(Credit: Jeffrey Hazelwood/Ruben Circelli/PCMag; Warframe)
The practice of using AI to help you code, better known as vibe coding, is not just commonplace but also straightforward enough to try without any prior experience. But what if you want to vibe code without knowing anything about programming or ever actually seeing a single line of code? To test the limits of AI-assisted app creation, I set out to create a build planner for the Warframe video game using only natural-language prompts via Claude's standard web interface. Everything worked much smoother than I expected.
Claude Makes Vibe Coding Ridiculously Easy
I liked Claude's interface and app-creation potential when I reviewed it, so I decided to use it for this project. However, I didn’t want to use Claude Code, a coding assistant for programmers, since it requires a decent amount of setup and some technical know-how. Instead, I was more interested in the ability to create, compile, and use apps via prompting in Claude's chatbot interface.
I exclusively used Claude’s latest reasoning model, Opus 4.6, with extended thinking enabled to get the absolute most power out of the chatbot and give it as much time as possible to chew through all my requests. I used Claude’s Max plan (the 5x usage variant) to avoid running into usage caps. Of course, you don't have to spend $100 per month to build an app with Claude, especially if it's simpler.
Why I Wanted the Ultimate Warframe Tool
I love the video game Warframe, a live service, third-person shooter set in a high-concept sci-fi world that launched all the way back in 2013. But its lackluster in-game interface forces me to use third-party apps, such as Overframe, to plan out character builds. These apps have their own limitations, so I decided to use Claude to make a better version. I wanted my app to do all the core things these apps do (such as allowing me to equip any number of different items and see how they impact my Warframe’s stats), as well as some new things (such as include even more kinds of items and better support for conditional effects that influence stats).
(Credit: Digital Extremes)Such an app requires a complicated series of interdependent calculations, a large database of in-game items, and a user interface with a ton of different elements, among other things. Put simply, it’s not an easy app to make, whether you’re vibe coding or designing it from scratch.
Opus 4.6 Makes a Noticeable Difference
I expected to encounter some pain points during the app creation process, since I’ve run into issues with Claude before. When I tested this capability in mid-2025, I frequently hit the context window limit on conversations, which required me to start new chats. Oftentimes, Claude accidentally removed things from an app when I asked it to include something else, necessitating a rewrite from scratch. Even when everything seemed fine, Claude would often bug out or crash.
But all of those problems are now gone. I never hit the context window limit because Claude automatically compacts my chat so we can keep talking. It now has no trouble making changes to apps without accidentally removing other things. Furthermore, I didn’t experience any crashes or major bugs. I sometimes find it difficult to appreciate the upgrades in chatbot reasoning models, but Opus 4.6 really shines from an app-creation standpoint.
FrameBuilder: From Idea to Fully Functional App
My app-making adventure spanned hundreds of prompts explaining what I wanted, along with requests for changes, new features, and refinements. At the beginning, I focused on asking Claude to build out my app’s core features and interface, which meant a searchable library of different items (Arcanes, mods, and Archon Shards) and a way to track all of the various stats.
(Credit: Anthropic/PCMag)Then, I had Claude add the functionality to track mod costs, polarities (mod types), and Forma spent (the currency used to change polarities). I made my library of items searchable by category, description, or polarity, rather than just by name. I even added what has become my favorite feature: tooltips that appear when you click a stat, showing the value's underlying calculations.
(Credit: Anthropic/PCMag)Next, I turned my attention to the interface. I added descriptions to library items, consolidated mod slots into a central area, insisted on robust drag-and-drop functionality, made each section adjustable and minimizable, and trimmed excess spacing and text throughout. My efforts left me with a much cleaner app that felt significantly better to use.
(Credit: Anthropic/PCMag)Lastly, I asked Claude to refactor my code so that adding additional Warframes (build platform) beyond the one I started with would be easy. Then, I added support for four more Warframes, alongside a major expansion of my library to accommodate significant more items. I even had Claude design a mobile interface to make the app more pleasant to use on phones.
Now, the above all sounds great, but does my app actually work? And what about its underlying code? Is it barely functional spaghetti?
How My App Compares to Existing Tools
To put my app to the test, I looked at Overframe’s top build for Warframe’s latest character, Uriel, and I recreated it in my app. All the values in Overframe matched mine, except for effective hit points and strength. However, that’s not because my app is wrong. Overframe is wrong about hit points, as the site never updated its underlying calculation to reflect a change introduced by a prior Warframe update. As for strength, even when Overframe applies conditionals, it doesn’t take every mod into account. But my app does.
Aside from being more accurate than Overframe, my app also offers more features. I can outfit Warframes with Archon shards and see a Warframe’s stats reflect those bonuses. My app also conveniently lists all active effects on a given Warframe, unlike Overframe. Furthermore, my app is more transparent about its math, thanks to the clickable tooltips.
To be fair, my app doesn’t support as many items or Warframes as Overframe, nor can it build companions or weapons. However, adding more items and Warframes is trivially easy with a single prompt, and I could always expand my app to do more. Nonetheless, my app is more of a fun experiment and a personal-use tool than a full-fledged replacement for popular community resources.
As for my app’s underlying code, well, I wouldn’t know. But I did reach out to a senior programmer and asked them to take a look at it. They called out some inefficiencies, such as that my item library isn't stored in a dedicated database. And they pointed out that I would need a second opinion about possible security vulnerabilities if I were to ever formally release my app. But they didn't find any major red flags with the code, which I find quite impressive for something I built just by prompting a chatbot.
Try Vibe Coding Yourself
Not only is Claude good at making apps, but it’s really fun to do so. I never liked actual coding, but I do enjoy making things, so I found prompting Claude to create something I actually plan to use unexpectedly satisfying.
If there’s a digital tool you've always wanted, go try making it yourself with Claude. Sure, you won’t be making the next hit video game or a sprawling piece of software only possible with a team of dedicated developers, but what you can actually do might surprise you.