mcp-server-demo (AI sticky notes)

mcp-server-demo (AI sticky notes)

AI Sticky Notes: A Model Context Protocol (MCP) application that enables Claude Desktop to create, read, and manage persistent sticky notes.

Overview

AI Sticky Notes leverages the Model Context Protocol to provide a seamless integration between Claude Desktop and a simple note-taking system. This implementation allows Claude to:

• Save notes to persistent storage
• Retrieve all saved notes
• Access the most recently added note
• Generate prompts to summarize existing notes

Installation

Prerequisites

• Python 3.7+
• MCP Python SDK
• Claude Desktop application

Setup

1. Clone this repository:

   • git clone https://github.com/aymaneELHICHAMI/mcp-server-demo.git
   • cd mcp-server-demo

2. Install uv

   • Documentation: https://docs.astral.sh/uv/getting-started/installation/

3. Create a virtual environment

   • uv venv venv

4. Activate the virtual environment

   • uv venv

5. Install dependencies from uv.lock

   • uv sync

6. Run the MCP server

   • uv run mcp install main.py If the configuration is not reflected in Claude Desktop (Menu Bar -> Settings -> Developer), Try to force shutdown Claude Desktop from task Manager et reopen it. When working, you should see "AI Sticky Notes MCP Server" running.

Core Components

from mcp.server.fastmcp import FastMCP
mcp = FastMCP("AI Sticky Notes")

The server registers itself as "AI Sticky Notes" in the MCP ecosystem, making it discoverable by clients like Claude Desktop.

Notes are stored in a simple text file (notes.txt), with each note on a separate line. The file is automatically created if it doesn't exist.

The server exposes the following MCP capabilities:

Type	Name	Description
Tool	add_note	Append a new note to storage
Tool	read_notes	Retrieve all existing notes
Resource	notes://latest	Access the most recent note
Prompt	note_summary_prompt	Generate a prompt for summarizing notes

Using with Claude Desktop

Once the server is running, Claude Desktop will automatically discover and connect to it. You can interact with your notes through natural language:

You: "Please save this as a note: Remember to schedule a team meeting for Friday"
Claude: [Uses add_note tool] "Note saved!"

You: "What notes do I have?"
Claude: [Uses read_notes tool] "Here are your notes: ..."

You: "What was my most recent note?"
Claude: [Accesses notes://latest resource] "Your most recent note is: ..."

You: "Can you summarize my notes?"
Claude: [Uses note_summary_prompt] "Here's a summary of your notes: ..."

Resources

• Model Context Protocol Documentation: https://modelcontextprotocol.io/introduction

• FastMCP API Reference: https://github.com/jlowin/fastmcp

• Claude Desktop Documentation: https://docs.anthropic.com/en/docs/welcome