Show HN: Round 1 of a vibe-coder ladder, judged by an LLM that only sees the UI (vibeladder.dev)

In an innovative competition dubbed "vibe-coder ladder," two participants showcased their coding skills by developing a personal notes app within a 90-minute timeframe. The app, which had to feature Markdown support, live previews, and persistence, was evaluated by an LLM that only interacted with the user interface and never inspected the source code. @P1s0 emerged as the winner, utilizing a lengthy, iterative approach with the Claude model, while @schwarzkopfb took a more rapid, direct route with GPT-5.6 Sol, ultimately highlighting the significant impact of testing and refinement in app development. This competition underscored the importance of iterative design and user testing in AI-assisted coding. While both entries met the functional requirements, the winner's attention to detail and user experience—with features like effective layout handling and enhanced task management—demonstrated that thorough preparation and an in-depth understanding of user needs can lead to superior results. The event also revealed challenges in judgment processes, prompting improvements to the testing harness based on feedback. With such lessons and a focus on iterative refinement, this round sets a meaningful precedent for future AI-driven coding competitions and emphasizes the value of comprehensive design cycles in developing robust applications.