Ask HN: Does Claude Code remove the need for so many front-end frameworks?

yes

It really depends on what you are building. I remember asking a Jr Dev to put up a simple static website of just a few pages and they were getting lost in react. Being an OG dev, I had to show them how to get it done faster and in a more lightweight manner with plain HTML, javascript and CSS. It was a simple site but the dev didn't know better cuz everything was done in react at the code camp they went to.

Conversely, I had Claude build a webapp for internal purposes to help translate some content we were creating. It was done in HTML, JS and tailwind with no framework. After a while, I could see that the jank it kept adding would be a problem so I had Claude refactor it using sveltekit.

I don't feel AI assisted coding has removed the need for framework adoption.

Instead, I feel it's enabled us to more freely choose what frontend framework(s) we want to implement. Based on the problem we're trying to solve.

There's only so much that 1 human can become an expert at. Before AI assistants, we had to make some choices: I cannot become an expert at React, Vue, Angular, Svelte, SolidJS, Quik, and Astro/Alpine/HTMX simultaneously. Too much. So I must pick one or two that I think are best-suited for me. Learn them deeply. Then apply them everywhere.

With AI assistants, there's more freedom. I'm not an expert at Svelte. But if there's a web problem I feel Svelte would solve best? Then that's what I can use. If I really need React's virtual DOM? Then go that direction. If what I'm building is so simple that a static website with basic HTML and JS and a CSS framework is sufficient? I can go that direction.

Granted, there are absolutely risks for relying on AI assistants to write code you don't comprehend or understand. There are times I'm okay with it. And times I am not.

But as for your question, am I less-likely to use frontend frameworks now? No, I'm actually using them more than I did previously. I'm just being choosier about what I truly need, versus what's overkill.

[flagged]

Same tbh, but perf and UX still bite you eventually. frameworks aren't dead just.. delayed

It's been like this all along! We can write complex applications with HTML/CSS/Vanilla JS... But then.. React happened! UI rendering JS is bad - thats just my opinion. If server can render HTML and serve, whats better than that? Its ok for apps where browser is the source of truth like Figma, Google Sheets.. But using React for everything ... that is overkill. Most apps has server as the source of truth, data lives in server, and it can generate HTML right there.. No need for JSON+Hydrate, which is unnecessary- because server can generate HTML, the browser can understand HTML... Virtual DOM makes React Fast, not our apps fast..

AI didn't eliminate the need for frameworks.. It just picks the right one for the Job.. when plain HTML is ok, it chooses plain HTML.

I started using my own frontend framework from 2022. It is a simple jsx based, directly DOM modification, in just about 3k lines of code inspired from React/Solid/Vue.

Have try various approach since Claude Code came out. It works well in any of the styles vanilla, React, Vue for simple pages. So for simple one page demo, I mostly just ask to use vanilla style. Because it can be fully self contained in one file and just be opened without any external steps.

While once it across certain complexity level and some manual change needed. It is super hard with pure vanilla style, due to so many procedure code.

So I usually just prompt to generate using my own framework with a distilled skill. I then change a bit to make it simpler and follow my own style.

Maybe. Frameworks and abstractions whether jquery or JSX are more about comfort of the person than actually solving a real technology problem. That said, when you hope for something like Claude to replace so many front end frameworks that implicitly means replacing so many people at such variant levels of confidence.

Don't think so, I've got better results when I use a framework, specially with svelte

Like Rich Harris said in a Svelte talk once: “frameworks are for organizing your mind”. As far as I can tell, the same is true for LLMs

It all depends what you want to build. If you build a simple website, sure go with plain html/css/js. However, if you build something that needs let's say performant DOM updates, React or similar will give you that for free, and otherwise you'd have to develop and test it yourself.

[flagged]

I would rather maintain a vibe coded frontend that uses well organised react components than some huge vanilla js codebase.

I think the needs are perhaps a bit different now. Syntactic sugar is perhaps less important for example. But code structure and architecture are still critical.

In my initial prompt I explicitly tell DeepSeek to use Node, Express, EJS and PostgreSQL only, plain HTML, JS and CSS. Nothing else, not a single extra package.

So far apps are the best, easy to read, easier to maintain.

[flagged]

It depends on what you want your frontend to do.

The main reason frontend frameworks like React/Vue existed was to manage complexity, like state, DOM updates, component reuse. If your target is simple interactions, Claude can absolutely remove the need for a framework by generating clean vanilla JS without you needing the abstraction layer. But for complex interactions, using a framework is still the better choice, you can ask Claude to build on top of a given framework rather than letting it generate custom implementations that may break on edge cases.

So yes, Claude Code does reduce the need for frontend frameworks, but there are still good reasons to use them.

Yes. Memorizing syntax is now obsolete. Frameworks are generally somewhat constrained, now they can be auto-generated to be exactly the UI you want. Only caveat would be saving tokens but tokens are cheap. RIP

I've been a rails centric developer for 10+ years and dabbled in React, but lately have felt "supercharged" in my domain, and doing side projects that are explicitly front-end React/Next.js, etc. For my primary job, I work in the enterprise where I'll use AI to assist but I'm still coding 50% there. I think you just need to resist the "letting it rip" and doing everything on its, where properly managed it's a great tool. Not to self-promote but the first ai-assisted next.js typescript project I did which was a revelation: https://flip.watch

Realistically, no need. Just a fact, in China right now, front end engineers or programmers can't find jobs....

For me, it’s gone both ways, leaning towards less framework use.

In some cases, I’ve chosen to use frameworks I wouldn’t have otherwise because unfamiliarity is no longer a barrier. I can lean on the AI for the rote memorisation.

In other cases, I have stripped entire frameworks out of my projects because I can now get AI to produce the 250 lines of code that are perfectly tailored to my project, instead of importing someone else’s 5000 lines of (effectively) black box code.

Just the other day I implemented an audio player in HTML5 (progress, playlist, interactions, etc) and didn’t even consider a library. I have code that does what I want and is small enough for me to read in entirety and debug by hand if needs be. (One of the things I like to do is insist that AI comments most lines of code, and then I go through and trim comments to leave what’s useful to me.)

[flagged]

[dead]

[flagged]

[flagged]