Python MCP Servers: Complete Catalog from awesome-mcp-servers

The awesome-mcp-servers repository maintains a curated index of 28 MCP (Model-Centered Programming) servers implemented in Python, each identified by the 🐍 emoji in the README.

The punkpeye/awesome-mcp-servers repository serves as the definitive community registry for MCP server implementations. Within this catalog, the 🐍 emoji functions as the authoritative marker for Python MCP servers, indicating that the codebase is written primarily in Python or provides first-class Python SDK support.

How Python Servers Are Identified in the Repository

The repository uses a simple emoji-based taxonomy in README.md to denote implementation languages. Every entry bearing the 🐍 (snake) symbol represents a Python-based MCP server. These entries contain direct links to their respective GitHub repositories, with line-specific anchors pointing to the exact location in the README where each server is documented (e.g., line 171 for depwire/depwire, line 576 for pythonanywhere-mcp-server).

Complete List of Python MCP Servers

The following catalog organizes the 28 Python MCP servers by functional category, preserving the exact line references from README.md in the source repository.

Development and Code Analysis Tools

These servers provide static analysis, dependency management, and code intelligence capabilities.

• depwire/depwire (README.md line 171) — Generates dependency graphs and exposes 15 MCP tools for AI coding assistants. Parses Python alongside TypeScript, JavaScript, Go, Rust, and C.
• xray by srijanshukla18 (README.md line 1450) — Progressive code-intelligence MCP server supporting Python, JavaScript/TypeScript, and Go.
• Diffcontext by trakshan-mishra (README.md line 1139) — Static-analysis-powered context compiler specifically focused on Python LLM agents.
• reuse-before-generate by aradar46 (README.md line 1102) — Python-enabled discovery engine that finds existing alternatives before scaffolding new projects.
• bumpguard-mcp by appcreationsca (README.md line 1104) — Performs pre-flight dependency-upgrade impact analysis using pure Python (with optional .NET support).

Cloud Platforms and Infrastructure

Servers that interface directly with cloud hosting environments and Linux administration.

• pythonanywhere-mcp-server (README.md line 576) — Official MCP server implementation for the PythonAnywhere cloud platform.
• mcp-linux-tools by gerard-kanters (README.md line 726) — Administers Linux servers from MCP clients, providing Python-based tools for file, cron, and service management.
• OpenAaaS by Wolido (README.md line 236) — Python adapter connecting Claude, Cursor, and Cline to the OpenAaaS scientific-agent network.

Secure Execution and Sandboxing

These implementations focus on running untrusted or isolated Python code safely.

• PRIMS by hileamlakB (README.md line 625) — Python Runtime Interpreter MCP Server that executes user-submitted code in isolated environments.
• capsule → mcp-server by mavdol (README.md line 627) — Executes untrusted Python and JavaScript code inside WebAssembly sandboxes.
• pydantic-ai → mcp-run-python (README.md line 631) — Secure sandbox for executing Python code via MCP tool calls, maintained by the Pydantic team.
• mcp_safe_local_python_executor by maxim-saplin (README.md line 680) — Safe Python interpreter based on Hugging Face Smolagents' LocalPythonExecutor.
• text_file_read_and_refactor_mcp by FI-Mihej (README.md line 686) — Token-efficient Python stdio MCP server offering safe text-file search, read, and refactoring tools.

Database and Storage Connectivity

• mcp-sqlalchemy-server by OpenLinkSoftware (README.md line 1001) — Generic DBMS connectivity via SQLAlchemy, utilizing Python and pyodbc drivers.
• memvid-mcp-server by ferrants (README.md line 957) — Small Python HTTP server for local interaction with the memvid storage and semantic search system.

Automation and Browser Control

• selenium-mcp by seleniumboot (README.md line 486) — Comprehensive Python MCP server for Selenium WebDriver, exposing 84 tools for browser automation, code generation, and CI pipelines.
• x402-discovery-mcp by rplryan (README.md line 237) — Runtime discovery layer for x402-payable APIs, including a Python SDK and CLI.

Frameworks and Developer SDKs

High-level tools for building new MCP servers in Python.

• FastMCP (README.md line 3909) — High-level framework specifically designed for building MCP servers in Python. Also referenced as jlowin/fastmcp at the same line.
• mcp-server-boilerplate by shellsage-ai (README.md line 1433) — Production-ready starter templates for MCP servers in Python and TypeScript.
• ticktick-sdk by dev-mirzabicer (README.md line 3756) — Comprehensive async Python SDK for TickTick with built-in MCP server support.

AI Safety and Compliance

• air-blackbox-mcp by air-blackbox (README.md line 3219) — EU AI Act compliance scanner specifically designed for Python AI agents.
• aimarket-mcp by alexar76 (README.md line 3194) — Python-supported MCP server for web-gateway management, API verification, and SSRF hardening.
• agent-trust-stack-mcp by alexfleetcommander (README.md line 3197) — Cryptographic provenance and reputation scoring system shipping with a Python SDK.

Multimedia and Specialized Tools

• kinocut by KyaniteLabs (README.md line 2846) — Guard-railed video editing MCP server offering both a Python client and CLI interface.
• openai-websearch-mcp by ConEchoAI (README.md line 3014) — Python-based MCP server that exposes OpenAI's web_search capability as a built-in tool.
• python-docs-mcp-server by ayhammouda (README.md line 1202) — Local read-only retrieval system for the official Python standard-library documentation.

Multi-Language Servers with Python Support

• mcp-server-js by yepcode (README.md line 638) — While primarily a JavaScript server, this implementation fully supports Python tool execution via NPM/PyPI integration.

Installation and Usage Examples

Installing Python MCP Servers via pip

Most Python MCP servers distribute through PyPI. Install them using standard pip commands:

pip install bumpguard-mcp
pip install pythonanywhere-mcp-server
pip install fastmcp

Running a Standalone Server

After installation, start the server with command-line arguments specifying host and port. The following example uses bumpguard-mcp:

# Start the MCP server (defaults to 127.0.0.1:5000)

bumpguard-mcp --host 127.0.0.1 --port 5000

Calling MCP Tools from Python Clients

Interact with running MCP servers using standard HTTP requests from Python applications:

import json
import requests

# Example: ask bumpguard-mcp to analyze a dependency upgrade

payload = {
    "tool": "bumpguard.analyse",
    "arguments": {
        "repo": "github.com/owner/repo",
        "dependency": "requests",
        "new_version": "2.32.0"
    }
}

resp = requests.post("http://localhost:5000/tools/call", json=payload)
print(json.dumps(resp.json(), indent=2))

Building Custom Servers with FastMCP

Use the FastMCP framework (referenced at line 3909 in README.md) to create production-ready MCP servers with minimal boilerplate:

# fastmcp_example.py

from fastmcp import MCPServer, tool

app = MCPServer()

@tool
def echo(message: str) -> str:
    """Return the same message back."""
    return message

if __name__ == "__main__":
    app.run(host="0.0.0.0", port=8000)

Execute the server:

python fastmcp_example.py

The server now exposes an MCP endpoint at http://localhost:8000/tools/call, compatible with any MCP-compliant client.

Summary

• The awesome-mcp-servers repository uses the 🐍 emoji in README.md to identify Python implementations, with 28 confirmed Python MCP servers catalogued as of the current main branch.
• Python MCP servers cover diverse domains including cloud infrastructure (PythonAnywhere), secure code execution (PRIMS, pydantic-ai), browser automation (selenium-mcp), and developer frameworks (FastMCP).
• Each entry in the repository links to specific line numbers (ranging from line 171 to line 3909) for precise verification of the Python implementation status.
• Most Python MCP servers install via pip and expose HTTP or stdio interfaces that conform to the MCP protocol specification.
• The FastMCP framework provides the standard high-level abstraction for building new Python MCP servers with decorator-based tool definitions.

Frequently Asked Questions

How can I verify that an MCP server is implemented in Python?

Look for the 🐍 emoji next to the repository name in the punkpeye/awesome-mcp-servers README. The repository maintains strict tagging conventions where this emoji exclusively denotes Python-based implementations. Each entry links to a specific line number in README.md (such as line 576 for pythonanywhere-mcp-server) for direct verification.

What is the recommended framework for building new Python MCP servers?

FastMCP is the community-standard high-level framework for Python MCP server development, referenced at line 3909 in the repository. It provides decorator-based syntax for defining tools and handles the underlying protocol negotiation, allowing developers to expose Python functions as MCP tools with minimal boilerplate code.

Are these Python MCP servers compatible with Claude Desktop and other MCP clients?

Yes. All Python MCP servers listed in the repository implement the standardized Model-Centered Programming protocol. Whether using the Selenium WebDriver server (line 486) or the SQLAlchemy database connector (line 1001), these implementations communicate via the standard MCP transport layer (stdio or HTTP), making them compatible with Claude Desktop, Cursor, Cline, and other MCP-compliant clients.

How do I install and run a Python MCP server locally?

Most Python MCP servers distribute through PyPI. Execute pip install <server-name> to install, then invoke the server binary or Python module to start the service. For example, bumpguard-mcp (line 1104) starts via command-line interface with configurable host and port parameters, while framework-based servers like those built with FastMCP run as standard Python scripts exposing HTTP endpoints at configurable addresses.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}