‹ YumoTab

The YumoTab import format

A tiny, open JSON format for moving organized bookmarks into YumoTab. Anyone can produce it — a script, another app, or an AI assistant.

At a glance

{
  "format": "yumotab.import/1",
  "categories": [
    { "name": "Work", "sites": [
      { "title": "Notion", "url": "https://www.notion.so" },
      { "title": "Figma",  "url": "https://www.figma.com" }
    ] },
    { "name": "Recipes", "sites": [
      { "title": "Serious Eats", "url": "https://www.seriouseats.com" }
    ] }
  ]
}

Fields

Field	Type	Meaning
`format`	string	Always `"yumotab.import/1"`. Recommended; YumoTab also accepts files without it.
`categories`	array	The categories to create or fill, in order.
`categories[].name`	string	Category name, up to 40 characters. A name matching a built-in category (“Work”, “AI”, “工作”…) fills that category instead of creating a new one.
`categories[].sites`	array	The sites in this category, in the order they should appear.
`sites[].title`	string	The name shown under the icon. Keep it short — “GitHub”, not “GitHub - Where the world builds software”.
`sites[].url`	string	An `http` or `https` address. Anything else is skipped.

What YumoTab does with it

Removes duplicate addresses, ignoring www., trailing slashes, #fragments and tracking parameters such as utm_source.
Tidies titles and uses each known site’s official icon; other sites get a clean letter tile.
Shows a preview first — people can untick sites and rename categories before anything is saved.
Never adds a site twice to the same category, and the whole import can be undone in one step.
Text before or after the JSON (and a ```json fence) is ignored, so an AI’s answer can be pasted as is.

Organize with AI

Export your bookmarks from your browser (every browser can save them as an HTML file), then give that file and this prompt to ChatGPT, Claude, Doubao or any other assistant. Paste its answer into YumoTab → Import Bookmarks → Paste from AI.

You are organizing my browser bookmarks so I can import them into YumoTab, a new-tab page. My exported bookmarks file is attached (or pasted below).

Rules:
1. Remove duplicates, login/redirect/tracking links and anything that isn't a normal web page.
2. Group the rest into 4–10 clear categories with short names (1–2 words), written in the same language as my bookmarks.
3. Give every site a short, clean name — "GitHub", not "GitHub - Where the world builds software".
4. Put the sites I'm most likely to use first in each category.
5. Reply with JSON only — no explanation, no code fence — in exactly this format:

{
  "format": "yumotab.import/1",
  "categories": [
    { "name": "Work", "sites": [
      { "title": "Notion", "url": "https://www.notion.so" },
      { "title": "Figma", "url": "https://www.figma.com" }
    ] }
  ]
}

Large collections may not fit in one conversation. Split the export into a few parts, or let YumoTab’s built-in “Sort Automatically” do the first pass — it needs no AI at all.

Also accepted

YumoTab reads the standard bookmarks HTML file (the “Netscape bookmark format”) that Chrome, Edge, Safari, Firefox and others export, and — inside the extension — can read the browser’s bookmarks directly with your permission.