stepkeeper

Turn videos into documents, recipes, and user manuals. Ambiguous spoken steps get the actual video frame.

stepkeeper turns a YouTube how-to video into a follow-along document. Spoken instructions like "cut it bite-sized" or "simmer until the sauce reduces" don't mean much as text. stepkeeper finds the moment where that state is actually visible, captures candidate frames directly from the player, lets you pick the right one, and builds a markdown document with the frames embedded. - Works on cooking, repair, crafts, beauty, fitness, software tutorials - Three candidate frames (before / center / after) per ambiguous step; you pick - Nothing is downloaded from YouTube — frames come from the player you're watching - Bring your own Gemini API key (free tier works); no account, no tracking - Output: document.md + images, ready for Obsidian or any markdown app How to use: 1. Enter your Gemini API key in the extension options (free at aistudio.google.com/apikey) 2. Open a YouTube how-to video, play it for a moment 3. Click the stepkeeper button (bottom right) → pick frames → download document