mcp-manager
CLI tool for managing Model Context Protocol (MCP) servers in one place & using them across them different clients
claude mcp add --transport stdio nstebbins-mcp-manager python -m mcp_manager
How to use
MCP Manager is a Python-based CLI tool that simplifies managing MCP servers across multiple clients (such as Claude Desktop, Cursor, and Claude Code). It lets you search for available MCP servers, view detailed information, and install or uninstall servers for specific clients. The tool orchestrates installation and configuration so you can quickly deploy and manage the servers you need without manual setup on each client. Available server endpoints include Playwright (browser automation), Filesystem (local file access), Fetch (HTTP requests), Git, GitHub, and Memory (Claude's memory management), with client-specific installation options.
To use MCP Manager, first install the package, then invoke its commands to discover and manage servers. For example, you can search for servers by keyword, inspect a server's details, install a server for Claude Desktop or Cursor, and then manage or uninstall as needed. The CLI also provides commands to show or set the path to client configuration files, making it easier to route each server to the correct client environment.
How to install
Prerequisites:
- Python 3.9 or higher
- pip (comes with Python)
Install MCP Manager from PyPI:
pip install mcp-manager
Basic usage after installation (examples refer to the default CLI entry point):
mcp-manager search <keyword> # Find available MCP servers matching a keyword
mcp-manager info <server-name> # Show detailed information about a server
mcp-manager install <server-name> --client=claude-desktop|cursor|claude-code # Install for a specific client
mcp-manager list # List installed servers
mcp-manager uninstall <server-name> --client=claude-desktop|cursor|claude-code # Uninstall
mcp-manager config path [--client=...] # Show current client config path
mcp-manager config set-path <new-path> [--client=...] # Set new client config path
If you plan to contribute or develop locally, you can clone the repository and install development dependencies:
# Clone the repository
git clone https://github.com/nstebbins/mcp-manager.git
cd mcp-manager
# Install dependencies and development tools
poetry install
pre-commit install
For running tests and code quality checks:
poetry run pytest
poetry run ruff format .
poetry run ruff check .
Additional notes
Notes:
- The MCP Manager supports multiple clients (Claude Desktop, Cursor, Claude Code); when installing, specify the target client with --client to ensure proper configuration.
- Some servers may require additional dependencies or runtime environments (e.g., Node.js for certain servers, Docker for others). Ensure those prerequisites are installed if you plan to use those servers.
- The tool includes commands to locate the client config file path and set a new path, which helps keep server configurations organized per client.
- If you encounter permission or environment issues, consider using a virtual environment (venv) or running with appropriate permissions. Ensure your Python environment matches the supported version (Python 3.9+).
- Regularly update MCP Manager to benefit from new servers and bug fixes, and review available servers using the search and info commands before installation to understand dependencies and client compatibility.
Related MCP Servers
LLaMa -Streamlit
AI assistant built with Streamlit, NVIDIA NIM (LLaMa 3.3:70B) / Ollama, and Model Control Protocol (MCP).
mcp-playground
A Streamlit-based chat app for LLMs with plug-and-play tool support via Model Context Protocol (MCP), powered by LangChain, LangGraph, and Docker.
dynamic-shell
Dynamic Shell Command MCP Server
mcp-aoai-web-browsing
A minimal Model Context Protocol 🖥️ server/client🧑💻with Azure OpenAI and 🌐 web browser control via Playwright.
mcp-community
Easily run, deploy, and connect to MCP servers
oauth-callback
Lightweight OAuth 2.0 authorization code capture for CLI tools & desktop apps. Works with Node.js, Deno, Bun. MCP SDK ready.