
Pinboard bookmarks MCP server
概览
Read-only access to Pinboard.in bookmarks for LLMs via Model Context Protocol (MCP). This server provides LLMs with the ability to search, filter, and retrieve bookmark metadata from Pinboard.in at inference time. Built on FastMCP, it offers five tools for bookmark interaction while respecting Pinboard's rate limits and implementing intelligent caching. - to Pinboard bookmarks - : search_bookmarks, search_bookmarks_extended, list_recent_bookmarks, list_bookmarks_by_tags, list_tags - with LRU cache and automatic invalidation using posts/update endpoint - respects Pinboard's 3-second guideline between API calls - converts Pinboard's legacy field names to intuitive ones (description→title, extended→notes) - for MCP Apps hosts: bookmarks and tags render as interactive cards in Claude, ChatGPT and other hosts that support MCP Apps; other hosts see plain results as before - with integration test harnesses and CI validation 1. from https://pinboard.in/settings/password 2. : 3. : 4.
README
以下为原文 README 节选,完整内容请查看原文。
Pinboard MCP Server
Read-only access to Pinboard.in bookmarks for LLMs via Model Context Protocol (MCP).
Overview
This server provides LLMs with the ability to search, filter, and retrieve bookmark metadata from Pinboard.in at inference time. Built on FastMCP, it offers five tools for bookmark interaction while respecting Pinboard’s rate limits and implementing intelligent caching.
Features
- Read-only access to Pinboard bookmarks
- Five MCP tools:
search_bookmarks,search_bookmarks_extended,list_recent_bookmarks,list_bookmarks_by_tags,list_tags - Smart caching with LRU cache and automatic invalidation using
posts/updateendpoint - Rate limiting respects Pinboard’s 3-second guideline between API calls
- Field mapping converts Pinboard’s legacy field names to intuitive ones (description→title, extended→notes)
- Result cards for MCP Apps hosts: bookmarks and tags render as interactive cards in Claude, ChatGPT and other hosts that support MCP Apps; other hosts see plain results as before
- Comprehensive testing with integration test harnesses and CI validation
Installation
Via pip (recommended)
pip install pinboard-bookmarks-mcp-server
From source
git clone https://github.com/rossshannon/pinboard-bookmarks-mcp-server.git
cd pinboard-bookmarks-mcp-server
pip install -e .
Quick Start
- Get your Pinboard API token from https://pinboard.in/settings/password
- Set environment variable:
export PINBOARD_TOKEN="username:1234567890ABCDEF" - Start the server:
pinboard-mcp-server - Verify it’s working:
# Test help command (works without token) pinboard-mcp-server --help # Server should show "Starting MCP server" when run with token
Usage with Claude Desktop
Add this configuration to your Claude Desktop settings:
{
"mcpServers": {
"pinboard": {
"command": "pinboard-mcp-server",
"env": {
"PINBOARD_TOKEN": "your-username:your-token-here"
}
}
}
}
Available Tools
1. search_bookmarks
Search bookmarks by query string across titles, notes, and tags. Recent-focused with automatic expansion.
Parameters:
query(string): Search querylimit(int, optional): Maximum results (default: 20, max: 100)
Example:
Search for "python testing" bookmarks
2. search_bookmarks_extended
Extended search for comprehensive historical results across titles, notes, and tags.
Parameters:
query(string): Search query
安装
This server does not publish a one-line install command.
Open the repository installation guide 新手?先看 mcpskill.org 教程 →配置
{
"mcpServers": {
"pinboard": {
"command": "pinboard-mcp-server",
"env": {
"PINBOARD_TOKEN": "your-username:your-token-here"
}
}
}
}