This project demonstrates how to build a custom MCP (Model Context Protocol) server that exposes tool-powered functionality for creating, updating, listing, and deleting AI-generated sticky notes.

The MCP server integrates seamlessly with Claude Desktop, where:

• Claude (LLM) acts as the MCP client
• Claude Desktop acts as the MCP host
• Your MCP Server registers custom tools
• Users interact naturally through chat
• Claude intelligently invokes your server tools to manage sticky notes

This project is a real example of Agentic Tool Use, powered entirely by MCP.

🚀 Features

✔️ Build your own MCP server (Model Context Protocol)

✔️ Claude automatically detects and calls your tools

✔️ Create, list, update, and delete sticky notes

✔️ Stored notes managed directly through your server

✔️ Zero manual commands — everything triggered by AI

✔️ Shows real-world tool use + agent execution

---

🧠 How It Works

1️⃣ You define tools in your MCP server

Examples:

• create_note
• get_all_notes
• update_note
• delete_note

Tools specify:

• Name
• Description
• JSON schema
• Handler function

Example tool definition:

{
  "name": "create_note",
  "description": "Create a new sticky note with title and content",
  "input_schema": {...}
}

---

2️⃣ Claude Desktop loads your MCP server

In the claude_desktop_config.json, you point Claude to your MCP server.

Claude now understands your custom tools.

---

3️⃣ User speaks naturally:

“Create a sticky note with 5 points explaining what MCP is.”

Claude interprets it → Plans the action → Calls your MCP tool → Sends JSON payload → Your server executes → Response returned → Claude shows result.

---

4️⃣ Notes stored through your backend

You can store notes:

• In memory
• In a JSON file
• In SQLite / MongoDB
• Or any storage (your choice)

---

🧩 Architecture Diagram

Below is an ASCII diagram showing how everything interacts:

               ┌─────────────────────────────┐
               │         User Query           │
               │  ("Create sticky note...")   │
               └───────────────┬─────────────┘
                               │
                               ▼
                     ┌──────────────────┐
                     │    Claude AI     │
                     │ (MCP Client LLM) │
                     └───────┬──────────┘
                             │  Natural language → Tool plan
                             ▼
                  ┌────────────────────────┐
                  │   Claude Desktop App    │
                  │     (MCP Host)          │
                  └──────────┬─────────────┘
                             │ Loads MCP server
                             ▼
                ┌──────────────────────────────┐
                │        Your MCP Server       │
                │ Registers custom tools:      │
                │  • create_note               │
                │  • list_notes                │
                │  • update_note               │
                │  • delete_note               │
                └───────────┬──────────────────┘
                            │ Executes tool call
                            ▼
                 ┌────────────────────────────┐
                 │       Notes System         │
                 │ (in-memory / JSON / DB)    │
                 └───────────┬────────────────┘
                             │ Returns result
                             ▼
                     ┌──────────────────┐
                     │    Claude AI     │
                     │ Returns response │
                     └──────────────────┘

---

📁 Project Structure (Example)

mcp-sticky-notes/
│
├── server.py                 # MCP server containing all tools
├── notes_manager.py          # In-memory or DB-backed notes logic
├── protocol/                 # MCP protocol handlers
├── requirements.txt
├── README.md
└── claude_desktop_config.json

---

⚙️ Installation & Setup

1️⃣ Clone the project

git clone https://github.com/AshutoshRajGupta/AI-Sticky-Notes-Using-MCP
cd AI-Sticky-Notes-Using-MCP

2️⃣ Install dependencies

pip install -r requirements.txt

3️⃣ Run the MCP server

python server.py

4️⃣ Configure Claude Desktop

Add your MCP server in:

~/Library/Application Support/Claude/claude_desktop_config.json

Example:

{
  "mcpServers": {
    "stickyNotes": {
      "command": "python",
      "args": ["server.py"]
    }
  }
}

5️⃣ Restart Claude Desktop

Claude will now recognize your Sticky Notes tools.

---

💬 Example Interaction

User: “Create a new sticky note explaining what MCP is in 4 points.”

Claude: (Invokes your MCP server tool automatically) **“Created a sticky note titled ‘About MCP’ with 4 bullet points.”*

User: “Show all my notes.”

Claude calls list_notes → Returns your notes.

---

SS Attached

🧪 Tools Implemented

📌 create_note

Create a new note with title & content.

📌 list_notes

Return all notes stored in the system.

📌 update_note

Modify an existing sticky note.

📌 delete_note

Remove a note by its ID or title.

---

🚀 Why This Project Is Powerful

This is not a normal chatbot. It demonstrates:

• Agentic AI
• Tool-calling automation
• Real MCP integration
• External function execution
• AI manipulating structured data
• Full local agent ecosystem

This is exactly the kind of project AI/ML recruiters love.

---