ScrollSplice

• 

  Light Mode

• 

  Dark Mode

Inspiration

I recently learned about the WEBTOONS website and scrolling comics blew my mind, so I decided to make an app that would make WEBTOONS for me.

What it does

It creates WEBTOONS via an OpenAI OAuth into a chat window built into the app. A user can also manually add images and do it themselves.

How we built it

I brainstormed what I wanted with GPT Live-1 and built on top of that with GPT 5.6 using my Bootstrap Documents skill and iterated as I went. July 12 was planning only and implementation occurred July 13–21. We validated the build with 423 unit tests, 11 companion checks, and 24 browser stories.

Challenges we ran into

Ran into a bug where the agent wouldn't place the text in the actual speech bubble. We had to build a a visual correction into the adapter along with many other bugs.

Accomplishments that we're proud of

Getting the OAuth using OpenAI authentication was especially a big point for me. I had to convince GPT 5.6 and GPT Live-1 that it was indeed possible using the Harnesses Codex, OpenClaw, and Hermes as examples for it to understand what I wanted to do.

What we learned

There are so many little things I learned along the way. How OAuth integration works, how painstakingly detailed you have to be to have things work the way you want. You have to be very articulate. This whole project was vibe coded. I do have some experience with programming but not enough to write my own programs.

What's next for ScrollSplice

Clean up, more functionality, better assets, tweaked custom instructions, and adding a wiki-like memory so it remembers everything about your project(s).

Built With

• codex
• css3
• eslint
• github
• gpt-5.6
• konva
• node.js
• playwright
• pnpm
• react
• react-konva
• typescript
• vite
• vitest
• zustand