SiftMarks

English | 简体中文

Turn your messy browser bookmarks into a local AI memory library.

Clean your bookmark bar. Search by memory. Give your AI tools a private knowledge base.

SiftMarks is a local-first bookmark manager for people who save too much and find too little. It imports your Chrome bookmarks, cleans up duplicate and scattered folders, makes the library searchable, and exposes your saved knowledge to AI tools through MCP.

The goal is simple: your bookmarks should feel like memory, not a junk drawer.

SiftMarks is built for developers, researchers, operators, founders, and anyone whose browser has quietly become a second brain with no search function.

Preview

Demo Video

Watch the 20-second product walkthrough:

If your Markdown viewer does not render embedded video, open docs/demo/siftmarks-demo.mp4 directly.

Product Screenshots

Review Bookmark Cleanup Like A Pull Request

Accept, dismiss, or batch-apply cleanup suggestions before anything touches Chrome.

Browse A Clean Local Bookmark Library

Filter by status, folder, duplicate state, missing metadata, and saved context.

Search By What You Remember

Search saved pages by keyword now, with memory-style search ready for configured AI providers.

Connect Bookmarks To AI Tools

Expose your local bookmark library to Claude, Cursor, Windsurf, and other MCP clients.

What It Does

• Import browser bookmarks from Chrome through the local web app or Chrome extension.
• Search by meaning or keywords instead of remembering exact titles.
• Rescue messy bookmark bars by detecting duplicates, vague titles, broken links, and folder chaos.
• Review cleanup suggestions first before applying them back to Chrome.
• Sync accepted cleanup back to Chrome through the extension, including duplicate removal and empty-folder cleanup.
• Generate tags and summaries with mock, OpenAI-compatible, or Ollama-compatible AI providers.
• Serve bookmarks to AI agents through an MCP server for Claude, Cursor, Windsurf, and other clients.
• Stay local by default with SQLite storage and no account requirement.

Why SiftMarks

Browser bookmarks are easy to save and hard to use. After a few months, the bookmark bar becomes a mix of imports, duplicates, vague titles, abandoned folders, and links you cannot search by memory.

SiftMarks treats bookmark cleanup like a code review:

1. Import the real browser state.
2. Generate cleanup suggestions.
3. Review and accept the changes you want.
4. Apply them back to Chrome with the extension.
5. Keep the cleaned library searchable for yourself and your AI tools.

Quick Start

Requirements:

• Node.js 18+
• npm
• Google Chrome if you want browser import/sync

npm install
npm run build
npm run dev

Open the local dashboard:

http://localhost:4399

SiftMarks stores its local database at:

~/.siftmarks/siftmarks.sqlite

Recommended Workflow

1. Start The Local App

npm run dev

Then open:

http://localhost:4399

The dashboard includes:

• Import for bookmark HTML files and local browser detection.
• Library for browsing bookmarks by folder, tag, and status.
• Search for finding saved pages.
• Rescue for reviewing cleanup suggestions.
• Settings for AI provider configuration.
• MCP for AI-client setup examples.

2. Load The Chrome Extension

The extension talks to the local app at http://localhost:4399.

1. Open chrome://extensions.
2. Enable Developer mode.
3. Click Load unpacked.
4. Select:

apps/chrome-extension

After loading it, click the SiftMarks extension icon.

3. Import Chrome Bookmarks

In the extension, click:

一键导入浏览器书签

This reads your current Chrome bookmarks and imports them into the local SiftMarks database.

4. Rescue The Library

Open the web dashboard:

http://localhost:4399/rescue

Click Generate or run rescue from the extension. SiftMarks will create cleanup suggestions such as:

• move bookmarks into clearer folders
• rename vague titles
• detect duplicate URLs
• mark broken links
• add tags

Review the suggestions, accept the ones you want, then sync them back.

5. Sync Accepted Changes Back To Chrome

In the extension, click:

同步回 Chrome

The extension applies accepted changes through the Chrome Bookmarks API. It can also:

• merge duplicate URLs
• remove empty folders
• consolidate scattered folders into readable top-level categories

Chrome will show a confirmation before modifying bookmarks. If Chrome Sync is enabled, those changes may sync to your Google account.

CLI

Build first:

npm run build:packages

Then use the local CLI:

npm run cli -- init
npm run cli -- stats
npm run cli -- doctor
npm run cli -- search "mcp browser automation"
npm run cli -- rescue
npm run cli -- export ./siftmarks-export.json

Import a browser-exported bookmark HTML file:

npm run cli -- import ./bookmarks.html

Index bookmarks for summaries, tags, and embeddings:

npm run cli -- index --limit 100

By default, SiftMarks uses the mock AI provider, so indexing does not send data to an external AI service unless you configure one.

MCP Server

SiftMarks can expose your bookmark library to AI clients through MCP.

Build the server:

npm run build:packages

Start it manually:

npm run cli -- mcp

Claude Desktop example:

{
  "mcpServers": {
    "siftmarks": {
      "command": "node",
      "args": ["/absolute/path/to/SiftMarks/apps/mcp-server/dist/index.js"]
    }
  }
}

Available MCP tools include:

Tool	Purpose
search_bookmarks	Search saved bookmarks
read_bookmark	Read one bookmark in detail
list_tags	List tags and counts
list_folders	List folders and counts
find_related_bookmarks	Find related saved pages
summarize_collection	Summarize by tag or folder
save_bookmark	Save a new bookmark
run_bookmark_rescue	Generate cleanup suggestions
get_bookmark_stats	Get library statistics

AI Providers

SiftMarks supports three AI modes:

Mode	Use Case
Mock	Default. No external AI calls. Good for local testing.
OpenAI Compatible	OpenAI, Azure OpenAI, Groq, Together, or compatible endpoints.
Ollama Compatible	Local models running through Ollama.

Configure providers in the web Settings page.

No external AI calls are made unless you explicitly configure a provider.

Privacy Model

SiftMarks is designed to be local-first:

• No account is required.
• Bookmark data is stored in local SQLite by default.
• The Chrome extension talks to your local app on localhost:4399.
• No telemetry is sent by the app.
• API keys are not logged.
• External AI calls are disabled unless you configure an AI provider.

Important: when you apply cleanup back to Chrome, Chrome itself may sync bookmark changes to your Google account if Chrome Sync is enabled.

Project Layout

siftmarks/
  apps/
    web/              Local Next.js dashboard and API
    cli/              Command-line interface
    mcp-server/       MCP stdio server
    chrome-extension/ Chrome extension for import and sync-back

  packages/
    shared/           Shared types and utilities
    db/               SQLite schema and data access
    core/             Import, search, rescue, cleanup logic
    ai/               Mock, OpenAI-compatible, and Ollama providers
    indexer/          FTS, summaries, tags, embeddings

Development

Common commands:

npm install
npm run dev
npm run build:packages
npm run build
npm run typecheck

For tests or experiments that should not touch your real bookmark library, point SiftMarks at a temporary home:

SIFTMARKS_HOME=/tmp/siftmarks-test npm run cli -- init

Current Status

SiftMarks currently includes:

• local web dashboard
• Chrome extension import
• Chrome sync-back for accepted cleanup
• SQLite storage
• keyword search and FTS indexing
• duplicate detection
• cleanup suggestions
• AI summaries and tags
• MCP server
• CLI workflow

Upcoming work may include richer semantic search, better folder policy editing, Firefox support, and full-page local archives.

License

MIT