Discover Awesome MCP Servers

MCP Bridge API

O MCP Bridge é um proxy leve, rápido e independente de LLM para conectar-se a múltiplos servidores Model Context Protocol (MCP) através de uma API REST unificada. Ele permite a execução segura de ferramentas em diversos ambientes, como dispositivos móveis, web e edge. Projetado para flexibilidade, escalabilidade e fácil integração com qualquer backend de LLM.

Redshift MCP Server

A Model Context Protocol server that enables AI assistants to interact with Amazon Redshift databases, allowing for schema exploration, query execution, and statistics collection.

ITSM Integration Platform

MCP para integração de ferramenta ITSM

Macroforge MCP Server

Enables AI assistants to access Macroforge documentation, validate TypeScript code with @derive decorators, expand macros, and retrieve macro documentation.

OpenFeature MCP Server

Provides OpenFeature SDK installation guidance through MCP tool calls. Enables AI clients to fetch installation prompts and setup instructions for various OpenFeature SDKs across different programming languages and frameworks.

RobotFrameworkLibrary-to-MCP

Okay, here's a breakdown of how to turn a Robot Framework library into an MCP (MessagePack-RPC) server, along with explanations and considerations: **Understanding the Goal** The core idea is to expose the keywords of your Robot Framework library as remote procedures that can be called over a network using MessagePack-RPC. This allows you to run your Robot Framework tests and libraries in a distributed manner, potentially offloading resource-intensive tasks to dedicated servers. **Key Components and Technologies** 1. **Robot Framework Library:** This is your existing library containing the keywords you want to expose. 2. **MessagePack-RPC (MCP):** A binary serialization and RPC protocol. MessagePack is efficient for data transfer, and RPC allows you to call functions on a remote server as if they were local. 3. **Python (or other language):** You'll need a server-side implementation (likely in Python) to host your Robot Framework library and handle the MCP requests. 4. **MCP Server Framework (e.g., `mcp` Python package):** A library that simplifies the creation of MCP servers. 5. **MCP Client (in Robot Framework):** A Robot Framework library that acts as a client, making calls to the MCP server. **Steps to Implementation** **1. Install Necessary Packages** * On the server (where your Robot Framework library will run): ```bash pip install robotframework pip install mcp ``` * On the client (where your Robot Framework tests will run): ```bash pip install robotframework pip install mcp ``` **2. Create the MCP Server (Python)** ```python # server.py import mcp import robot.libraries # Import the robot.libraries module from robot.api.deco import keyword from robot.libraries.BuiltIn import BuiltIn # Replace 'YourLibrary' with the actual name of your Robot Framework library # and the path to it. If it's in the same directory, you can use a relative path. # Example: # from my_robot_library import MyRobotLibrary # library = MyRobotLibrary() # Dynamically load the library library_name = "YourLibrary" # Replace with your library's name library_path = "./YourLibrary.py" # Replace with the path to your library file try: # Attempt to import the library dynamically spec = importlib.util.spec_from_file_location(library_name, library_path) module = importlib.util.module_from_spec(spec) spec.loader.exec_module(module) library_class = getattr(module, library_name) # Assuming the class name matches the library name library = library_class() except Exception as e: print(f"Error loading library: {e}") exit(1) class RobotLibraryService: def __init__(self, library): self.library = library def __dir__(self): # Expose only the keywords as callable methods return [name for name in dir(self.library) if callable(getattr(self.library, name)) and not name.startswith('_')] def __getattr__(self, name): # Delegate calls to the Robot Framework library's methods (keywords) try: attr = getattr(self.library, name) if callable(attr): return attr else: raise AttributeError(f"'{type(self.library).__name__}' object has no attribute '{name}'") except AttributeError: raise AttributeError(f"'{type(self.library).__name__}' object has no attribute '{name}'") # Create an instance of the service, passing in your Robot Framework library service = RobotLibraryService(library) # Create the MCP server server = mcp.Server(service) if __name__ == '__main__': server.run(('localhost', 6000)) # Listen on localhost, port 6000 (or choose a different port) print("MCP Server started on localhost:6000") ``` **Explanation of `server.py`:** * **Imports:** Imports the necessary libraries (`mcp`, `robot.libraries`, `robot.api.deco`). * **Library Loading:** This is the crucial part. It dynamically loads your Robot Framework library. You'll need to replace `"YourLibrary"` and `"./YourLibrary.py"` with the actual name and path of your library file. The code attempts to import the library dynamically using `importlib`. This is more flexible than a direct `from ... import ...` statement because it allows you to specify the path to the library file. It then retrieves the class representing your library from the loaded module. * **`RobotLibraryService` Class:** This class acts as a wrapper around your Robot Framework library. It's essential for exposing the library's keywords as callable methods to the MCP server. * `__dir__`: This method is important for introspection. It tells the MCP server which methods (keywords) are available for remote calling. It filters the attributes of your library to only include callable methods (functions) that don't start with an underscore (to avoid exposing internal methods). * `__getattr__`: This is the magic. When the MCP server receives a request to call a method (keyword) that doesn't directly exist in the `RobotLibraryService` class, Python calls `__getattr__`. This method then looks up the method in your *actual* Robot Framework library and returns it. This effectively delegates the call to your library. * **Server Creation:** Creates an instance of the `RobotLibraryService`, passing in your Robot Framework library. Then, it creates an `mcp.Server` instance, passing in the service object. * **`server.run()`:** Starts the MCP server, listening on the specified address and port. **3. Create the Robot Framework Client Library** ```python # McpClientLibrary.py import mcp from robot.api.deco import keyword class McpClientLibrary: def __init__(self, host='localhost', port=6000): self.host = host self.port = port self.client = None def connect_to_server(self): self.client = mcp.Client((self.host, self.port)) def disconnect_from_server(self): if self.client: self.client.close() self.client = None @keyword def call_remote_keyword(self, keyword_name, *args): """Calls a keyword on the remote MCP server.""" if not self.client: raise Exception("Not connected to the MCP server. Call 'Connect To Server' first.") try: result = self.client.call(keyword_name, *args) return result except Exception as e: raise Exception(f"Error calling remote keyword '{keyword_name}': {e}") ``` **Explanation of `McpClientLibrary.py`:** * **Imports:** Imports `mcp` and `robot.api.deco`. * **`McpClientLibrary` Class:** * `__init__`: Initializes the client with the server's host and port. * `connect_to_server`: Creates an `mcp.Client` instance to connect to the server. * `disconnect_from_server`: Closes the connection to the server. * `call_remote_keyword`: This is the key keyword. It takes the name of the keyword you want to call on the server and any arguments. It uses `self.client.call()` to make the RPC call. It handles potential exceptions and re-raises them with more informative messages. The `@keyword` decorator makes this method available as a Robot Framework keyword. **4. Robot Framework Test Case** ```robotframework ***Settings*** Library McpClientLibrary ***Variables*** ${SERVER_HOST} localhost ${SERVER_PORT} 6000 ***Test Cases*** Call Remote Keyword Connect To Server ${result} = Call Remote Keyword your_keyword_name arg1 arg2 # Replace with your keyword and arguments Log Result: ${result} Disconnect From Server ``` **Explanation of the Robot Framework Test Case:** * **`Library McpClientLibrary`:** Imports the client library you created. * **`Connect To Server`:** Calls the `connect_to_server` keyword to establish a connection to the MCP server. * **`Call Remote Keyword`:** Calls the `call_remote_keyword` keyword, passing in the name of the keyword you want to execute on the server (e.g., `"your_keyword_name"`) and any arguments that the keyword expects. The result of the remote keyword execution is stored in the `${result}` variable. * **`Log Result: ${result}`:** Logs the result to the Robot Framework report. * **`Disconnect From Server`:** Closes the connection to the server. **5. Running the Code** 1. **Start the MCP Server:** Run `python server.py` on the server machine. Make sure the server is running *before* you run the Robot Framework test. 2. **Run the Robot Framework Test:** Execute your Robot Framework test case. **Important Considerations and Improvements** * **Error Handling:** The example code includes basic error handling, but you should add more robust error handling to catch potential network issues, exceptions in your Robot Framework library, and other problems. * **Security:** MCP itself doesn't provide built-in security features like encryption or authentication. If you're transmitting sensitive data, you'll need to add security measures, such as using TLS/SSL to encrypt the communication channel or implementing authentication mechanisms. Consider using a more secure RPC framework if security is critical. * **Data Serialization:** MessagePack is generally efficient, but be mindful of the types of data you're passing between the client and server. Complex objects might require custom serialization/deserialization. * **Library Loading:** The dynamic library loading is more flexible, but it relies on the library being structured in a way that allows it to be loaded dynamically. If your library has complex dependencies or initialization logic, you might need to adjust the loading code accordingly. * **Asynchronous Operations:** For long-running tasks, consider using asynchronous operations (e.g., using `asyncio` in Python) to prevent the MCP server from blocking while waiting for the task to complete. * **Alternative RPC Frameworks:** While MCP is a good choice for its simplicity and efficiency, other RPC frameworks like gRPC or Thrift might be more suitable for complex applications with strict performance or security requirements. * **Robot Framework's `Remote` Library:** Robot Framework has a built-in `Remote` library that uses XML-RPC. While it's simpler to set up initially, it's generally less efficient than MCP. The `Remote` library is a good starting point for simple remote execution scenarios. **Example: A Simple Robot Framework Library** ```python # YourLibrary.py from robot.api.deco import keyword class YourLibrary: @keyword def add_numbers(self, a, b): """Adds two numbers and returns the result.""" a = int(a) b = int(b) return a + b @keyword def say_hello(self, name): """Returns a greeting.""" return f"Hello, {name}!" ``` In this case, you would replace `"YourLibrary"` in `server.py` with `"YourLibrary"` and `"./YourLibrary.py"` with `"./YourLibrary.py"`. Then, in your Robot Framework test case, you could call the `add_numbers` or `say_hello` keywords remotely. **Summary** Turning a Robot Framework library into an MCP server involves creating a server-side component that hosts the library and exposes its keywords as remote procedures, and a client-side library that allows Robot Framework tests to call those procedures. The `mcp` Python package simplifies the creation of MCP servers and clients. Remember to handle errors, consider security, and choose the right RPC framework for your specific needs.

REMnux Documentation MCP Server

Enables AI assistants to search and retrieve information from the REMnux malware analysis toolkit documentation, allowing users to ask questions about reverse-engineering tools and get accurate, up-to-date answers from the live documentation.

Fusion 360 MCP Integration

Enables AI assistants to interact programmatically with Autodesk Fusion 360 for creating parametric 3D models through simple API calls.

A2AMCP

A Redis-backed MCP server that enables multiple AI agents to communicate, coordinate, and collaborate while working on parallel development tasks, preventing conflicts in shared codebases.

Anki MCP Server

Enables interaction with the Anki desktop app for spaced repetition learning. Supports reviewing due cards, creating new flashcards, and managing study sessions through natural language commands.

Google Chat MCP Server

Enables posting text messages to Google Chat spaces through webhook-based integration, providing simple and secure message delivery without OAuth setup requirements.

grasp

Self-hosted Browser Using Agent with built-in MCP, A2A support.

w3c-mcp

MCP Server for accessing W3C/WHATWG/IETF web specifications. Provides AI assistants with access to official web standards data including specifications, WebIDL definitions, CSS properties, and HTML elements.

Simple MCP POC

A proof-of-concept MCP server that enables reading local files and performing basic arithmetic operations. It provides a simple foundation for understanding how tools are exposed to MCP clients.

Codebase MCP Server

Enables AI assistants to semantically search and understand code repositories using PostgreSQL with pgvector embeddings. Provides repository indexing, natural language code search, and development task management with git integration.

GitHub Configuration

Um servidor de Protocolo de Contexto de Modelo (MCP) para o aplicativo de gerenciamento de tarefas TickTick.

Knowledge Tools Mcp

Um servidor MCP para interagir com vários aplicativos de pesquisa e anotações.

MCP-Foundry

MCP Foundry

makefilemcpserver

An MCP server that exposes Makefile targets as callable tools for AI assistants, allowing Claude and similar models to execute Make commands with provided arguments.

PDF MCP Flow

Enables AI-driven PDF document processing including PDF to Markdown conversion, intelligent text and table extraction, image extraction, format conversion between PDF/Word/Markdown, batch processing, and fuzzy search - optimized for LLM context and RAG workflows.

MCP TS Toolkit

A comprehensive MCP server providing secure tools for filesystem operations, Git management, web search, document conversion, npm/.NET project management, and AI generative capabilities (image/video/audio generation and processing) via PiAPI.ai integration.

Pilldoc User MCP

Provides access to pharmaceutical management system APIs including user authentication, pharmacy account management, and advertising campaign controls. Enables querying pharmacy information, updating account details, and managing ad blocking through natural language interactions.

HubAPI Auth MCP Server

An MCP server for interacting with HubSpot's authentication API, enabling secure authentication and authorization operations through natural language.

Cross-System Agent Communication MCP Server

Espelho de

ThinkDrop Vision Service

Provides screen capture, OCR text extraction, and visual language model scene understanding capabilities with continuous monitoring and automatic memory storage integration.

FSS Pension MCP Server

Provides FSS (Korean Financial Supervisory Service) pension data to AI models like Claude through Model Context Protocol, enabling real-time pension product information retrieval and personalized AI-based pension consultation services.

Channel Talk MCP Server

An unofficial MCP server that integrates with the Channel Talk Open API to allow AI assistants to access and manage chat information. It enables users to retrieve chat lists and message histories filtered by status, tags, and customer details.

Jinko

Jinko

Kali MCP Server

Provides a containerized Kali Linux environment that gives AI assistants access to a comprehensive suite of security and penetration testing tools. It enables automated vulnerability scanning, network reconnaissance, and secure command execution through the Model Context Protocol.