
What happened
A Zenn guide says vague prompts like "add a nice animation" carry no spec, so Claude or ChatGPT decides element, motion, duration and distance itself — producing uniformly flashy results — and it prescribes a four-part request: name (Japanese and English), location, numbers, reduced-motion note.
Why it matters
Without that structure, the same prompt yields a different, over-the-top result each time, so users keep rephrasing without getting closer to the intended motion, the guide says.
Summaries like this, in your inbox every morning.
The Zenn article is written for people who use Claude or ChatGPT to build landing pages and keep getting motion they did not ask for. It traces the problem to the prompt itself: words like "nice" or "stylish" leave the AI to choose which element moves, which motion to use, how long it takes and how far it travels. The result, the guide says, is a uniformly flashy output rather than a targeted one.
The proposed fix is to name the motion in both Japanese and English, state where it goes, give speed and distance as numbers, and add a note that respects the OS reduced-motion setting. The guide presents this as a repeatable request pattern rather than a one-off trick, and it illustrates the pattern with six common web motions, each paired with a sample prompt and CSS. It also points readers to a separate reference site that lists 32 motions with live samples and adjustable sliders, so a user can settle on a numeric value by eye and copy the matching prompt.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Docker published an analysis of major AI coding agent incidents from October 2024 to February 2026, including…

Using ClickCapture screenshots of Robocopy Scheduler, Gemini 2.0 Flash-Lite produced a structured parameter ta…

After two ✓-marked items turned out wrong post-publication — a nonexistent "創業枠(200万円)" and a 3-month photo ru…

A weekend solo engineer asked Claude Code to move a skill that triggers CodeRabbit reviews every 65–70 minutes…

Drawing on 11 years as a 臨床工学技士, the author proposes six fields for AI agent approval logs: when, who, what ta…

A non-engineer built an AI receptionist app for the 5th Agentic AI Hackathon with Google Cloud by vibe coding…
