I’ve been using Gemini to create infographic cards for my blog — typically one canvas with many images per set. Every time I had to screenshot them one by one, and if my hand shook, I’d spend forever adjusting. Absolutely maddening. Finally, I used Claude Code + Playwright MCP to completely solve this problem. Now I just let Claude Code run in the background taking screenshots. What a relief.
Watch the video — running one set takes under 4 minutes, video at 4x speed. About the same time as manual screenshots, maybe even slightly slower. But since it can run in the background while I do other things, it’s actually pretty convenient. The benefit is having CC handle position detection without needing complex OCR. Downsides: slow + burns tokens. But it’s consistent and precise — no more struggling to get perfectly aligned manual screenshots.
Just send it the link and claude.md each time. Initial setup requires configuring the Playwright MCP browser extension to connect to your default browser and maintain Gemini account login state. I hit some bumps early on using Chrome Dev MCP — that thing can maintain login states for other sites and products, but can’t log into Google accounts (probably restricted). Don’t recommend it — Playwright is more reliable.
I think this approach is quite the unorthodox hack. Vibe Operation — you could say I’ve discovered a non-coding use case for Claude Code, haha.
