# codevember 2026 - day 1: Fold

I'm doing Codevember this year, following [Jaume Sanchez Elias' initiative](https://x.com/thespite/status/2116974998212772095). I will probably not complete every day, but I'll try to do as much as possible and document the journey here.

Today I worked on the first prompt, "fold". Here's the result: https://www.grgrdvrt.com/work/codevember-2026/01_fold

I'm not touching the code, this is purely vibe-coded. Last time I did Codevember was in [2019](https://grgrdvrt.com/works/codevember-2019/). I created 15 pieces (out of 30 prompts) and found the pace brutal. I remember learning a few things, but they were not strictly technical. I think I can do and learn more without writing the code myself.

Finding interesting ideas is still difficult because it's too tempting to pick the first one and implement it. This is what I did today. The main difference compared to before AI is that I don't consider the technical feasibility but rather how I would describe the idea to the agent.
For the rest of the month I'll spend more time elaborating on the idea before I start the implementation.

I gave a lot of directions to the agent, but still let it fill the gaps in almost all areas. For example I never mentioned the camera or the lighting. This is probably a mistake, but I'm still satisfied with the result.
While I think I could have implemented it myself, it would have taken me several days.

What I lack the most in my current AI setup is an easy way to send sketches to the agent. I'll work on that the coming days.