Subhokkhon/AI-Sticky-Notes-Custom-MCP-server

🧠 AI Sticky Notes – Custom MCP Server

This repository contains a custom MCP (Model Context Protocol) server built using fastmcp — a lightweight Python library for quickly creating MCP servers.

The server, called “AI Sticky Notes”, allows AI models (like Claude or other MCP-compatible clients**) to:

Add new notes

Read existing notes

Retrieve the latest note

Generate summaries of all notes

All notes are stored persistently in a local file called notes.txt.

🗂️ Repository Structure AI-Sticky-Notes/ │ ├── main.py # Main MCP server logic ├── notes.txt # File where all notes are stored └── README.md # Documentation (this file)

⚙️ Features 📝 1. Add a Note

Adds a new note (string) to notes.txt.

Function: add_note(message: str)

Returns: "Note saved!"

📖 2. Read All Notes

Reads and returns all stored notes.

Function: read_notes()

Returns: A concatenated string of all notes, or "No notes yet." if empty.

📌 3. Get the Latest Note

Retrieves the most recently added note.

Resource URI: notes://latest

Returns: The latest note or "No notes yet." if no notes exist.

🧾 4. Generate a Summary Prompt

Creates a natural language prompt for summarizing all notes using the AI client.

Function: note_summary_prompt()

Returns:

"Summarize the current notes: [content]"

"There are no notes yet." if empty

🧩 How It Works

The server uses the FastMCP framework to define a Model Context Protocol interface.

It exposes tools and resources that can be used by any AI model (like Claude) that supports MCP.

When the model calls a function like add_note or read_notes, FastMCP handles the interaction and updates notes.txt accordingly

If your MCP-compatible client (like Claude Desktop) is set up properly, it will detect the server and display it as “AI Sticky Notes”.

Restart Claude.

You can now use commands like:

“Add a note saying ‘Finish project report by Monday.’”

“Show me all my notes.”

“Summarize my notes.”

“What was the last note I added?”

🧠 Example Interactions with Claude

You: “Add a note saying ‘Schedule meeting with team.’” Claude: “Note saved!”

You: “Show me all notes.” Claude:

1. Finish project report by Monday.
2. Schedule meeting with team.

You: “Summarize my notes.” Claude: “You have two tasks: finalize a project report and arrange a team meeting.”

🪶 File Details main.py

Implements:

MCP server initialization

Tools (@mcp.tool)

Resource (@mcp.resource)

Prompt (@mcp.prompt)

Ensures that the notes.txt file exists using the helper function ensure_file().

notes.txt

Plain-text file used to store all notes. Each note is stored on a separate line.

🧰 Technologies Used

Python 3.8+

FastMCP — for building the MCP server

Model Context Protocol (MCP) — for communication with AI models

Claude AI — as an example MCP client