be accurate. Not only I failed to get the consistency right, the wheat-buckwheat ratio was nowhere near what you'd get in France. I say, write researched articles about what makes recipes work. I can read it, I can bounce my LLM off it. If it's a fiddly recipe, I'll have to fiddle with it no matter what. If I can have a conversation with an LLM about principles at work, that's much better to me than a bunch of "human recipes".
Also, I often have questions about alternatives or things I need advice on as I'm preparing food. I'll also look at the recipe a gazillion times to check the instructions, quantities, etc. I'll set and check a timer often too. A voice-assistant is the obvious answer to this, which I'll try at my earliest convenience.
> In my experience, most "human recipes" are just random variations on some baseline.
This just shows your lack of experience and curiosity. The art of cooking is a cultural achievement with thousands of years of history, with vast differences between different countries and cultures. Just visit a store (yes, a real one) with a good cooking book section. You might be surprised.
You're on a discussion forum. Where you express your opinion. Then the other party expresses their opinion. That's what makes this discussion forum great. Sometimes the opinions of the other party aren't going to agree with yours. It's a normal part of discourse on online forums.
You're implying that the only quality a conversation has is (dis)agreement. I have absolutely nothing against disagreement. Why are you saying this via a throwaway?
I read your comment before the edit so please accept my apologies
Saying that posting a simple recipe for cocaine was probably illegal sounded very "puritan american"
As a European (French) myself, I'm surprised that you feel that way because I remember reading in a chemistry book in high-school how to make explosives and our chemistry teacher took the complete timeslot to respond in details to someone who asked if, like Walter White, they could make meth.
Wow, I can't believe how many people seem to be missing the point here and being overly negative. The fact that you can have an ai assistant code something like this by giving it instructions would be utterly miraculous 10 years ago, even less.
Whatever the ultimate usefulness of the website is, the point is using it is slick. It works and it works well.
Very nice demo of vibe coding Tom. I appreciate it.
Thanks for sharaing your journey, I have few observations
1. When I click on a recipe from home page it is maintaining the scroll position, so I am not seeing the top of the screen. Is this deliberate?
2. "Recipe Ninja was vibecoded by Tom in San Francisco." Will it increase of decrease trust in your system for users?
3. To remove AI changing random files, I use "Copy relative path" to tell AI which file to change (there is a keyboard shortcut too). Not fully vibe coding but can be useful for precision bug fixing.
"Still, I was determined not to write any code ... I just sat laughing as the computer wrote code."
I can pop over to Midjourney and be determined not to draw a single line and "sit there laughing" as it draws the Mona Lisa in the style of Salvador Dali but with a turnip instead of a person.
How is this any different? What is ultimately notable about it? Did any of it make you a better programmer?
I'm always deeply impressed when people devote significant chunks of their time to achieving extraordinary results. I'm entirely baffled, however, that there's anything at all interesting about using an AI interface to build an AI interface to connect you to AI slop.
You could have spent 20 hours planting trees or doing some kind of community serivce, and the world would have been a far better place.
Programming is not art. In the end, nobody will care whether software was painstakingly hand-chiseled by humans in a dungeon or "vibe coded", as long as the result is good enough. (I'd argue that art will "suffer" from the same "problem".)
What is notable here is that someone is demonstrating that the systems are reaching a quality where this is possible.
> Did any of it make you a better programmer?
By conventional metrics, if the job got done well enough in less time, yes, even if less skill is involved.
I don't know about that, Roller Coaster Tycoon is fondly remembered and sits on a special place culturally because it was written in assembly by a single person.
I mean I guess kudos to you for being creative and building something, even if its with the so-called vibe-coding. But LoC is not the metric I would use to describe any app I've built, or that I am building.
The only obvious problem here is showing all of them publicly. Let people do silly things but only show the sensible ones instead of being user hostile by default with false positives
Leaving them public has the advantage that no one will think the site is serious, which will help them avoid the recipes that look okay but are poisonous.
By practice. It requires some skill too, to constantly remove slop and keep the project from deteriorating. Usually by suggesting good architecture, asking for tests, and rewriting / cleaning up some bad code endlessly. Still much faster than manual development.
Please don't attack others or their work like this on this site, regardless of who or what you have a problem with. It's the opposite of the curious, respectful conversation we're looking for, and always has been.
A cooking web app 35k LoC? Wow, the author might be surprised to know that an entire OS kernel can be programmed at a fraction of that count. I can safely say that the days where we compained about JS bloat can be put aside, this is a whole new league by itself.
Sane advice: learn to program, put the AI hype/drug aside and do yourself a favor. It's an invaluable lifetime skill knowing to program from scratch and perhaps unassisted coding will be a looked-after skill in the years to come.
I kinda wish I could look at the code, but it's not open source and the frontend sources are minified/obfuscated, but it's long and looks like it's using React and a css-in-react approach, which also generates a ton of code.
The term "vibe coding" gives me bad vibes, I feel that it is in some way an insult to my craft, the way a carpenter might feel about Ikea. But if only LLM's are going to read the code, the LoC aren't a big problem, increasing cost significantly less than if a real person would have to read it.
The two biggest issues I can think of with this approach is performance and security, with the first one only being a problem if you "make it", and the second one being too often ignored with or without vibe coding.
I keep thinking of Visual Basic... It was scoffed at as not real programming... It was the drag and drop thing in the Visual Basic IDE that made it worse. Suddenly "anyone" could do it without really knowing programming. This new trend (if it is a trend) is sort of like drag and drop.
You probably got downvoted because you admitted your failure of using the WWW. The problem you describe only happens on "recipe" websites that are 99% advertising and written like a blog, with every other word set in bold to maximize the ad revenue.
These websites should be avoided entirely. Proper recipe sites include en.wikibooks.org/wiki/Cookbook:Table_of_Contents, based.cooking and of course https://web.archive.org/web/20110323153937/http://www.ottavi...
Impressive!
35 kLOC is quite a bit. I wonder how straightforward and maintainable this app ended up to be. This would require taking a look at the sources. While good Rails code tends to be very terse, frontend may be quite voluminous.
> I believe within a couple of months, when things like log tailing and automated testing and native version control get implemented
This sounds a bit too optimistic, especially around automated testing, but yes, eventually this all will be there.
> an extremely powerful tool for even non-technical people to write production-quality apps
But why would non-technical people would even think in terms of log tailing and version control, any more than they think about the gauge of wiring in their walls, or the kind of modulation their Wi-Fi device use? For really non-technical audience to make a good use of such tools, it won't just take the AI to be a competent coder. The AI should become a competent architect and a competent senior SWE to translate from the product management language to the software development language, without even surfacing it when not explicitly asked. It's going to be quite a bit of a challenge to make it work, and work about as reliably as with a human team.
Since I wrote the blog post, I actually added automated tests which Windsurf / Claude Code runs before committing any change. It saved me a lot of headaches when the LLM decided to make random logic changes for no reason.
Would you mind describing the process for this in a bit more detail?
The LLM swaps out business logic without you knowing it? As someone who worries about determinism and doesn’t do much vibe coding, I am curious what’s going on here, thanks!
Absolutely, very often. It typically happens because it's trying to do more than you asked for.
Example scenario: you have a codebase that you iterated on with LLM, and it contains let's say 15 features with various implementation details. You continue tomorrow and want to make a small change to handle an edge-case. While making changes to one of the 3 required files, suddenly it will decide to also rewrite / "improve" other parts of the code that have nothing to do with your request, where pieces of previous logic will no longer exist - since it made 18 changes to the file, and there are 3 such files, good luck spotting it without thorough and detailed change review.
Also, if you made manual changes to generated code and than you ask it to add something to the code (within the same "conversation" / context, it will often replace your changes with how it originally wanted code to look like.
Do things like Cline's memory bank solve this?
I experienced this, too. It is quite annoying. You have to explicitly tell the LLM to leave everything else as-is.
use conventions.md to have it not do that.
What the fuck
"Let's take a different approach. Instead of trying to fix the existing file incrementally, let's rewrite it completely with a simpler approach that uses the --print-config flag to test the configuration:"
Yeah, without trying to be dismissive I'm a bit unsure where the 35kLOC are going?
The vibes? It was vibecoded.
probably includes npm packages. Actual loc is probably 2-3k.
with npm packages it's probably 35k - Files -
The 35k lines of code is what made me think this was either a joke or the state of vibe coding, but no it turns out they're serious
I have entire codebases of embedded software in C without the shortcuts of modern programming languages in way fewer than 35k lines
I cannot find a link to the source code. Is there any?
https://www.recipeninja.ai/assets/index-B7TAGfYd.js
Not 35k lines, but there's one line that's about 48,000 characters long. So either they've intentionally obfuscated it, or chatgpt just churned out one long line
Edit: Running it through a prettifier the code comes about 33k lines
Oh, this. I thought there was more into it.
Edit: Run it through another prettifier, might turn into 35k lines. :D
It’s obviously obstructed, but throw it into an LLM to attempt to clean it up
> 35 kLOC is quite a bit. I wonder how straightforward and maintainable this app ended up to be. This would require taking a look at the sources. While good Rails code tends to be very terse, frontend may be quite voluminous.
I think people will have to recalibrate on this. The LOCs do things that you otherwise would not do. Features and details that simply would not happen — because they are too code/time intensive for most projects. It just won't matter anymore.
> But why would non-technical people would even think in terms of log tailing and version control
They won't! They won't have to. The obvious good stuff that everyone thinks the AI tool should be able do, will just work, because the people building the tools, will mostly obviously focus on making them work.
That or we're completely ignoring principles like DRY and producing a head spinning amount of tech debt.
I can't really imagine producing that much code in that short amount of time and holding any amount of it in my head. I'd bet money there's code in there that does the same thing but different, leading to all kinds of little inconsistencies that make this code worthless in any serious context.
I’m an old school coder. I haven’t vibed. I’m guessing the “who is going to maintain this” could only be another tool similar to what built it. I can’t imagine trying to get my head around that much code in a timeframe acceptable to someone demanding product. So are we headed to a place where everything is written and maintained this way or a lucrative future of having to get these things to work properly.
Given amount of duplication and variation there is no way to guarantee any level of security in such codebase. The bright future ahead!
Yeah 35k of code is a red flag. There's no reason for this to be that large.
I also wrote a recipe simplifying website, so this is a fun project to see. Incredible work!
Probably the main value engineers have for a maintenance project is context. I wonder what happens when we fully cede context to the machines...
Today, I got a request at work for a feature ("let's offer coupons!") that I thought would take a week. That was until I found out that another engineer wrote most of the code last year, and it'd take him a day to dust off.
I'm totally onboard with, and grateful for, larger-scale experiments like this...thanks for putting the effort in. I wonder how well Cursor (or similar) would handle a situation in which large amounts of code are _almost_ being used. What if 3k LOC accidentally get duplicated? Can our automated systems understand that and fix it? Because if they can't, a human is going to spend a _long_ time trying to figure out what happened.
Over the next 18 months, I expect we'll hear a few stories of the LLM accidentally reimplementing an entire feature in a separate code path. It's a whole new class of bugs! :D
I have similar thoughts and I have come to conclusion that that's the beauty and the curse of this technology. If one relies too much on it - it's gonna be a curse. However, if technology is used with care - it's a beauty. Not only does it keep SWE jobs "secure" - it really helps a lot for those who know what they are doing.
I think in the end AI will be more advanced tool, but a tool nonetheless. Like methodologies and principles, good practises etc. - they only work if you use it with care and added thought and adaptation to your case. DRY it a great principle. But sometimes it's better if you repeat yourself. For one reason or another. And these are the the tradeoffs that human in the loop should be making imho.
> I have similar thoughts and I have come to conclusion that that's the beauty and the curse of this technology.
I agree. When I read these articles on vibe coding I can't help to think that these guys are basking in the glory of the impressive maze they built around themselves. Of course running these things in production and having them reach the state of legacy code is an entirely different thing. Building a maze is one thing, having to run around it is an entirely different challenge.
It's like one of those world expos: everything looks fantastic, but the moment the event ends everything just crumbles.
Help! I just made the Diarrhea Walnuts (https://www.recipeninja.ai/recipe/r_N1VSPtXzCJVV3l/diarrhea-...) recipe and it turns out I am allergic to Walnuts! I am throwing up everywhere! You’re gonna be hearing from my lawyers!
> Preheat oven to 950 Kelvin
I believe I'm going to need a new oven...
You’re going to need a new home as a bonus
they don't call it extreme cuisine for nothing
Pyrolysis Mode can come pretty close on some ovens I think.
Try out the Gluten Free Shark Teeth once you feel better!
https://www.recipeninja.ai/recipe/r_ttOB5xyqpOLXCL/gluten-fr...
This sounds like an order from the Key and Peele "Soul Food" sketch: https://www.youtube.com/watch?v=3zDHSLDY0Q8
For anyone saddened by the recipe being removed: https://web.archive.org/web/20250402064135/https://www.recip...
Thanks, I wanted to give it a try. Today is a good day to cook.
Aw no it's gone!
bro that is part of the vibe, you just gotta roll with it
It feels like Jian Yang and Big Head are making a new app
'Silicon Valley' was a practically a documentary. It's just a gold mine of absurdity out here.
Yeah. The more the years go by the more the oracle writers of the show are proved right. Frightening
> using OpenAI's realtime API for voice-to-voice responses
Does it mean it use this expensive open ai audio model in the app? Don't you worry this will make it bankrupt if app goes viral and not monetised?
Can you share what's your strategy here, like topup $2000 open ai account as kind of marketing expenses for users to try for free? Genuine questions since planning to use openai audio API in other case and this kind of expensive price worry me a lot even if switching to new mini-transcribe and mini-tts
I have a paid subscription to Windsurf using Claude. I struggle to find much utility in terms of actually writing code. It spends a lot of time trying / retrying and glitching out. I'm sure in a couple of years it will be amazing (assuming a fundamental breakthrough isn't required) but it isn't quite there yet.
LLMs are super useful but currently, the primary use case is teaching, not doing. For this reason, I think ChatGPT is really just as good as an AI enabled editor (or both if you don't mind paying for two subscriptions).
To correctly vibe code something useful, I find that I need to religiously give faith to the tool and forget all about SWE principles and best practices and instead treat it like a child who makes mistakes and corrections while the adult must not intervene or admonish too much but rather nudge it to right direction.
Also vibe code has a parallel feature, while the code is generating, you are also doing live review and correcting it towards right direction, so depending on your experience, the end product can be a bad mess or wonderful piece of creation and maintenance dream.
The issue with seasoned SWE is that, the moment a mistake(or bad pattern) is made, the baby is thrown with bath water.
For a tiered app like the one presented, 35k LOC is not really that impressive if you think about it. A generic react based front end will easily need a large number of LOC due to modular principle of components, various amounts of hooks and tests(nearly makes us 25-40% of LOC). A business layer will also have many layers of abstractions and numerous impl. to move data between layers.
The vibe code shines, when you let it build one block at a time, limit the scope well and focus. Also, 2-3 weeks is a lot of time to write 35k LOC. at start of any new project, LOC generation rate is very high. But in maintenance phase it significantly falls as smaller changes are more common.
I've had a bit more success on the front end as it is possible to see the results of a change very quickly. In fact I would prefer it to just auto-apply the changes and I would visually inspect. It isn't bad but the workflow is pretty slow. The resultant code is also very verbose - likely 3X more code than an experienced engineer would create (this is one part that I am certain will improve dramatically in time). While I do use this workflow, sometimes I feel like I am just being lazy as opposed to productive.
I'm just being honest. For my use case, I would be much better off if LLMs could just do everything.
My experience matches yours.
Lots of apps are quite repetitive: for building APIs for example you generate one controller and the ask the app to generate more using the first ones as a pattern. For frontend you do the same for forms or lists.
Tests are often quite good, but I think they were already great even back in the first ChatGPT release.
With this strategy and the fact that some patterns are quite verbose (albeit understandable for an AI or a reader), it is quite easy to get to a big LoC while still maintaining consistency.
> LLMs are super useful but currently, the primary use case is teaching, not doing.
For code? Autocomplete on steroids is the killer-app.
The other things the LLMs give me are prone to be over-engineered/overly verbose code or similar.
I went through a lot of "Why are you also doing $FOO then $BAR? Doesn't seem necessary if we skip them and do $BAZ which will make one or both of those redundant" and it responding "You're right! Lets use $BAZ instead".
And giving them code to make a small change to was pointless - they would often, but not always, make an incidental change far from the point where you asked for the change.
But autocomplete? That works just great and because I've already got context of the code I am writing I can check it in (at most) two seconds and move on.
I think you're missing the value of rubber-ducking with the chatbot on a fairly well-defined/well-known problem in a somewhat specific/novel scenario—this is the vast majority of software development being done today.
Depending on the situation this can be invaluable. If you're experienced in the domain you probably know generally what you need to do but you might get a better result by reasoning through the best solution with the constraints and requirements you have. Or maybe you missed something obvious when you write out the full context—which is a required step for getting a good output from the chatbot, and generally isn't a required step if you aren't explaining your approach to someone else and you don't want to be rigorous.
> I think you're missing the value of rubber-ducking with the chatbot on a fairly well-defined/well-known problem in a somewhat specific/novel scenario—this is the vast majority of software development being done today.
I actually do use ChatGPT for rubber-ducking, but in that context there is no (or very little) code. In a coding context, I've resigned myself to purely autocomplete-on-steroids.
The thing is, in the vibe-coding context (having the LLM write the code for you), I've had atrocious results across all of the popular LLMs.
This does seem like something that is worth getting to the bottom of. Why are some people having such good results while others do not. What is the secret sauce?
I have very similar experience.
After seeing how people like Andrej Karparthy used vibe coding to generate applications https://x.com/karpathy/status/1903671737780498883?s=61 I realize that
you need to be clear on what you want the LLM to do break down the tasks and give byte sized tasks to llm to do specific thing and sometimes I had to tell it not go and change random files because it found the need to refactor them.
Full Disclosure: I'm building a LLM chat app for software developers and domain experts
> I struggle to find much utility in terms of actually writing code.
I personally feel you need to give up some control and just let the LLM do its thing if you want to use it to help you build. It honestly does a lot of things in a more verbose way and I've come to the conclusion that it is an LLM writing code for another LLM. As long as I can debug it, I'm okay with the code, as I can develop at a pace that is truly unreal.
I finished my "Recent" contexts feature in a half a day, today. Without the LLM, this would have taken me a week I think. I would say 98% of my code in the past few months has been AI generated. You can see a real life work flow here:
https://app.gitsense.com/?chat=eece40e2-6064-46d2-9bf1-d868c...
I truly believe if you provide a LLM with the right context, it can meet your functional specs 90% of the time. Note the emphasis on functional and not necessary style. And if *YOU* architecture your code properly, it should be 100% maintainable.
I do want to make it clear that what I am doing right now is not novel, but I believe most problems are not. If the problem is not well understood, it can be a challenge like my my chat bridge feature. This feature allows you import Git repos for chatting but I will probably need to rewrite 50% of the LLM code since the solution it built is not scalable.
Glad you finally found Claude Code useful, Tom ;)
On a more serious note: I've found that for debugging difficult issues, o1 Pro is in a league of it's own.
Claude Code's eagerness to do work will often fix things given enough time, especially for self-contained pieces of software, but I still find myself going to o1 Pro more often than I'd expect.
A coworker and I did a comparison the other day, where we fired up o1 Pro and Claude Code with the same refactor. o1 Pro one-shotted it, while Claude Code took a few iterations.
Interestingly enough, the _thinking_ time of o1 Pro led us to just commit the Claude Code changes, as they were both finished in around the same time (1 min 37s vs. 2+ minutes), however we did end up using some feedback from o1 to fix an issue Claude hadn't caught. YMMV
I think this needs a NSFW added to the title, on the first page at the moment there are at least 50% of NSFW recipes.
Is the key feature here just the voice control? I’m wondering what you feel is missing from other popular recipe websites/trackers, and why I would choose to use this over something with more care put into it.
Or, was this mostly just an exercise in engineering/testing AI?
Hands-free voice control and being able to access recipe ingredients and steps without 5 pages of SEO-optimised prose.
The real problem of your website is that the recipes are AI generated.
A second, minor problem of your website is that the images illustrating recipes are AI generated with a bad quality
You can't solve those issues by throwing more AI.. well maybe the second problem you can (AI images with later models are generally ok)
They just haven't done the "now monitise it" step in the vibe-coding journey.
with all respect to the man himself, not sure tom blomfield needs to think about that part..
It’s ok! He’ll just ask the AI! It’ll solve it for us!
Why would I want to do that when I can go to one of dozen AI chat interfaces and ask for recipes and further customization without having to look at any interface?
Could you 'vibecode' a website that tells us where to get the ingredients? I don't know where around here I'm going to find some werewolf-harvested honey...
https://www.recipeninja.ai/recipe/r_XbZvrH23kS6FwN/werewolf-...
While I'm not going to make any food according to the recipes yet, the recipes gave me good chuckle. I'm always amazed by the creativity people have, a feeling that dissipates the moment I know something is AI generated. Comprehensive JavaScript Tutorial https://www.recipeninja.ai/recipe/r_VDcP5QiJjAPVhV/comprehen... is my favourite so far.
"White Power Waffles"? You may want to double-check your examples.
https://www.recipeninja.ai/recipe/r_0a8wYxMgm1zFSw/white-pow...
https://news.ycombinator.com/item?id=43554945
I love how the top recipe shows a literal stuffed lamb, e.g. a plush toy. Lots of the generated images have an early deep dream vibe
I can tell. Ingredients. 1 skillet.
Is this a serious thing? If so, I'd like to know where I can purchase the wasp meat required to make this recipe: https://www.recipeninja.ai/recipe/r_NYrbOKZEyqrp7r/wasp-meat...
How about some dog sashimi? https://www.recipeninja.ai/recipe/r_I7a0kQHzgSexi8/dog-sashi... Maybe a human brain? https://www.recipeninja.ai/recipe/r_A6EsANYZ4uce2n/human-bra...
Care to taste some "cyanide ice cream"? [0]
[0] https://www.recipeninja.ai/recipe/r_WHeXRD7qXHV0Vr/cyanide-i...
Crunchy Plastic Bites in Windshield Fluid, anyone?
https://www.recipeninja.ai/recipe/r_3i5lSfWjUKq05m/crunchy-p...
80 liters of human diarrhea would be a big problem:
https://www.recipeninja.ai/recipe/r_N1VSPtXzCJVV3l/diarrhea-...
Is the AI also generating the contents of the recipes? Will it be Google telling us to eat rocks[1] again?
[1]: https://www.bbc.com/news/articles/cd11gzejgz4o
Also wow, pho from scratch in 30 minutes? https://www.recipeninja.ai/recipe/r_f0IwybXaymnGzp/spicy-pho...
diarrhea walnuts - https://www.recipeninja.ai/recipe/r_N1VSPtXzCJVV3l/diarrhea-...
What is a good place to find and purchase rat meat? https://www.recipeninja.ai/recipe/r_8cj5RcE1Bl0lt7/rat-burge...
There's also apparently a hairstyle? https://www.recipeninja.ai/recipes/hairstyle
This is the funnies thing I have read all week. :D
The Japanese tradition of raising and eating wasps https://www.splendidtable.org/story/2019/02/08/the-japanese-...
I think the recipes on your site are AI-generated? If you're looking to generate a large number of recipes, I highly recommend my tool: https://github.com/Yiling-J/tablepilot. It's specifically designed for this kind of task. Don't forget to check out the examples folder, I use recipes as examples a lot.
I still haven't done the AI vibe coding, but I think I get a similar effect with the Conjure plugin for Neovim with Clojure.
Being able to type out and immediately execute it directly in the window, and even have your code replaced by the output, is kind of life-changing. It fundamentally changes the way you write code, like the REPL isn't just a quick way to test your code, but a direct helper to test the stuff that you right.
I did a project in Clojure recently, heavily using Conjure, and then my next project was in Rust. Rust has nice Neovim plugins as well, but it still kind of felt like a step backwards; I found myself reaching for the "automatically evaluate" keystrokes that don't exist on Rust.
I type "lasagne", hit enter, some things pop up and disappear before I can read them, and then I get a solid white screen. (in Firefox)
Hmm it seems like recipe generation might be broken on Firefox. I'll look into it - thanks! It's working in Chrome
Search is working fine, I just tried https://www.recipeninja.ai/search?name=lasagne
[flagged]
Programmers are also human just came out with this:
Interview with Vibe Coder in 2025 https://youtu.be/JeNS1ZNHQs8?si=kQIVpEBUwK3pNvRw
I think this helps to understand the mindset of a vibe coder better
What is "vibecoded" supposed to mean?
See https://twitter.com/karpathy/status/1886192184808149383 for when it was coined by Andrei Karpathy.
I've written more about it here: https://simonwillison.net/2025/Mar/19/vibe-coding/ and here: https://simonwillison.net/2025/Mar/23/semantic-diffusion/
It is so hard to reconcile my view of Karpathy as a world class instructor who does great things like making you work out gradients via the chain rule even though you don't need to anymore with this version of Karpathy.
They're the same person. He's just a techno-optimist and interested in teaching.
Thanks for the reply.
But I really detest the ever-growing universe of unnecessarily obscure slang. It's not clever and not informative. Good slang is akin to a word that you can figure out from its Latin origins; it makes sense, might be clever, and you can figure it out if you think about it.
This is just douchey.
It's a term to describe using an agentic LLM based assistant to write the majority of the code but without even taking the bare minimum of time to review the output. I laughed when people naively claimed that it was intended for proof-of-concepts only and would never be used in actual "production".
It's a newish term that I have understood to mean coding by letting the AI do nearly all the work, just telling it what you want done in broad strokes. If the code returned seems to "vibe" with what you're trying to do, then accept it.
Contrasted to AI assisted coding, where you would give much more detailed prompts with technical specifications, and read over every line to make sure you understand it before accepting a response.
In theory, vibe coding can let someone with very limited technical expertise build complete apps, so understandably a lot of people are excited by it.
In practice, it doesn't seem like we're there yet. But each new step in AI development leads to people trying again, and it's hard to deny that the results are getting better. I think we're at the stage of where AI image generators were a few years ago. Very much in the uncanny valley.
<quote:esperent> In theory, vibe coding can let someone with very limited technical expertise build complete apps, so understandably a lot of people are excited by it. </quote:esperent>
Actually it is booming. In bsky, X and Linkedin, I see another recipe/todo/budget management/profit tracking/SaaS starter template/landing page/people-to-follow-directory etc. being pumped out every single day. Before GenAI, this would be more like one partial feature per user every month, now post GenAI, entire product in weeks or even hours.
I believe, the indie entrepreneurs are making the maximum bang for the bucks with AI codegen compared to any other groups.
With 35K LoC for a simple website, I'd consider this slopcoding
Diarrheactive™
"Hey AI build me a node app that does X" "Now add a page for this" "Change this field to be that" "Integrate it with service X, Y, Z"
"don't forget to make it secure"
It looks like OP will be learning a thing or two about content moderation.
Well this is a good start, I tried searching for “Anti-inflammatory” and the app crashed. Likely that scaling is way out of proportion considering it’s here on HN , but do you have any load balancing or caching setup? Debouncing on the inputs?
Weird - sorry to hear that. It's working fine for me.
I get this search result https://www.recipeninja.ai/search?name=Anti-inflammatory
What browser are you using? Voice mode or typing?
I heard about it from this article a friend posted to a group chat: https://www.404media.co/vibe-coded-ai-app-generates-recipes-...
There are a lot of really weird recipes still on there, including cyanide ice cream.
Nice work! The voice control is actually really, really cool. I was able to say "show me a chicken tikka masala recipe" which got me 3 recipes, and then I said "show me the third one" and it opened nearly instantly. Pretty awesome. I want this for my entire computer!
Nice work! I also made a recipe app, but after seeing the recipes people generated on yours, I'm choosing not to share it at this time.
Seriously though - vibecoding is great. Even better (or only feasible) as engineers who can dive in when we need to.
My app is iOS and I had never done any Swift. I do have AI generation but that was more of a fun afterthought. The main utility is extracting recipes from the web and having a synced shopping list that I can share with my wife.
I understand that this is just kind of a proof of concept exposition of AI capabilities, but I fucking hate it and it is emblematic of the kind of low effort drivel that is infecting the internet. This is totally unusable in every way possible. While technically impressive (that AI can make a pretty-ish and functional website) it is practically useless.
Also, the photos are some of the most un-appetizing, uncanny valley, shit I've ever seen.
Yeah, this is amusing but I would like to see the source that was “vibes” 34k LOC is an outrageous number for something so basic..
What is basic about it? It's packed with features (that you might not care about, but that seems entirely beside the point).
It has some basic features that are 'in-house' like a login, everything else seems to be using third party API's. As per the longer article :
> Eleven Labs API for text-to-speech conversion
> streamio-ffmpeg gem for audio file analysis
> Active Storage integration for audio file management
> OpenAI integration for recipe generation
etc. I'm not saying it's not a fun excersize in 'vibe coding', I'm just very curios the quality of the code that was actually produced. Its my feeling project maintainability is something that will be a big pain point in the future is people are relying on these tools.. 35k LOC for this recipe app smells like that to me at least.
> The app’s security is not perfect, but I’m pretty happy with it for the scale I’m at. If I continue to grow and get more abuse, I’ll implement more robust measures.
The app literally exposes his OpenAI key.
Not in position to judge the product, but had me thinking...
Perhaps this is 1998 again, when you could earn big money on creating a visitor-counter service, or a guest book service.
Perhaps, now is the time for a lot of smaller projects with AI, that will, in a few years, all be blasted off the market by big corporations and change in trends.
Some characters break links. Try searching "Луковый угар". None of the previews could generate any working links
The recipe still exists though: https://www.recipeninja.ai/recipe/r_UptD1AgJYvvXWm/%D0%9B%D1...
I assume the insane recipes are generated from users’ prompts, so you might want some content warning or moderation. Kinda cool though that it generates recipes for anything, regardless of if it involves food.
https://www.recipeninja.ai/recipe/r_WHeXRD7qXHV0Vr/cyanide-i...
Ah yes, they just don't make cyanide ice cream like grandma used to
This timeline is so cursed I can't tell if this is real or just an April Fools joke.
With Gluten-Free Plywood Waffles it must be real.
Because the app shows entirely unfiltered "crowdsourced" latest recipes which of course are ridiculous given this is public, I first thought this was some high effort april fools joke on vibe coding, AI slop and the insecurity / thoughtlessness of the results.
But there is some quality in it, I can't argue against that
I always wanted to know how to do an Uranium Bomb:
https://www.recipeninja.ai/recipe/r_SOv9sTmzAz3cg4/uranium-b...
I'm joking NSA.
Would love to see a git repo
I have a few qualms with this app:
It does not seem very "viral" or income-generating. I know this is premature at this point, but without charging users for the service, is it reasonable to expect to make money off of this?
I'm scaling the heroku database right now, which caused a little bit of downtime :-/
Among Heroku and OpenAI bills, the project is apparently not optimized for low-cost operation just yet. Should it become popular, this may become an issue.
Ok it's mostly back online!
Ah yes, exactly as Ottolenghi would've presented it! https://www.recipeninja.ai/recipe/r_h7nU7K03fLU1W7/ottolengh...
So much gristle
Interesting... Thanks for putting in the effort. I'm curious, did your previous knowledge of programming help you accept code changes, or do you just keep everything written by agent?
Quick note: In my opinion, the ai-generated food pictures are deep in the uncanny valley. Unappetizingly so. The hollandaise sauce image is disturbing in a way that's hard to describe, which somehow makes it worse.
I don't think I would want to cook and eat AI generated recipes. I don't like to read AI generated texts and I don't like to listen to AI generated music.
You can call me a snob, but I appreciate some things only if they are the result of work and creativity of humans.
> It now includes a fully interactive voice assistant so you don't need to get your dirty hands over your new iPad when you're cooking.
Always floored about the problems people think need fixing. The problem is not that you get your dirty hands on the iPad. The problem is that you want real recipes. You know, things people have actually cooked and found to be good. With real photos of how the result actually looks (instead of what an AI thinks it might look based on the description).
You might be lucky and find these for free someplace on the web. However, those LLMs that "vibecoded" this Rails app for you are now also used to flood the web with garbage recipes, so finding good recipes on the web will become much, much harder than it already is. I browsed through the recipes and could not find a single one that actually looks real, so at the moment, you are just adding to this problem. This is why people still buy physical cooking books. The good ones are made with sturdy, thick paper so that you can get your dirty hands on them. This is what cooking is all about. Only unused cooking books stay clean.
Vibe coders will be so busy solving the problem of making something work that they wont invest time into understanding the business problem at hand
My argument is: the problem is solved. The solution is called: books containing real recipes.
The "problem" people have is: yeah, I want those, but for free. But creating a cooking book is a ton of work and very expensive. You need to first find good, original recipes. Then you need to actually cook each of them to see if they are any good, at least once, sometimes several times to perfect the recipe. And unless you are a photographer as a side-hustle, you need to do this again and hire a professional to make good-looking photos of the result.
Printing the books is also expensive. Good cooking books need to have very good binding so that they can be flattened to show one page for a long time without disintegrating. The paper must be thick and the printing must be able to withstand stains and moisture without becoming unreadable.
I use a delivery service that packages the food and the recipe in a box. The recipe is 1 page long, on thick paper, and I have used it as temporary food storage when the cutting board gets too full.
Works like a champ. Pictures and everything.
Who cares if it solves a problem? Look how fast they made it! Bro it’s so easy! 20 minutes! Who needs to know how to code! Who needs skills? We just vibed it!!
Imo, this is not just a vibe coders problem. It's been a classic builders mindset issue, where many start with building and look to find problem to the solution.
To be fair the author of this did found two massively successful businesses so I imagine he does understand tackling a business problem.
Absolutely agree. Real recipes, tested by actual cooks (home or professional) with genuine photos, matter far more to me than avoiding messy screens. Does anyone know trustworthy sites that consistently offer recipes meeting these standards?
Seriouseats, Epicurious, Bonappetit, or anything with Kenji Lopez-Alt (for always going the distance in methodically and empirically testing all variations of whatever he is evaluating/developing).
If you want something in the short form video era, I do appreciate Andy Hearnden (andycooks) as he is both concise, consistent and always posts the full recipe in the video descriptions (all too rare).
+1 for Serious Eats. I also like NYT cooking. I do not like Bon Appetit, I find most of their recipes to not turn out well.
BA might have gone downhill, it's been a few years since the whole unfair pay debacle, and I know a lot of good people left, and I haven't checked in as much lately.
https://www.bbcgoodfood.com/ -> Search recipes...
If you want just recipes that were published in the magazine (that this website is the companion to), you can also filter results to author = Good Food team
If you like Indian food I’d say Swasthi’s Recipes is decent. A lot of what I have seen on there is similar to how my aunties in Punjab n aunties of my friends in UP make food/have taught me to make food.
https://www.indianhealthyrecipes.com/
Edit, if you’re veg also this site is decent:
https://www.vegrecipesofindia.com/
Not books, but youtube / blog:
* Chef John (foodwishes)
* Brian Lagerstrom
* Adam Ragusea
* Sip N Feast
* Kenji Lopez-Alt
* Spain On A Fork
* Alton Brown (of course!)
I cook every day and love trying new things. There's no reason to pick just one, but if I had to, Chef John is my go-to. I stumbled across him when I was trying out pretty much every birria recipe on the internet, and his (yes, a white dude) is by far my favorite. That pattern seems to repeat with his recipes.
It's not free but Americas Test Kitchen is very, very good
For Italian cooking, il cucchiaio d'argento is like a reference website, unfortunately it's in italian only
Cucchiaio.it
https://www.pickuplimes.com/ is pretty good.
Does anyone know trustworthy sites that consistently offer recipes meeting these standards?
I believe you can subscribe to the New York Times Cooking section/app without subscribing to the rest of the New York Times.
I used to know people who hated the Times, but still paid for Cooking.
nytcooking is kinda the gold standard.
As somebody who cooks quite a lot (for a diverse set of people), I have the opposite experience of what you describe (also I hate having to touch my screen while cooking).
The cooking book scene has been openly criticized for not actually trying the recipes, even before LLMs were a thing[0]. Regular cooking websites have always been somewhat unusable due to massive ads and fluff text because 1) SEO and 2) recipes are not copyrightable, but the fluff text is.
For quite some time I get my recipes directly from chatgpt, the instructions are very condense, they work quite well, and most importantly: It knows how to substitute ingredients. "My friend is vegan and allergic to heat-resistant soy protein" and it's going to adjust accordingly.
[0]: https://www.matchingfoodandwine.com/news/blog/recipes-that-d...
What the hell is the "cooking book scene"? You make it sound like some cabal of cooking book writers sitting in some basement, conspiring to lead people astray. Nobody is saying that all cooking books are great, of course there are loads of shitty cooking books, and of course you are free to criticize that, even openly if you dare. I somehow cope by buying the good ones.
My experience with LLM recipes is like with pretty much everything else these things generate: usually very mediocre, mixed with glaring errors in between. If you are an experienced cook, you'll be able to manage since you'll recognize the errors.
In another comment, you say photos are required. I would claim that the cooking book scene is the publisher interest in the limited shelf space for photo-heavy cookbooks. Celebrity and TV chefs thoroughly represented.
Books without photos (McGee, Hewitt, Potter, Julia Child) are not interesting to publishers.
So you're allegedly annoyed with untested recipes and decided to solve this problem with a tool that only generates untested recipes? You're happier to cook or bake something from a wall of text that wasn't authored by anyone with any amount of culinary experience, that nobody has ever read before, attempted to replicate, or publicly critiqued, why?
I don't buy it, that's going to be a strictly worse experience than trying out the first search result that has photos, is rated 4/5 stars or better and that has a few positive comments. You can always ask an LLM for substitution recommendations separately.
Yup. SEO slop is generally worse than AI generated, unfortunately
What do you think the AI was trained on?
Are you trying to tell me this Rubbery Lasagne isn't a real recipe?
https://www.recipeninja.ai/recipe/r_LZarKW1PMNlSlx/rubbery-l...
All Ingredients
- rubber cement, 2 cups
- water, 1 cup
- lasagne noodles, 1 box
- shredded mozzarella, 2 cups
...
Step 2
Prepare the rubber sauce by mixing 2 cups of rubber cement with 1 cup of water in a saucepan over low heat until thickened.
This is the funniest thing I've read in a while. Thanks.
Indeed. See Ann Reardon's recent video on the topic: https://www.youtube.com/watch?v=SVaYJgmPvCg
Agreed! That was a problem even before LLMs showed up. I found my subscription to Cooks Illustrated/America's Test Kitchen to be incredibly valuable. Bonus: You don't have to read through 15 paragraphs of SEO fluff to get to the recipe.
And without 14 paragraphs of how your great grandma first discovered this recipe on a trip to italy from a soldier she romanced
I'm convinced that people who complain about this don't use recipe sites very often. There are loads without any preamble, and many of those with narrative provide a 'jump to recipe' button. Even if it is a long article with no jump (rare), it's still not a big deal
1. Some people do want the narrative 2. It's not that much of a hardship to scroll
You're entirely right. I don't use recipe sites very often, and yet it still happens often enough that I have to keep scrolling and reading on different sites to get past the backstory
There's also sometimes a button for the "Printer Version" of the recipe which is very much no fluff.
Ex. Above all the fluff is the word "Print" in a yellow box [1] which takes you to a no fluff page [2].
[1]: https://www.seriouseats.com/buttermilk-vanilla-waffles-recip...
[2]: https://www.seriouseats.com/buttermilk-vanilla-waffles-recip...
I've got a shelf full of physical cookbooks -- The New Best Recipe, The Food Lab, an Ottolenghi book, On Food and Cooking, etc. I like them, I've read through all of them, I sometimes use them for inspiration.
It's rare that I actually cook directly from them -- usually, that'd be big and fancy stuff or stuff I'm very unfamiliar with; in both cases I usually take the time to cross reference whatever the cookbook says with additional resources from the internet.
ChatGPT, on the other hand, I frequently use when or before cooking (and I cook virtually every day).
It's great when I only have a vague idea based on stuff in the fridge; five minutes later I've got a checklist I can reference. If it hallucinates something that I flat out don't think will work or, much more likely, comes up with something that I don't want or cannot do for lack of ingredients or time or whatever, I'll tell it to adjust the recipe and it does.
It's also great when I feed it a couple of existing recipes (from real people) to compare and contrast and integrate and reformat in a way that's most useful to me, e.g. a tabular format, or scaled to a different serving size.
With all that said, the AI based recipe sites don't really do it for me, either. If I want to cook purely AI generated recipes, a chat interface works fine -- and probably better. What I really want is an AI tool that helps me curate my own recipe collection. E.g. I want to ask it "I'd like to make Ramen, how did I do it the last time, what were my notes" and when it's done I want to tell it "ok, this was fine, I decided to double the mirin and next time I'd marinate the eggs longer" and have it update the recipe.
Is the project anything more than a (fairly elaborate I must recognize) 1 April joke? I agree with your point even on a more general level, amateur techies thinking to fix stuff which never needed fixing are nothing unusual. We "disrupt" the economy, the finances, housing and more, every day, while being barely legal and sometimes outright abusive to some part of the society, all in the name of the Saviour Technology. Is this just another example of "solutionism"? Or it's just me waking up grumpy and without coffee...
Edit: and right after this, I run into another AI-related gem: https://artificialintelligencemadesimple.substack.com/p/ai-t...
You don't fancy Cyanide Custard? Radioactive Slime Jello made with actual radioactive waste? Sweet Tooth Delight made with real human teeth? Or a few other desserts with NSFW names that would go down a treat for a family dinner. It is hard to please some people /s.
But yes, websites will now be filled with these low-quality recipes, and some might be outright dangerous. Cyanide custard should ring alarm bells, but using the wrong type of mushroom is equally dangerous and much more challenging to spot.
I don't know why everyone is so salty. It's like putting "vibecode" in the headline attracted the worst in people.
For context, I'm not a cooking geek or virtuoso. I enjoy it to some degree, but mostly it's just about having a nice, nutritious experience, in line with whatever my mood might be. I only ever measure things super accurately when I'm baking things in the bread maker (because it doesn't let you make corrections). For most meals, I wing half the measurements and time estimates.
In my experience, most "human recipes" are just random variations on some baseline. I hate looking for recipes on recipe sites, youtube, etc. There are food bloggers that are exceptions, but usually I'll just end up scrolling for a long time with just frustration to show for it. If I sort of know some of the ingredients I want to use, have some sense of the type of eating experience I'm going for, and I want a bunch of recommendations based on that, regular "human recipe" sites are not the answer.
90% of my new recipes come from ChatGPT, and that ratio is increasing. I just marinated some chicken based on a recipe it magicked for me. I asked for insights on mixing mayo and yoghurt in the same marinade, because I had leftovers of both. It gave me 5 or so diverse recipes, and I just picked the one that best fit my pantry and mood. I also asked it to convert the recipe from volume to weight, not to mention scale it for my specific quantity of chicken, which was super handy.
I find that ChatGPT is great at providing common sense instructions and approximations. It's absolutely awesome at clobbering together a meal from ingredients I tell it I have. I can have an actual dialogue about any of it, get all kinds of recommendations and insights. That's been very useful to me. I'd go as far as to say that recipe generation is one of the easiest real problems for an LLM to solve. Or, at least for the kinds of recipes I use.
I've done my share of recipes from Serious Eats, but they weren't particularly good. I was doing Breton galettes the other week, which are notoriously fiddly to get right. Serious Eats had a huge article about it, interesting insights, but their final recipe sucked, and I was trying to be accurate. Not only I failed to get the consistency right, the wheat-buckwheat ratio was nowhere near what you'd get in France. I say, write researched articles about what makes recipes work. I can read it, I can bounce my LLM off it. If it's a fiddly recipe, I'll have to fiddle with it no matter what. If I can have a conversation with an LLM about principles at work, that's much better to me than a bunch of "human recipes".
Also, I often have questions about alternatives or things I need advice on as I'm preparing food. I'll also look at the recipe a gazillion times to check the instructions, quantities, etc. I'll set and check a timer often too. A voice-assistant is the obvious answer to this, which I'll try at my earliest convenience.
Kudos to the author!
> In my experience, most "human recipes" are just random variations on some baseline.
This just shows your lack of experience and curiosity. The art of cooking is a cultural achievement with thousands of years of history, with vast differences between different countries and cultures. Just visit a store (yes, a real one) with a good cooking book section. You might be surprised.
Seriously, what's with this thread? I'd rather not have a discussion in this base tone.
You're on a discussion forum. Where you express your opinion. Then the other party expresses their opinion. That's what makes this discussion forum great. Sometimes the opinions of the other party aren't going to agree with yours. It's a normal part of discourse on online forums.
You're implying that the only quality a conversation has is (dis)agreement. I have absolutely nothing against disagreement. Why are you saying this via a throwaway?
> It's absolutely awesome at clobbering together a meal from ingredients I tell it I have.
You should be able to do this yourself easily, though.
Feedback:
I would tell the AI to avoid recipes that contain bodily fluids.
> https://www.recipeninja.ai/recipe/r_vBiLoIJK7qsUc7/cum-panca...
Love it! I went to fix an easy sandwich and ended up building my own recipe app:
https://www.recipeninja.ai/recipe/r_iEyaSAKCQlzl4Q/vibes-and...
But then, as with most llm tools, the fun wore off after a few minutes of playing with it.
Is this even legal? It was on the main page.
https://www.recipeninja.ai/recipe/r_dxF7OQ0O3IGXOw/actual-co...
I bet there might be a recipe of a bomb somewhere, too.[1]
April's Fools or not, I think you could get in legal trouble, but IANAL.
[1] Apparently there is: https://www.recipeninja.ai/recipe/r_SOv9sTmzAz3cg4/uranium-b...
Please inform your orange overlord to ban this gov website asap!
https://www.ojp.gov/pdffiles1/Digitization/132907NCJRS.pdf
It was just a note to the developer because he may get in legal trouble for it (it happened before), the Government obviously will not.
Edit: Both "recipes" have been deleted (the URL should tell you what it was about).
I read your comment before the edit so please accept my apologies
Saying that posting a simple recipe for cocaine was probably illegal sounded very "puritan american"
As a European (French) myself, I'm surprised that you feel that way because I remember reading in a chemistry book in high-school how to make explosives and our chemistry teacher took the complete timeslot to respond in details to someone who asked if, like Walter White, they could make meth.
I do not think it should be illegal though. It might be (hence the note), but I do not agree with it.
No hard feelings! :)
Opened and the first recipe was “Thick white cum soup”, glad I’m my mobile phone, not at work
https://www.recipeninja.ai/recipe/r_6e0Low2SZhbkoD/thick-whi...
Now updated to include real cum!
Sounds spicy...
Dish: Fek Yerr AI Slop Garbage Plate All Ingredients
Thank you for that ice water recipe, really wouldn't have known how to do that: https://www.recipeninja.ai/recipe/r_8Pq0RPfvVGf2Xw/ice-water...
Couldn't you just have installed wordpress and a theme?
Ai slop for generating more ai slop!
Missed an opportunity to call it "Vibe Cooking"
This is so crazy. How did you gather recepies
It's AI generated.
Wow, I can't believe how many people seem to be missing the point here and being overly negative. The fact that you can have an ai assistant code something like this by giving it instructions would be utterly miraculous 10 years ago, even less.
Whatever the ultimate usefulness of the website is, the point is using it is slick. It works and it works well.
Very nice demo of vibe coding Tom. I appreciate it.
I am horrified that you've produced that much code in such a short amount of time.
Thanks for sharaing your journey, I have few observations
1. When I click on a recipe from home page it is maintaining the scroll position, so I am not seeing the top of the screen. Is this deliberate?
2. "Recipe Ninja was vibecoded by Tom in San Francisco." Will it increase of decrease trust in your system for users?
3. To remove AI changing random files, I use "Copy relative path" to tell AI which file to change (there is a keyboard shortcut too). Not fully vibe coding but can be useful for precision bug fixing.
Good luck with the project.
Obviously your fish recipes are the best, well done sir :) Apparently most people aren't aware of the date...
Every recipe is cum bro
"Still, I was determined not to write any code ... I just sat laughing as the computer wrote code."
I can pop over to Midjourney and be determined not to draw a single line and "sit there laughing" as it draws the Mona Lisa in the style of Salvador Dali but with a turnip instead of a person.
How is this any different? What is ultimately notable about it? Did any of it make you a better programmer?
I'm always deeply impressed when people devote significant chunks of their time to achieving extraordinary results. I'm entirely baffled, however, that there's anything at all interesting about using an AI interface to build an AI interface to connect you to AI slop.
You could have spent 20 hours planting trees or doing some kind of community serivce, and the world would have been a far better place.
Programming is not art. In the end, nobody will care whether software was painstakingly hand-chiseled by humans in a dungeon or "vibe coded", as long as the result is good enough. (I'd argue that art will "suffer" from the same "problem".)
What is notable here is that someone is demonstrating that the systems are reaching a quality where this is possible.
> Did any of it make you a better programmer?
By conventional metrics, if the job got done well enough in less time, yes, even if less skill is involved.
I don't know about that, Roller Coaster Tycoon is fondly remembered and sits on a special place culturally because it was written in assembly by a single person.
Programming is more art then anything since it's definitely not a science, nor is it mathematics or engineering.
It's similar in principle to architecture or interior design.
April fools day has messed with my ability to identify satire at this time of year. Let's look at the evidence:
Posting date: 2025-04-02T01:57:13 1743559033 <-- too late
LoC: 35,000 <-- That's a _lot_
Front page: "Elon Musk Dirty Pants", "Heroin Hashbrowns", "AI Slop Stew", "Sweet Tooth Delight Made with Human Teeth" <-- WTF?
This is a joke, right?
took forever to open.
I mean I guess kudos to you for being creative and building something, even if its with the so-called vibe-coding. But LoC is not the metric I would use to describe any app I've built, or that I am building.
People are putting disgusting nonsense in there. Maybe add an AI review step to block that stuff. Let them do the whole recipe, then just block it.
The only obvious problem here is showing all of them publicly. Let people do silly things but only show the sensible ones instead of being user hostile by default with false positives
Leaving them public has the advantage that no one will think the site is serious, which will help them avoid the recipes that look okay but are poisonous.
A wrong fixing another wrong lol
How does one learn to vibe code
By practice. It requires some skill too, to constantly remove slop and keep the project from deteriorating. Usually by suggesting good architecture, asking for tests, and rewriting / cleaning up some bad code endlessly. Still much faster than manual development.
But how exactly !
1. Download Cursor
2. Tell it to make you something
3. Get frustrated when it doesn't work
4. Think about how to revise your prompt
5. Repeat from step #2
I prefer Aider, but that's a matter of taste.
Any free versions of Cursor or Aider?
Is this satire making fun of 'AI'? I cannot tell. But it looks and feels like satire.
the microphone symbol should be animated while listening.
This is great. Want to come demo it in the weekly EarthPilot.ai AI Playground tomorrow at 11est?
Http://earthpilot.com/play and then join at AnthonyDavidAdams.com/zoom at 11 for show and tell.
I’m making a non-fiction book writing agent and I’d love to better understand how you used function calling to navigate the website!
Basically you declare to the AI which functions (tools) are available for it to call: https://platform.openai.com/docs/guides/function-calling?api...
Then you handle those function calls in your javascript.
``` if (function_name === 'search_recipes') { const searchParams = new URLSearchParams();
```Afraid I have a day job, so I will politely decline your kind invitation
Thanks! I’m going to show the folks what you built. We have VCs, founders, and all Sorts of folks who join us every week.
Amazing! Truly the frontier of technology. Now imagine if every recipe was an NFT, stored in the blockchain!!
One of the websites of our lifetimes for sure!
Make it a Spacial NFT.
[flagged]
This is against both the HN guidelines and show Show HN guidelines.
https://news.ycombinator.com/newsguidelines.html
https://news.ycombinator.com/showhn.html
Please don't attack others or their work like this on this site, regardless of who or what you have a problem with. It's the opposite of the curious, respectful conversation we're looking for, and always has been.
A cooking web app 35k LoC? Wow, the author might be surprised to know that an entire OS kernel can be programmed at a fraction of that count. I can safely say that the days where we compained about JS bloat can be put aside, this is a whole new league by itself.
Sane advice: learn to program, put the AI hype/drug aside and do yourself a favor. It's an invaluable lifetime skill knowing to program from scratch and perhaps unassisted coding will be a looked-after skill in the years to come.
I kinda wish I could look at the code, but it's not open source and the frontend sources are minified/obfuscated, but it's long and looks like it's using React and a css-in-react approach, which also generates a ton of code.
The term "vibe coding" gives me bad vibes, I feel that it is in some way an insult to my craft, the way a carpenter might feel about Ikea. But if only LLM's are going to read the code, the LoC aren't a big problem, increasing cost significantly less than if a real person would have to read it.
The two biggest issues I can think of with this approach is performance and security, with the first one only being a problem if you "make it", and the second one being too often ignored with or without vibe coding.
I keep thinking of Visual Basic... It was scoffed at as not real programming... It was the drag and drop thing in the Visual Basic IDE that made it worse. Suddenly "anyone" could do it without really knowing programming. This new trend (if it is a trend) is sort of like drag and drop.
That IKEA comparison is an awesome, perfect analogy of vibe coding as cheap shitty furniture.
- can't be maintained over time, gonna need to replace it
- mostly simple/basic furniture
- out of place in any existing environment
- easy to break (cardboard tables!)
Sometimes that's all you need though.
Vibe convert it to rust and it's even more bloat.
please stop saying vibecoding it makes me wanna kms
It makes you want to... kilometers? I've heard of people "going postal" when they're angry, but I never heard of someone "going metric".
Less tongue-in-cheek: there's no word censorship in HN. You can say "kill myself" here.
That is nice actually. Recipe sites are so annoying where I have to scroll for an hour to see the ingredients. Thanks!
I never wonder about downvotes (and we are not supposed to ask) but why? Downvoted for a compliment.
You probably got downvoted because you admitted your failure of using the WWW. The problem you describe only happens on "recipe" websites that are 99% advertising and written like a blog, with every other word set in bold to maximize the ad revenue. These websites should be avoided entirely. Proper recipe sites include en.wikibooks.org/wiki/Cookbook:Table_of_Contents, based.cooking and of course https://web.archive.org/web/20110323153937/http://www.ottavi...
Ok, I think this is actually the best recipe app I've used. Well done!
Haha, someone downvoted me for saying this!
Stop ruining the commons please, it is a horrible ethical crime to flood the internet with AI-generated garbage.