mcp-repl 0.1.1

Add your description here

MCP-REPL

A lightweight REPL (Read-Eval-Print Loop) for interacting with various services through MCP (Model Control Protocol).

Overview

mcp-repl is designed for efficient development, debugging, and testing of MCP servers. It provides an intuitive command-line interface that's simpler than using the Cloud Desktop app, allowing developers to quickly:

• Send queries to MCP servers
• View detailed tool execution
• Automatically save chat history
• Test multiple MCP services simultaneously

Installation

Install via pip or uv:

uv add mcp-repl

Development Installation

Clone and install in editable mode:

git clone https://github.com/yourusername/mcp-repl.git
cd mcp-repl
uv venv

Databases Example

This example demonstrates how you can:

• Set up 3 databases (PostgreSQL, MySQL, Redis) in Kubernetes with mock data
• Run MCP servers for each dataset
• Interact with all databases from a single REPL

Key Features

• Natural language queries across multiple databases
• Comparison of data structures across systems
• Execution of complex operations seamlessly

Running the Example

Setup infrastructure (requires kind and helm):

bash examples/databases/setup.sh

Generate mock data:

python examples/databases/generate_mock_data.py

Start the REPL:

python -m src.mcp_repl.repl --config examples/databases/config.json --auto-approve-tools

Sample Queries

You can perform queries like:

• "Find all tables in PostgreSQL and MySQL"
• "Compare the structure of the 'users' table in PostgreSQL with the 'customers' table in MySQL"
• "Count the number of records in each database"

Usage

Start the REPL:

python -m src.mcp_repl.repl --config path/to/config.json

Optional Flags

• --auto-approve-tools: Automatically approve all tool executions
• --always-show-full-output: Always display complete tool outputs
• --chat-history-dir PATH: Directory to save chat history (default: ./chat_history)

Testing

Comprehensive integration tests are included to verify functionality:

Run all tests:

pytest

Run specific integration tests:

pytest ./test/integration/