MCP -Unified-Deployment

This MCP server acts as a unified deployment manager for multiple MCP servers. It provides a single interface to start, stop, restart, and monitor MCP servers regardless of their underlying deployment type (uvx, node, source_code, docker, etc.). The core controller is managed via Python scripts, primarily scripts/manage_mcp.py, which reads the mcp_servers.json configuration and orchestrates the configured servers through SSE-based communication when appropriate. You can also generate client-specific configurations using integrate_config_generator.py so that each client can connect to the unified SSE endpoint and control or observe the deployed servers. Typical usage involves starting the unified deployment service, then using the CLI to manage individual servers by name, or querying status for an overview of the ecosystem. The documentation highlights how to configure different server types and how to run the manager in development or production modes.

Prerequisites:

• Python 3.12+
• Git
• uvx (for uvx-based servers) or Node.js/npm (if you plan to manage Node-based servers separately)
• pip (comes with Python)
• Optional: Docker and Docker Compose for containerized deployment

Installation steps:

1. Clone the repository and navigate into it:

git clone https://github.com/BigUncle/MCP-Server-Unified-Deployment.git
cd MCP-Server-Unified-Deployment

2. Set up a Python virtual environment with Python 3.12+:

# Create a virtual environment and activate it
python -m venv venv
# Activate (Linux/macOS)
source venv/bin/activate
# Activate (Windows)
venv\Scripts\activate

3. Install dependencies (as described in the repo):

pip install -r requirements.txt
# Alternatively, install via provided setup script
python scripts/setup_env.py

4. Install mcp-proxy using pipx (recommended):

# Ensure pipx is installed
pip install pipx
pipx ensurepath
# Install mcp-proxy
pipx install mcp-proxy

5. Prepare your configuration by copying the example and editing as needed:

cp config/mcp_servers.example.json config/mcp_servers.json

6. Start the unified deployment manager:

python scripts/manage_mcp.py start

Optional Docker-based route:

• If you prefer Docker, follow the Docker deployment docs in the repository, which explain how to run the unified deployment manager in a containerized environment using pre-built images and docker-compose files. Ensure Docker is installed and configured before proceeding.

Tips and notes:

• The unified deployment tool relies on a central mcp_servers.json file. Each entry should specify the server name, type (uvx, node, source_code, docker, etc.), and the appropriate start command template for that type. The manager then exposes a consistent interface via SSE where applicable.
• Environment variables for individual servers can be specified under the env key per server entry. Use them for things like NODE_ENV, DEBUG, or any service-specific configuration.
• If you modify the configuration, you may need to reload or restart the manager to apply changes. Use the status command to verify the health of all managed servers.
• For local development, using uvx-based servers requires uvx to be installed; for Node-based servers, ensure Node.js and npm are installed. Docker users should reference the dedicated Docker deployment options for production-grade setups.
• The integrate_config_generator.py script can auto-generate client-specific config files under config/client_configs/ based on the mcp_servers.json file; these can be distributed to clients to configure their SSE endpoints.