Bookmark Smart

For everything worth finding again.

Save the good stuff.Find the right thing.

Save pages to a browser-local collection and find them by meaning. Bookmark Smart uses a small embedding model on your device to bring relevant pages back into view.

Download Preview 1 Manual installation
3.0.0 Preview 1Independently built by Nick Dudczyk.
Bookmark Smart semantic search finding balcony gardening pages for the query grow food in a small space.
Actual extension · Local model search over 24 sample pagesOpen image to view full size ↗
Bookmark Smart ranking a bicycle puncture guide first for a query about fixing a flat tire.
A second search over the same synthetic collectionOpen image to view full size ↗
Browser extensionLocal model inferenceSemantic search

Less digging. More discovering.

Your bookmarks, with a sense of direction.

01

Start with what you saved.

Capture the text of a webpage and save it to your local collection as you browse.

02

Search beyond the title.

Semantic search uses local embeddings to connect your query with related saved content.

03

Pick up where you left off.

Saved pages stay in your browser profile across restarts, ready for your next search.

04

Keep the model close.

An embedding model runs in your browser. Model files are downloaded and cached during setup; inference happens locally.

The everyday workflow

Three steps.
One place.

  1. 01

    Capture

    Save an ordinary webpage from the extension popup.

  2. 02

    Search

    Describe the idea you want to find.

  3. 03

    Return

    Reopen your browser and search your saved collection.

Thoughtful by design

A useful place for a small model.

Bookmark Smart uses embeddings to represent the meaning of saved content. A quantized sentence model runs locally through Transformers.js, while IndexedDB stores the collection. The first use downloads model files; subsequent searches run on your device.

  • Local collection storage
  • Semantic similarity search
  • Saved pages across restarts
  • No API key or paid account required
3.0.0 Preview 1

3.0.0 Preview 1 is available.

Download the built extension from GitHub and install it manually in Chrome. This public preview focuses on page capture, semantic search, and a persistent local collection.

Download Preview 1 ↗View on GitHub ↗
  1. Download bookmark-smart-3.0.0-preview.1.zip from the release assets, then extract it to a permanent folder. Choose the built ZIP, not a source-code archive.
  2. In a separate Chrome profile, open chrome://extensions, enable Developer mode, and select Load unpacked. Choose the folder containing manifest.json.
  3. Pin Bookmark Smart, open a webpage, then choose Capture → Save Current Page. Keep the popup open until the save completes. The initial model download needs an internet connection.
Full installation & update guide ↗

A few useful details.

Can I install it today?

Yes. Download 3.0.0 Preview 1 from GitHub and follow the manual installation guide. It is not listed in the Chrome Web Store.

Does it run locally?

The collection is stored locally and the embedding model runs in the browser. Initial setup downloads model assets, so a fresh installation needs an internet connection.

What is included in this preview?

Capture, semantic search, and saved collections were exercised in isolated Chromium. Clustering, history import, and browser-bookmark tools are experimental. Use a separate profile; the manual Chrome installation flow still needs owner verification.

How do updates work?

Updates are manual. Keep the same installation folder and browser profile, and follow the update guide. There is no automatic updater or supported collection backup/export in this preview.

How is it related to BookmarkTools?

BookmarkTools is a companion data-preparation project. Bookmark Smart is the browser product, with its own embedding-based search and organization workflow.