Bloom Pinboard MCP Server

A Model Context Protocol (MCP) server that connects Claude to Pinboard, enabling AI-assisted bookmark management and collective knowledge discovery for the Bloom network.

The Bloom Tagging Convention

bloom is a signal tag that means: "I want this resource surfaced to others in the network."

Pair it with a location tag to help people find local resources:

• bloom + houston
• bloom + austin
• bloom + detroit
• bloom + global (for non-location-specific resources)

This creates a decentralized, federated knowledge layer. Anyone can participate by tagging their bookmarks with bloom. No central authority required — just a shared convention.

What This Does

This MCP server lets Claude:

• Search and analyze your Pinboard bookmarks
• Discover what others tagging bloom are saving
• Find connections between community members based on shared interests
• Save new bookmarks with smart tagging
• Track new members joining the bloom network

Prerequisites

• Python 3.10+ installed on your computer
• Claude Desktop app (not the web version)
• Pinboard account with API access

Installation

1. Clone this repository

git clone https://github.com/YOUR-USERNAME/bloom-pinboard-mcp.git
cd bloom-pinboard-mcp

2. Install Python dependencies

pip install httpx mcp

3. Get your Pinboard API token

1. Log into Pinboard
2. Go to https://pinboard.in/settings/password
3. Copy your API token (looks like: username:ABC123DEF456...)

4. Find your Python path

Windows:

where python

Mac/Linux:

which python3

Copy the full path that appears.

5. Configure Claude Desktop

Claude Desktop needs a configuration file to know about MCP servers.

Config file location:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

Create or edit this file with the following (replace the placeholders):

{
  "mcpServers": {
    "pinboard": {
      "command": "C:\\Path\\To\\Your\\python.exe",
      "args": [
        "C:\\Path\\To\\Your\\bloom-pinboard-mcp\\server.py"
      ],
      "env": {
        "PINBOARD_API_TOKEN": "your-username:YOUR_API_TOKEN_HERE"
      }
    }
  }
}

Important:

• Use double backslashes (\\) on Windows
• Use the full absolute path to both Python and server.py
• Keep your API token private — never commit it to git!

6. Restart Claude Desktop

Completely quit Claude Desktop (don't just close the window — use Cmd+Q on Mac or right-click the system tray icon on Windows) and reopen it.

Verify It's Working

In Claude Desktop, try asking:

• "Show me my Pinboard tags"
• "Search my bookmarks for houston"
• "Find bloom-tagged resources about mutual aid"

If Claude can answer these using your Pinboard data, you're all set!

Available Tools

Personal Bookmarks

Tool	Description
search_my_bookmarks	Search your personal Pinboard collection
get_my_tags	List all your tags with counts
save_bookmark	Save a new bookmark with tags
suggest_tags	Get tag suggestions for a URL

Bloom Collective Discovery

Tool	Description
search_collective	Search bloom-tagged bookmarks across the network
get_recent	Get recent bloom bookmarks
get_collective_overview	Full snapshot of the collective in one call
who_knows_about	Find contributors by topic
find_connections	Find interest overlaps between members
get_network_map	Generate relationship graph

User Discovery

Tool	Description
get_user_profile	Complete profile of any Pinboard user
search_user_bookmarks	Search a specific user's public bookmarks
find_common_ground	Compare two users' shared interests
who_else_saved	Check who else saved a URL

Member Tracking

Tool	Description
discover_new_members	Find new contributors using bloom tags
list_known_members	List all tracked bloom network members
reset_known_members	Clear the known members list

Utilities

Tool	Description
get_feed_url	Generate RSS/JSON feed URLs for any user or tag
clear_cache	Clear the in-memory cache
search_public_pinboard	Search all public Pinboard bookmarks

Customizing for Your Location

The server is configured with a COLLECTIVE_TAG variable in server.py. You can change this to match your local bloom chapter or search for bloom broadly:

COLLECTIVE_TAG = "bloom"  # Find all bloom-tagged resources globally

The tools will find anyone using the bloom convention, and you can filter by location in your queries.

Generate Your Own Viewer

Example viewer: https://claude.ai/public/artifacts/c5a8be0f-f34d-4749-809f-432fe4995613

Want a custom version for your city or network? Ask Claude:

Prompt: "Connect to the Pinboard MCP and create a Bloom Network directory artifact. Fetch all bookmarks tagged with bloom and [your-city]. Include:

Generative AI search that matches user intent to resources Topic tag filtering A way for people to see how to contribute Clean, accessible design

Pull the real data from my Pinboard account and embed it in the artifact."

Or for a simpler static version:

"Query my Pinboard for all bookmarks tagged bloom and [your-city], then create a searchable directory artifact with those resources."

Troubleshooting

"python.exe is not recognized"

Your Python path is wrong. Run where python (Windows) or which python3 (Mac/Linux) and use that exact path.

Server not showing up in Claude

1. Check your JSON syntax (missing commas, quotes, etc.)
2. Make sure paths are absolute, not relative
3. Completely quit and restart Claude Desktop

"No bookmarks found"

• Check that your API token is correct
• Try clear_cache to force fresh API calls

Check the logs

Claude Desktop logs MCP errors to:

• Windows: %APPDATA%\Claude\logs\mcp-server-pinboard.log
• Mac: ~/Library/Logs/Claude/mcp-server-pinboard.log

Security Notes

• Never commit your API token to git — use the example config as a template
• Your token is stored only in your local Claude Desktop config
• The server makes read/write calls to Pinboard's API on your behalf

About Bloom Network

Bloom Network is a grassroots movement building regenerative communities worldwide.

The bloom tag on Pinboard serves as a shared knowledge layer — a way for organizers, activists, and community builders to surface resources to each other without any central platform. Tag something bloom and it becomes discoverable. Add your city and it becomes locally findable.

This is infrastructure for collective intelligence: simple, decentralized, and owned by no one.

License

MIT

Contributing

Pull requests welcome! Ideas for improvement:

• Dynamic location filtering (search bloom + any location)
• Cross-location discovery (find resources tagged in multiple cities)
• Digest/notification tools for new bloom content
• Integration with other knowledge-sharing tools

If you start a bloom chapter in your city, let us know!