Vibe Coding with Nuxt
summary
Builds a Nuxt app that scans a photo of a crossword, runs OCR and computer‑vision to detect the grid, then renders an editable, responsive crossword grid; uses Copilot agent (agent mode), Nuxt/Nitro, Tailwind v4, and a pulled‑in OpenCV OCR model. The live session generates a working upload/processing flow and an interactive grid within about 20 minutes, but grid detection and some template errors remain and need manual fixes and tests before production.
AI-written summary · about
chapters (14)
- 0:00Vibe Coding with Nuxt
- 1:29What is Nuxt
- 3:37Vibe Coding and LLMs
- 7:07Working with LLMs
- 8:33Nuxt project setup
- 10:02Copilot instructions config
- 13:17Crossword app prompt
- 15:01LLM generates project
- 18:36Refining LLM code
- 24:16Switching models and errors
- 27:26Testing LLM code
- 28:28Progress update
- 31:24Crossword app demo
- 31:46Closing thoughts
related talks
35mVue-doo Magic: AI Development Tricks
55mNuxt.js: isomorphic Vue.js Apps made easy, fast and fun
27mUniversal Vue applications with Nuxt.js
45mAn AI-Enabled Framework? Using LLMs to Put People First
11mNuxt js + Netlify CMS
26mBe Lazy, Be Smart, Be Nuxt.
23mFull Stack with Nuxt
44mNuxt.js - Why? How? Oh yeah!