🧠 Custom MCP Server for AI-Powered Sticky Notes

---

📖 What is NoteFlow?

NoteFlow is a custom MCP (Model Context Protocol) server designed to bring persistent sticky notes into AI workflows.

Instead of manually managing notes in separate apps, users can simply ask their AI assistant to create, store, or retrieve notes using natural language.

When integrated with AI clients like Claude Desktop, NoteFlow allows the assistant to:

• Save reminders
• Store ideas
• Retrieve previously saved notes
• Maintain persistent context across conversations

All notes are stored locally in a simple .txt file, making the system lightweight, transparent, and fully customizable.

---

✨ Features

Feature	Description
🧠 Custom MCP Server	Implements a fully functional MCP server using FastMCP
📝 Natural Language Note Creation	AI assistants can add notes directly through commands
📖 Note Retrieval	Retrieve previously saved notes instantly
💾 Persistent Storage	Notes are stored locally in a .txt file
⚡ Fast Python Environment	Uses the modern uv package manager
🔗 Claude Desktop Integration	Automatically registers tools inside Claude
🧩 Extensible Design	Easily add new tools like delete, tag, or categorize notes

---

🏗️ System Architecture

Request Flow

graph TD
    A[👤 User Prompt] --> B[Claude Desktop AI]
    B --> C[MCP Client]
    C --> D[Custom MCP Server - FastMCP]
    D --> E{Command Type}

    E -->|Add Note| F[Write to notes.txt]
    E -->|Retrieve Notes| G[Read notes.txt]
    E -->|List Notes| H[Parse Stored Notes]

    F --> I[Return Confirmation]
    G --> I
    H --> I

    I --> B

---

MCP Workflow

The MCP server exposes custom tools that the AI client can call.

1️⃣ The user asks the AI assistant to save or retrieve a note 2️⃣ The assistant sends a tool request through MCP 3️⃣ The FastMCP server processes the request 4️⃣ Notes are written to or retrieved from notes.txt 5️⃣ The response is returned to the AI client

This architecture allows AI assistants to interact with local tools safely and efficiently.

---

🛠️ Technology Stack

Core Backend

Component	Technology
Programming Language	Python 3.12
MCP Server Framework	FastMCP
Package Manager	uv
Storage	Local .txt file
AI Client Integration	Claude Desktop

---

📂 Project Structure

ai-sticky-notes-mcp/
│
├── main.py            # MCP server logic (tools, prompts, handlers)
├── notes.txt          # Persistent storage for sticky notes
└── README.md          # Project documentation

---

🚀 Installation & Setup

Prerequisites

• Python 3.12+
• Claude Desktop (optional but recommended)

---

1️⃣ Install uv

Windows (PowerShell)

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Mac / Linux (curl)

curl -LsSf https://astral.sh/uv/install.sh | sh

Or with wget

wget -qO- https://astral.sh/uv/install.sh | sh

---

2️⃣ Initialize the Project

uv init .

---

3️⃣ Add MCP Dependency

uv add "mcp[cli]"

---

🏃 Running the MCP Server

Place all server logic inside main.py.

Then install the MCP server:

uv run mcp install main.py

This registers the MCP server so AI clients can discover it automatically.

---

🔗 Claude Desktop Integration

If Claude Desktop is installed:

1️⃣ Run:

uv run mcp install main.py

2️⃣ Claude will automatically detect your MCP server.

3️⃣ Restart Claude if tools do not appear:

• Open Task Manager
• End the Claude Desktop process
• Restart the app

Your MCP tools should now be available.

---

🌐 MCP Tools Example

Your server can expose tools such as:

Tool	Description
add_note	Save a new sticky note
get_notes	Retrieve all notes
search_notes	Find notes containing a keyword

Example interaction:

User: Remember that I need to finish my ML assignment tonight.

Claude: Note saved successfully.

---

⚙️ Configuration

Setting	File	Description
Notes Storage	notes.txt	Local file where notes are stored
MCP Tools	main.py	Define tools and commands
Client Integration	Claude Desktop	Registers MCP server automatically

---

🐛 Known Issues & Troubleshooting

MCP server not visible in Claude

Restart the Claude desktop application:

1. Open Task Manager
2. End the Claude process
3. Restart the app

---

Notes not saving

Ensure that:

• notes.txt exists in the project directory
• The MCP server has permission to write to the file

---

🔮 Future Improvements

• 📂 Note tagging and categorization
• 🧠 Semantic search using embeddings
• 🗑️ Delete or edit notes
• ☁️ Cloud storage integration
• 🔗 Multi-agent MCP workflows

---