# Bookmarks Mcp

> Chrome Bookmarks MCP to allow searching with AI

- **Type:** MCP server
- **Install:** `agentstack add mcp-robamills-bookmarks-mcp`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [RobAMills](https://agentstack.voostack.com/s/robamills)
- **Installs:** 0
- **Category:** [Integrations](https://agentstack.voostack.com/c/integrations)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [RobAMills](https://github.com/RobAMills)
- **Source:** https://github.com/RobAMills/bookmarks_mcp

## Install

```sh
agentstack add mcp-robamills-bookmarks-mcp
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Bookmarks MCP Server

A Model Context Protocol (MCP) server that provides AI agents with read-only access to Chrome bookmarks, enabling intelligent search, analysis, and organization of bookmarked content.

## Features

- 🔍 **Search Bookmarks**: Find bookmarks by keywords in title, URL, or folder path
- 📊 **Analyze Collection**: Get statistics about your bookmark collection
- 🌐 **Domain Analysis**: See which websites you bookmark most frequently
- 📁 **Filter by Root**: Access Bookmarks Bar, Other Bookmarks, or Synced bookmarks
- ⚡ **Fast Performance**: JSON-based (no SQLite locking, works while Chrome is open!)
- 🔒 **Read-Only**: Safe to use - no modifications to your bookmarks

## Use Cases

- **Find old bookmarks**: "What did I save about Python testing?"
- **Discover patterns**: "Which sites do I reference most often?"
- **Organize insights**: "These 50 dev bookmarks are scattered across 10 folders"
- **Rediscover content**: "Show me all my AI/ML bookmarks"

## Requirements

- Python 3.10 or higher
- Google Chrome (with bookmarks)
- MCP-compatible AI agent (e.g., Goose Desktop)

## Installation

### Use with Goose Desktop

1. **Open Goose Desktop** and go to **Settings → Extensions**

2. **Click "Add custom extension"**

3. **Fill in the form:**
   - **Name**: `Bookmarks MCP`
   - **Command**: 
     ```
     uvx --with mcp[cli] mcp run [path to mcp code]/server/main.py
     ```
   
   Replace `[path to mcp code]` with your actual path, for example:
   ```
   uvx --with mcp[cli] mcp run [mcp install dir]/bookmarks_mcp/server/main.py
   ```

4. **Click "Add"** - you're ready to go!

### Use with MCP CLI

```bash
# Install dependencies
pip install mcp

# Run the server
python server/main.py
```

## Quick Start

### Step 1: Check Status

First, verify that your Chrome bookmarks can be accessed:

```python
check_bookmarks_file_status()
```

Expected response:
```json
{
  "status": "ready",
  "bookmarks_path": "/Users/rob/Library/Application Support/Google/Chrome/Default/Bookmarks",
  "file_size": 1234567,
  "total_bookmarks": 1234,
  "total_folders": 56
}
```

### Step 2: Get All Bookmarks

Retrieve all bookmarks:

```python
get_all_bookmarks()
```

Or get specific root only:

```python
get_all_bookmarks(root="bookmark_bar")
```

### Step 3: Search Bookmarks

Find bookmarks by keyword:

```python
search_bookmarks_by_query("github")
```

Search in specific fields:

```python
search_bookmarks_by_query("docs.python.org", search_in=["url"])
```

### Step 4: Analyze Collection

Get overview statistics:

```python
analyze_bookmarks_overview()
```

See top domains:

```python
analyze_bookmarks_domains(limit=20)
```

## Available Tools

### 1. `health_check()`
Quick test to verify the MCP server is running.

### 2. `check_bookmarks_file_status()`
Check if Chrome bookmarks file is accessible and get basic information.

**Returns:**
- `status`: "ready", "file_not_found", "permission_denied", or "error"
- `bookmarks_path`: Path to Chrome bookmarks file
- `total_bookmarks`: Total number of URL bookmarks
- `total_folders`: Total number of folders
- `roots`: Available root folders

### 3. `get_all_bookmarks(root="all", include_folders=True, max_depth=None)`
Retrieve all Chrome bookmarks with filtering options.

**Parameters:**
- `root`: "all", "bookmark_bar", "other", or "synced"
- `include_folders`: Include folder entries in results
- `max_depth`: Maximum folder depth to traverse

**Returns:**
- `bookmarks`: List of bookmark entries
- `summary`: Statistics about retrieved bookmarks

### 4. `search_bookmarks_by_query(query, search_in=None, case_sensitive=False, root="all", max_results=100)`
Search bookmarks by keywords.

**Parameters:**
- `query`: Search term
- `search_in`: List of fields - ["title", "url", "folder"]
- `case_sensitive`: Enable case-sensitive search
- `root`: Which root to search
- `max_results`: Maximum results to return

**Returns:**
- `results`: Matching bookmarks
- `total_found`: Number of results
- `search_time_ms`: Search duration

### 5. `analyze_bookmarks_overview()`
Get comprehensive statistics about your bookmark collection.

**Returns:**
- `total_bookmarks`: Total URL bookmarks
- `total_folders`: Total folders
- `max_depth`: Maximum folder hierarchy depth
- `date_range`: Oldest and newest bookmark dates
- `roots_summary`: Breakdown by root
- `average_bookmarks_per_folder`: Average bookmarks per folder

### 6. `analyze_bookmarks_domains(limit=20)`
Analyze which domains you bookmark most frequently.

**Parameters:**
- `limit`: Maximum number of top domains to return

**Returns:**
- `top_domains`: List of domain statistics with counts and examples
- `unique_domains`: Total unique domains
- `bookmarked_domains_count`: Domains with multiple bookmarks

## Data Structure

Each bookmark entry contains:

```python
{
    "id": "unique_identifier",
    "name": "Bookmark Title",
    "url": "https://example.com",
    "type": "url",  # or "folder"
    "folder_path": ["Bookmarks Bar", "Dev Resources"],
    "date_added": "2024-01-15T10:30:00",
    "date_added_timestamp": 13278388739951719,
    "indent_level": 1
}
```

## Chrome Bookmarks Location

The server automatically detects Chrome bookmarks file location:

- **macOS**: `~/Library/Application Support/Google/Chrome/Default/Bookmarks`
- **Windows**: `%LOCALAPPDATA%\Google\Chrome\User Data\Default\Bookmarks`
- **Linux**: `~/.config/google-chrome/Default/Bookmarks`

## Advantages Over Browser History

- ✅ **No file locking**: Works while Chrome is running
- ✅ **JSON format**: Simple parsing (no SQLite)
- ✅ **Tree structure**: Preserves folder hierarchy
- ✅ **Permanent**: Bookmarks don't auto-delete like history
- ✅ **Curated**: Only content you intentionally saved

## Project Status

**Version**: 0.1.0 (Alpha)

**Implemented**:
- ✅ Chrome bookmarks parsing
- ✅ Search by title/URL/folder
- ✅ Overview analysis
- ✅ Domain analysis
- ✅ Cross-platform support (macOS, Windows, Linux)

**Planned** (Phase 2):
- ⏳ Duplicate detection
- ⏳ Folder structure analysis
- ⏳ Caching layer (5-minute TTL)
- ⏳ Fuzzy search

**Future** (Phase 3):
- ⏳ Organization suggestions
- ⏳ Firefox/Safari support
- ⏳ Integration with browser history

## Development

### Project Structure

```
bookmarks_mcp/
├── server/
│   ├── __init__.py
│   ├── main.py              # MCP server setup
│   ├── bookmarks_parser.py  # JSON parsing and tree traversal
│   └── local_types.py       # Type definitions
├── tests/
│   └── fixtures/
├── README.md
├── TODO.md
├── CHANGELOG.md
└── pyproject.toml
```

### Running Tests

```bash
# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest
```

## Troubleshooting

### "file_not_found" Error

**Cause**: Chrome is not installed or bookmarks file doesn't exist.

**Solution**: 
1. Verify Chrome is installed
2. Open Chrome and create at least one bookmark
3. Check the path in the error message

### "permission_denied" Error

**Cause**: Insufficient permissions to read bookmarks file.

**Solution**:
- **macOS/Linux**: Check file permissions with `ls -l ~/Library/Application Support/Google/Chrome/Default/Bookmarks`
- **Windows**: Run Goose Desktop as administrator

### Empty Results

**Cause**: No bookmarks match your criteria or bookmarks file is empty.

**Solution**:
- Verify you have bookmarks in Chrome
- Try `get_all_bookmarks()` without filters
- Check `analyze_bookmarks_overview()` for statistics

## Changelog

See [CHANGELOG.md](CHANGELOG.md) for version history.

## Contributing

This is just a fun side project done with Goose and GMT-4.7 

I have no plans to support this but feel free to fork it. 

## License

MIT License - see LICENSE file for details

## Acknowledgments

- Inspired by [browserhistorymcp](https://github.com/mixophrygian/browser_history_mcp)
- Built with [MCP Python SDK](https://github.com/modelcontextprotocol/python-sdk)

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [RobAMills](https://github.com/RobAMills)
- **Source:** [RobAMills/bookmarks_mcp](https://github.com/RobAMills/bookmarks_mcp)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/mcp-robamills-bookmarks-mcp
- Seller: https://agentstack.voostack.com/s/robamills
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.