MCP Todo Server

MCP server developed in Python for basic task management stored locally in a JSON file.

The project implements the main capabilities of the Model Context Protocol (MCP): Resources, Tools, and Prompts, allowing you to query pending tasks, register new tasks, mark tasks as completed, and generate a daily summary template.

Features

The server exposes the following capabilities:

Resource — tasks://pending

Returns the tasks that are pending in tasks.json.

Completed tasks remain stored, but are not included in the Resource result.

Tool — add_task

Allows registering a new task by providing:

• name;

• description;

• priority (alta, media, or baja).

Each task automatically receives an identifier with the format:

T-001
T-002
T-003

The tool also validates that the task has a name and that the priority corresponds to one of the allowed values.

Tool — complete_task

Allows marking a task as completed using its identifier.

For example:

T-001

Upon finding the task, its completada status changes from false to true and the change is saved in tasks.json.

Prompt — daily_summary

Generates a template based on the current state of the tasks.

It considers:

• total tasks;

• completed tasks;

• pending tasks;

• high-priority pending tasks.

The template allows generating a summary of overall progress and focusing attention on the highest-priority activities.

Related MCP server: Tasks MCP Server

Project structure

mcp-todo-server/
│
├── server.py
├── tasks.json
├── requirements.txt
├── README.md
└── .gitignore

The .venv directory corresponds to the local virtual environment and is not included in the repository.

Data model

Tasks are stored in JSON format with the following structure:

{
    "id": "T-001",
    "nombre": "Revisar documento del extraclase",
    "descripcion": "Comprobar que las capturas y explicaciones estén completas",
    "prioridad": "alta",
    "completada": true
}

The id field identifies each task, while completada determines whether it should appear in the pending tasks Resource.

Requirements

To run the project you need:

• Python 3

• pip

• Model Context Protocol SDK for Python

Installation

1. Clone the repository

git clone URL_DEL_REPOSITORIO
cd mcp-todo-server

2. Create the virtual environment

On Windows:

python -m venv .venv

3. Activate the virtual environment

In PowerShell:

.\.venv\Scripts\Activate.ps1

4. Install the dependencies

python -m pip install -r requirements.txt

Execution

With the virtual environment active, run:

python server.py

The server uses communication via standard input and output (stdio) and stays running waiting for an MCP client connection.

To stop it:

Ctrl + C

Testing with MCP Inspector

The server's operation can be verified using MCP Inspector.

Using pnpm:

pnpm dlx @modelcontextprotocol/inspector python server.py

Once the connection is established, the client allows discovering and interacting with the exposed capabilities:

Resources
└── tasks://pending

Tools
├── add_task
└── complete_task

Prompts
└── daily_summary

The tests allow verifying task queries, modifications to tasks.json through the Tools, and the generation of the daily summary template.

Technologies used

• Python

• Model Context Protocol (MCP)

• JSON

• MCP Inspector

• Git

Author

Jendry Linneth Murillo Pérez
Programming IV
II Term, 2026