miniMCP

A lightweight project that demonstrates how to implement and connect a local AI model (using Ollama) with custom tools through the Model Context Protocol (MCP).

In this project, the AI acts as a database assistant that can query and manage an in-memory database using CRUD (Create, Read, Update, Delete) operations.

Project Structure

• server.py: This is the MCP server created with FastMCP. It maintains an in-memory database and exposes tools for:

  • Creating items (create_item)

  • Reading individual items (read_item)

  • Reading multiple items at once (read_multiple_items)

  • Updating items (update_item)

  • Deleting items (delete_item)

• client.py: This is the client that interacts via the terminal. It connects to the MCP server through standard input/output streams (stdio), reads the available tools, and establishes a chat loop with the user and the Ollama LLM (gemma4:e4b).

• requirements.txt: File with the project's main dependencies.

Related MCP server: Modular MCP Server with Python Tools

Requirements and Dependencies

Before starting the project, make sure you have installed:

1. Python 3.10+

2. Ollama installed and running on your machine.

3. The model used by this repository, which by default is gemma4:e4b. You can download it with the command:

   ollama run gemma4:e4b

Installation

1. Clone the repository and navigate to its folder:

   git clone <url-del-repo> miniMCP
   cd miniMCP

2. Create and activate a virtual environment (optional but highly recommended):

   python -m venv .venv
   # En Windows:
   .\.venv\Scripts\activate
   # En macOS/Linux:
   source .venv/bin/activate

3. Install the necessary dependencies:

   pip install -r requirements.txt
   pip install gradio

Usage

The project consists of a Client/Server configuration, but the client is responsible for launching the server itself, so you only need to run the client:

python client.py

Type your questions directly into the console.

Interaction Example

You can try telling the AI:

• "Create an item with ID 'task1' and content 'Buy milk'"

• "Show the content of 'task1'"

• "Update 'task1' and tell me the milk should be skimmed"

• "Read items 'task1' and 'task2'"

• "Delete ID 'task1'"

The AI will automatically execute the server.py tools without you having to write the code or calls directly.