Hey HN, I'm Daniel, founder of Onlook (https://github.com/onlook-dev/onlook and onlook.com).
Onlook is a desktop app that puts a canvas on top of your real codebase, letting you design directly in code with the agent you already use. Over 25,000 designers and developers have supported the project, 100+ have contributed, and it's been the #1 trending repo in the world twice. We just rebuilt the entire product as an agentic design tool, and open-sourced it to the community.
Here’s a demo video: (https://youtu.be/BogcYWf4siM)
If you're an engineer, traditional engineering tools are largely built for convergence, whereas designers tend to explore and diverge on ideas. Onlook gives designers a familiar canvas while managing all of the dev processes that get between them and making mind-blowing designs in code. Designers can work on a multiple simultaneous worktrees that run on the real app locally, and every change lands as a PR you can review or throw away, so there aren’t mockups to rebuild.
We built Onlook because we believe there needs to be a developer tool that works the way designers do.
More and more, designers are expected to work in code. Designers think by seeing, drafting, arranging variations, using references, and remixing ideas. That’s why the canvas has been the signature feature of design tools because it genuinely is useful for how messy the creative process is. Most of the challenge of using engineering tools for design work is that the first half of starting a project often takes mental bandwidth in of itself (What branch am I on? What page was that last exploration on?) and the second half is constrained by configuring your coded design to present itself (prompting design tools on top of what you’ve built or making temporary navigation for that specific state).
Some of you may remember our earlier Show HNs (July 2024: https://news.ycombinator.com/item?id=40904862, August 2024: https://news.ycombinator.com/item?id=41390449, May 2025: https://news.ycombinator.com/item?id=44127653). Since then, we built a web version (too early), then a Storybook-native build (a mistake). We noticed that our desktop product from July 2024 was still getting hundreds of downloads a month despite zero promotion, so we decided to rebuild it for the new era of designers working in code.
How it works…
Onlook shines at exploring tons of ideas at once. Ask it to lay out every flow in your product, show every state of a component, or riff on ten versions of one screen. Your designs all show up side by side on an infinite canvas, with each frame as a real webview running your actual app with real data. Not as a mockup, but the real deal.
When we built this editor, we had an epiphany that it could be a new kind of IDE for design. Design tools used to be limited to a proprietary setup or closed schema, but now you can push and pull the medium of code to get whatever result you want, so why not give agents an infinite canvas of browsers and see what they do? It turns out, the agents can figure out how to better present coded work by adjusting the urls of browser frames and arranging and resizing them on the canvas.
The concept is simple, but it requires a lot of product thinking. For example, because all of the frames on the canvas share a single browser session, and Chromium only allows six connections per host per session, ten frames of the same app end up fighting over limited connections. So, we had to plan around that, and there are still tons of things we need to do to make the canvas ultra performant.
Onlook is an Electron app where you can import a repo from GitHub or upload a local folder. Then, Onlook creates a git worktree, starts your dev server, and has the agent sort out missing dependencies or other requirements. You never open a terminal and everything the agent does is a normal git diff that can go to GitHub as a PR. We inject a small script into each frame that draws selection outlines and reads the page, so that way you can click an element, say "make this a card layout," and the agent knows exactly what "this" is.
This time we didn't build our own agent. Onlook runs Claude Code or Codex on your own subscription, and we turn whatever the agent emits into one typed event log and draw the UI from that in our own chat interface. On top of that we give the agent a bespoke set of design-specific tools, from arranging frames on the canvas, to making suggestions, to generating multiple iterations of a selection – which is especially useful because the agent decides how to get the app into a state that you need based on your prompt.
We also have our own Artifacts which are quick HTML sketches that live in the editor, but outside your repo. They’re almost like components in other design tools in that they can be carried and edited across every project but are detached from the project with no network access. As an example, you can ask the agent to turn your interface into an Artifact, and use it as a reference for another exploration, or download it as HTML and send it to a coworker as an interactive export of something you have in progress. I frequently go back to the artifacts I’ve made as visual snippets when I need new ideas or need to borrow something that’s already been built.
The desktop editor is free (you can download it from our website at onlook.com) and open source under Apache-2.0, and we'll eventually offer a paid cloud product for teams.
We'd love to hear which repos work and which don't, what you try to do on the canvas, and what's missing. We’re so excited for this to be a new way for designers to work in code!
The Onlook team (Joel - thebrubaker and Kris - krisodd), and I will be here all day!
this is very cool. how can I make use of it as someone who doesn't code but uses claude/ codex for everyday things? I do a lot of design work in claude or grok.
i use claude design for a lot of design work but it always struggles with keeping new components and designs in style with existing components and designs so i have to spend a lot of time telling it to constantly "use the same design system". how do you guys solve this?
That's great tool but the issue the design of the AI even if it like that easy to change it fix it what ever they still creating the same template of designs i think they needs a lot of samples that are real from real designers to use them as a source. For example when i was creating my platform about 2 years ago designing it was the most complex part because the AI was making a lot of mistakes in layers and z-index and content appearance and much more. But now a days those issues becomes much more rarely to appears and the big issue in the creativity of each design because each design now a days it looking same. I wonder if in the future you will use real designers designs as a source?
Big, big fan of open design tools that run real code! The "iterating on Onlook in Onlook" demo got me :)
I'm finding myself spending most of my time inside frontier labs' desktop apps lately and I'm always looking for ways to pull more of my workflow into them... any chance you guys are working on a Codex/Claude Code plugin so I can run Onlook there?
We'll add a MCP for using Onlook from any agent pretty soon, Codex and Claude plugins would be great additions when we have this in place!
Looks cool and I'd love to check it out, but there are a few hiccups even getting into the app.
1. It's not entirely clear from the on-boarding as to why I need an account 2. When trying to setup an account via email I keep getting hung screens or the error: Can‘t verify the user is human. Please try again.
This is on macOS 27.2.1
hmm that's concerning sorry to hear that's happening.
Did any of the other logins work for you? Appreciate you reporting this!
Wow this is fire. How do you handle the step after exploring ten versions, when the designer wants “the navigation from version 3, the layout from version 7, and the interactions from version 9”? In a canvas tool that’s mostly copying pieces around, but across worktrees it becomes a code integration problem. Can Onlook help combine those ideas?
Appreciate the kind words! The answer is going to sound absurdly simple – hold down SHIFT, select the frames you want, and say exactly what you just said! It's pretty wild what you can do when the AI is looking at the same canvas you are + has the context of everything that was created.
We do need to do a bit better of a job of persisting project iterations after the handoff is made though. A lot of design work relies on being able to reference past explorations, and the agents tend to lean on cleaning EVERYTHING up that isn't going to be made in the PR, so a lot of times project explorations can be swept up in preparation for an idea going live. But it's a challenge we're definitely able to solve with some tuning and UX.
I’ve been doing pretty much all of my design work in Codex lately, from trying out rough ideas to refining the UI in the actual codebase. What does your tool do better than Codex for that workflow? I’m especially curious about what happens after the first draft, when I’m dialing in the layout or trying to make a new screen feel like it belongs in the rest of the app. Is there a concrete example where you’d reach for your tool over Codex?
it depends on what you might be driving in codex, but for me it's just a great well designed canvas experience for the same kind of workflow, and I often jump between being in cc or codex, and being in onlook driving things. I think if you're working on a problem that really benefits from being in a canvas space (seeing many variations side by side, jumping between problems quickly) then you might like it more.
also: onlook connects to the same codex you drive, you'll see the same session when you jump from onlook back into codex
Awesome! Definitely a much needed tool. How does it compare to paper/pen? I've been preferring the former, since pen has its own format and I prefer not to use it. nit: it was pretty hard to find pricing info. I had to scroll down to the footer to see it. Would recommend adding it to the navbar
The key difference for Onlook is that we work on the real running application. Paper and Pen are excellent tools for working on abstractions of what you'd like to ship, but still are constrained in a way that favors the editor experience, not the building experience.
This is totally ok in our opinion – you want to choose the right tool for the job – we just think the job for designers is trending towards being right up against the real medium of code and how it gets represented when it ships.
Your point on the proprietary format is a good one. There are so many ways to build a design tool, and it felt like agents give us an opportunity to make an entirely new way to craft designs that doesn't require a strict format. Why can't we just select the thing we want to change and do it, regardless of what it is?
I think this makes sense, I have customers that still use figma, how does this differ from some of Figma's new capabilities?
Figma is of course the UI/UX design industry leader that we're all familiar with, but the flaw in their primary vector editing tool has been its disconnect from code. They are adding-on code-related features, but those features will always be secondary to their ultra-powerful editor as the foundation of their business.
Onlook will always be code-native and caters to the ways designers will need to work with code specifically, not vectors. That means being local-first, but also flexible for the challenges and diversity of code that will have to continue to be maintained in the future.
Onlook will feel a bit light to start when compared to Figma 1:1 on every feature, but the foundation enables us to focus on the future of designers coding and generating assets with AI, not drawing vectors.
also your cc / codex is the driver, so if you have Figma's MCP connected you'll see that in Onlook, and you can treat them as complimentary
Is this cc api only? This is exactly the workflow I've been imagining, so eager to try it out.
We detect your cc subscription automatically, so you get your subsidized intelligence right out of the box :)
The onboarding is super fast in Onlook you may be surprised how quickly you can go from starting to designing (which you'll see as a claim on our install page: https://www.onlook.com/install)
yup! we work with your existing cc setup / subscription, so if you're already logged in you're good to go. same claude you use every day, and onlook is just the canvas you build on top of
"World's best, top-most agent recommended #1 Developer tool for Designers and Visual Engineers to design with Real Code."
Did you agent also recommend this copy??? Fire it! :)
hahahaha we borrowed it from TLDraw! https://github.com/tldraw/tldraw
We also think it's a pretty hilarious idea... maximum agent-discovery.
Very very cool. Seems like there's a huge opportunity here. Cursor got stuck after the acquisition. And especially if paired with open models, I'd definitely use this to avoid using my limits on cc/codex
I do think that there is a great opportunity to bring in open models into this interface. We've also heard from a lot of designers how Cursor has fallen short when it comes to helping them get their design work done.
There are a lot of features that Cursor offers that we still need to meet parity on, but We do see an opportunity to build an excellent tool for creatives to work in code on an infinite canvas that aligns better to the divergent process of design vs. the convergent process of engineering.
we have MCP prepped as another connection, that should let you drive onlook from your open model / harness too. What open models are you using (and what client are you driving them from?)
Congrats on the launch. How does it differ from tools like pen.dev?
Pen.dev chose to go the route of a specific file format for design in favor of building an editor around that, which is a more constrained approach in our opinion. Meaning that the work you do in the canvas is always a "copy" or "mock" of what is actually being implemented.
For Onlook our goal is to make everything you do truly 1:1 with no translation layer in between what you see and what will ship. It's a small nuance but makes a big difference in the design process.
They certainly have an impressive product that is quite performant from what I remember.
Do you mean you render the code for perfect fidelity instead of having the LLM translate it to JSON?
What's the business model?
We're free for now but have collaboration + cloud features planned, and there's a lot of additional paid services that we'll bundle like image generation, vector generation, even font generation (which I'm super stoked for) and that will require some spend on our side that we'll charge for in a general subscription.
Also, as is popular with working on major codebases these days, we see a lot of interest in VMs and/or sandboxes, and there's an opportunity to do that in the canvas, but specialized for the ideas that people cook up quickly in Onlook.
cool, congrats on launch. the designers in my life are all loving claude code rn.
> Onlook runs Claude Code or Codex on your own subscription
this was first thing i looked for in the readme and didn't see it. recommend putting it in there :)
hah I agree, I think this should be higher up. under the hood we're built on the agent SDK and codex's app server, so the agent you drive is what you're already using every day, and onlook just brings the canvas / tools / workflows that help them be most effective in that space. the agent relationship is between you and anthropic / openai, agents and code are local, and the chat is basically a rendering layer for you to not have to jump between multiple apps.
thanks! we mentioned it on this line – but we can make it more prominent thanks for the suggestion
> Bring your own agent. We don't ship our own model. Connect the Claude Code or ChatGPT/Codex account you already have, and that agent does the work inside your project. Plug in more tools through MCP connectors to bring all of the context you need to your creative workflow. Work on multiple design tasks at once across repos and pages.
Isn't it illegal (TOS) to use subs outside of Claude/ChatGPT proper?
We work like other IDEs do by using the same subscription that is on-device
Nice demo video and love that’s open source!
Appreciate it! Took a lot of takes but Screenstudio made it much less painful to edit the final take.
what's the main benefit of using this over cc design?
CC design is a great product for sketching out ideas but works more like a canvas of artifacts that a dedicated dev environment for working on running apps and codebases.
There are also some odd UX decisions they've made with that product in favor of competing for a general design tool but constrained by how they render designs in code.
I get the sense that the mega-bundling of Design into Artifacts is an added step for creatives who want to just dive into a tool made for explicitly design work.
Do you use cc design a lot? What do you like about it?