mcp-studies

Open studio on MCP servers - experimental studies and prototypes using the Model Context Protocol (MCP) Python SDK. This repository provides a reproducible environment for testing, development, and integration of MCP features.

Project Setup and Structure

Environment Setup

To create a clean Python environment for this project, we use uv for fast dependency management and virtual environment creation.

Steps:

1. Create a virtual environment:

   uv venv

   This will create a .venv directory with an isolated Python environment.

2. Install dependencies:

   make env

   This command (defined in the Makefile) installs all required packages into the .venv environment.

3. Activate the environment:

   source .venv/bin/activate

   This sets your shell to use the project's Python and installed packages.

The spikes Folder

The spikes directory contains experimental code, demos, and proof-of-concept scripts. Each subfolder or script is a "spike"—a focused experiment or prototype to explore a specific feature, integration, or idea. Spikes are not production code; they are intended for learning, rapid iteration, and sharing findings.

Environment Variables

Use the included env.sh script to automatically generate a .env file with necessary API keys from 1Password:

./env.sh          # Silent mode
./env.sh -v       # Verbose mode
source .env       # Load environment variables

Configure your credentials in .env_vars file following the format:

ENV_VAR_NAME:1password_item_name:field_name