Discover Awesome MCP Servers

mcp-painter

Ferramenta de Desenho para Assistentes de IA

Tushare MCP Server

Um servidor baseado no Protocolo de Contexto de Modelo (Model Context Protocol) que permite que assistentes de IA consultem e pesquisem informações sobre ações usando a API Tushare.

Multi-Database MCP Server

A plugin-based MCP server that enables interaction with multiple databases (MySQL, PostgreSQL, MongoDB, SQL Server, Redis) through a unified interface with configurable read/write permissions and independent plugin architecture.

MCP Server

Um proxy de serviço de IA moderno que permite a interação com múltiplos provedores de IA (Anthropic Claude, OpenAI) através de uma API unificada, implantado globalmente usando Cloudflare Workers.

The Graph Token API MCP

Official MCP server that turns The Graph’s Token API into a plug-and-play web3 data tool. Exposes ERC-20 & NFT metadata, balances, transfers, top-holder stats, prices, and more, allowing LLMs to run SQL queries on structured and indexed blockchain data.

PokeAPI MCP Server

A Model Context Protocol server that interfaces with PokeAPI to provide Pokémon information to LLM applications through JSON-RPC over stdio.

Browserbase MCP Server

Enables cloud browser automation through Browserbase and Stagehand, allowing LLMs to interact with web pages, take screenshots, extract data, and perform automated actions with support for proxies, stealth mode, and parallel sessions.

JSON Editor MCP

Enables efficient JSON file editing with targeted read, write, delete, and deep merge operations using dot notation paths, optimized for managing multilingual projects and large configuration files.

WeChat Publisher MCP

A Model Context Protocol server that enables AI tools to automatically publish articles to WeChat Official Accounts, supporting Markdown-to-HTML conversion, image handling, and both preview and official publishing modes.

MCP Log Reader

Um servidor MCP especializado que ajuda a analisar e depurar logs do Protocolo de Contexto de Modelo, fornecendo ao Claude acesso direto a arquivos de log em múltiplas plataformas.

MCP Server POC

A proof-of-concept MCP server demonstrating various capabilities including mathematical calculations, URL fetching, system information retrieval, data processing, and file operations.

GoScry

GoScry é uma aplicação de servidor escrita em Go que atua como uma ponte entre um sistema de controle (como um LLM ou script) e um navegador web.

Run Command MCP Server

Provides tools for executing shell commands both synchronously and asynchronously with real-time output streaming and process management capabilities. It enables users to start background tasks, monitor progress, and manage long-running processes via Stdio or HTTP transports.

MCP Server Demo

A WebSocket-based Model Control Protocol (MCP) server that processes model requests and provides responses. Supports chat and text completion actions with a standardized JSON protocol for AI model communication.

BatchIt

A simple aggregator server that allows batching multiple MCP tool calls into a single request, reducing token usage and network overhead for AI agents.

MCP Config

Uma ferramenta CLI para gerenciar configurações de servidores MCP.

mcp-pandoc-ts: A Document Conversion MCP Server (TypeScript/Host Service Version)

Here's a breakdown of how to create an MCP (Message Control Protocol) server that allows a Docker container to control a Pandoc installation on the host machine, along with code snippets and explanations: **Concept** The core idea is to have a simple server running on the host machine that listens for commands from the Docker container. These commands will be Pandoc commands. The server then executes Pandoc locally and returns the output (or errors) back to the Docker container. MCP is a lightweight protocol that can be used for this purpose. **Components** 1. **Host Service (Pandoc Controller):** A Python script (or similar) running on the host machine. This script: * Listens on a specific port for incoming MCP messages. * Parses the MCP message to extract the Pandoc command. * Executes the Pandoc command using `subprocess`. * Captures the output (stdout and stderr) of the Pandoc command. * Formats the output into an MCP response and sends it back to the Docker container. 2. **Docker Container:** Your application running inside the Docker container. This application: * Constructs an MCP message containing the Pandoc command. * Sends the MCP message to the host service (using the host's IP address and the port the service is listening on). * Receives the MCP response from the host service. * Parses the MCP response to extract the Pandoc output (or error). 3. **MCP Protocol:** A simple protocol for communication. For example: * **Request:** `COMMAND:pandoc -s input.md -o output.pdf` * **Response (Success):** `STATUS:OK\nOUTPUT: [PDF data as base64 or a path to a file]` * **Response (Error):** `STATUS:ERROR\nOUTPUT: [Error message]` **Python Host Service (pandoc_controller.py)** ```python import socket import subprocess import base64 import os HOST = '0.0.0.0' # Listen on all interfaces PORT = 65432 # Port to listen on def execute_pandoc(command): """Executes the Pandoc command and returns the output.""" try: process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE) stdout, stderr = process.communicate() return_code = process.returncode if return_code == 0: return "OK", stdout.decode('utf-8') # Decode to string else: return "ERROR", stderr.decode('utf-8') # Decode to string except Exception as e: return "ERROR", str(e) def handle_client(conn, addr): """Handles a single client connection.""" print(f"Connected by {addr}") with conn: while True: data = conn.recv(1024) if not data: break message = data.decode('utf-8').strip() print(f"Received: {message}") if message.startswith("COMMAND:"): pandoc_command = message[len("COMMAND:"):].strip() status, output = execute_pandoc(pandoc_command) response = f"STATUS:{status}\nOUTPUT:{output}" conn.sendall(response.encode('utf-8')) else: response = "STATUS:ERROR\nOUTPUT:Invalid command format" conn.sendall(response.encode('utf-8')) with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.bind((HOST, PORT)) s.listen() print(f"Listening on {HOST}:{PORT}") while True: conn, addr = s.accept() handle_client(conn, addr) ``` **Explanation of `pandoc_controller.py`:** * **Imports:** Imports necessary modules: `socket` for network communication, `subprocess` for running Pandoc, `base64` for encoding binary data (if needed), and `os` for file operations. * **`HOST` and `PORT`:** Defines the host address and port the server will listen on. `0.0.0.0` means listen on all available network interfaces. * **`execute_pandoc(command)`:** * Takes the Pandoc command as a string. * Uses `subprocess.Popen` to execute the command. `shell=True` is used for simplicity, but be aware of the security implications if the command comes from an untrusted source. Consider using `shell=False` and passing the command as a list of arguments if possible. * Captures the standard output (`stdout`) and standard error (`stderr`) of the command. * Returns the status ("OK" or "ERROR") and the output (or error message). * **`handle_client(conn, addr)`:** * Handles a single client connection. * Receives data from the client. * Parses the message to extract the Pandoc command (if the message starts with "COMMAND:"). * Calls `execute_pandoc` to run the command. * Constructs an MCP response with the status and output. * Sends the response back to the client. * **Main Loop:** * Creates a socket. * Binds the socket to the specified host and port. * Listens for incoming connections. * Accepts a connection and calls `handle_client` to handle it. **Docker Container Code (Example in Python)** ```python import socket HOST = '172.17.0.1' # Replace with the actual IP address of your host machine PORT = 65432 def send_pandoc_command(command): """Sends a Pandoc command to the host service and returns the output.""" try: with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.connect((HOST, PORT)) message = f"COMMAND:{command}" s.sendall(message.encode('utf-8')) data = s.recv(4096) # Adjust buffer size as needed response = data.decode('utf-8') status = None output = None for line in response.splitlines(): if line.startswith("STATUS:"): status = line[len("STATUS:"):] elif line.startswith("OUTPUT:"): output = line[len("OUTPUT:"):] if status == "OK": return output else: raise Exception(f"Pandoc error: {output}") except Exception as e: print(f"Error communicating with host service: {e}") return None if __name__ == "__main__": pandoc_command = "pandoc -s input.md -o output.pdf" # Example command try: result = send_pandoc_command(pandoc_command) print("Pandoc Output:", result) except Exception as e: print(f"Error: {e}") ``` **Explanation of Docker Container Code:** * **`HOST` and `PORT`:** The IP address of the *host machine* (not the container's IP). You'll need to determine the host's IP address. A common way to find this is to use the gateway IP address of the Docker network. On Linux, you can often find this with `ip route show default`. The IP address after `via` is usually the host's IP. `172.17.0.1` is a common default, but it might be different on your system. The port should match the port the host service is listening on. * **`send_pandoc_command(command)`:** * Creates a socket and connects to the host service. * Constructs the MCP message: `COMMAND:pandoc ...`. * Sends the message to the host service. * Receives the response from the host service. * Parses the response to extract the status and output. * Returns the output if the status is "OK", otherwise raises an exception. * **Main Block:** * Defines the Pandoc command to execute. * Calls `send_pandoc_command` to send the command to the host service. * Prints the output or any error messages. **Steps to Run:** 1. **Save the code:** Save the host service code as `pandoc_controller.py` and the Docker container code as `app.py` (or similar). 2. **Install Pandoc on the Host:** Make sure Pandoc is installed and accessible in the host's `PATH`. 3. **Run the Host Service:** On the host machine, run the Python script: ```bash python3 pandoc_controller.py ``` 4. **Create a Dockerfile:** Create a `Dockerfile` for your Docker container: ```dockerfile FROM python:3.9-slim-buster WORKDIR /app COPY app.py . #COPY input.md . # If you have an input file # Install any dependencies your Docker container needs #RUN pip install ... CMD ["python", "app.py"] ``` 5. **Build the Docker Image:** Build the Docker image: ```bash docker build -t my-pandoc-app . ``` 6. **Run the Docker Container:** Run the Docker container, making sure to map the port and set the correct host IP: ```bash docker run --rm my-pandoc-app ``` **Important Considerations:** * **Security:** This setup has security implications. The Docker container can execute arbitrary Pandoc commands on the host machine. If the Docker container is compromised, the attacker could potentially use this to execute malicious commands on the host. Consider these mitigations: * **Input Validation:** Carefully validate the Pandoc commands received by the host service to prevent command injection attacks. Whitelist allowed options and arguments. * **User Permissions:** Run the host service under a user account with limited privileges. Don't run it as root. * **Firewall:** Use a firewall to restrict access to the host service's port. Only allow connections from the Docker container's IP address. * **Alternatives:** Consider using a more secure approach, such as a dedicated API with authentication and authorization, or running Pandoc in a separate, isolated container. * **Error Handling:** Implement robust error handling in both the host service and the Docker container. Log errors to help with debugging. * **File Access:** If the Pandoc commands need to access files on the host machine, you'll need to handle file paths carefully. Consider using absolute paths or mounting a volume from the host into the Docker container. * **Binary Data:** If Pandoc outputs binary data (like a PDF), you'll need to encode it (e.g., using base64) before sending it over the network. The example code includes `base64` import, but you'll need to add the encoding/decoding logic. Alternatively, you could have Pandoc write the output to a file on a shared volume and then have the Docker container read the file. * **Host IP Address:** The host IP address can change. You might need to use a more dynamic way to determine the host's IP address from within the Docker container. Environment variables or a discovery service could be used. * **MCP Complexity:** For more complex interactions, consider using a more robust messaging protocol like gRPC or REST. However, for simple command execution, MCP is often sufficient. **Example with Base64 Encoding (for Binary Output)** **Host Service (pandoc_controller.py - Modified)** ```python import socket import subprocess import base64 import os HOST = '0.0.0.0' # Listen on all interfaces PORT = 65432 # Port to listen on def execute_pandoc(command): """Executes the Pandoc command and returns the output.""" try: process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE) stdout, stderr = process.communicate() return_code = process.returncode if return_code == 0: # Check if the output is binary (e.g., PDF) try: stdout.decode('utf-8') # Try decoding as UTF-8 is_binary = False except UnicodeDecodeError: is_binary = True if is_binary: output = base64.b64encode(stdout).decode('utf-8') # Encode as base64 else: output = stdout.decode('utf-8') # Decode to string return "OK", output else: return "ERROR", stderr.decode('utf-8') # Decode to string except Exception as e: return "ERROR", str(e) def handle_client(conn, addr): """Handles a single client connection.""" print(f"Connected by {addr}") with conn: while True: data = conn.recv(1024) if not data: break message = data.decode('utf-8').strip() print(f"Received: {message}") if message.startswith("COMMAND:"): pandoc_command = message[len("COMMAND:"):].strip() status, output = execute_pandoc(pandoc_command) response = f"STATUS:{status}\nOUTPUT:{output}" conn.sendall(response.encode('utf-8')) else: response = "STATUS:ERROR\nOUTPUT:Invalid command format" conn.sendall(response.encode('utf-8')) with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.bind((HOST, PORT)) s.listen() print(f"Listening on {HOST}:{PORT}") while True: conn, addr = s.accept() handle_client(conn, addr) ``` **Docker Container Code (app.py - Modified)** ```python import socket import base64 HOST = '172.17.0.1' # Replace with the actual IP address of your host machine PORT = 65432 def send_pandoc_command(command): """Sends a Pandoc command to the host service and returns the output.""" try: with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.connect((HOST, PORT)) message = f"COMMAND:{command}" s.sendall(message.encode('utf-8')) data = s.recv(4096) # Adjust buffer size as needed response = data.decode('utf-8') status = None output = None for line in response.splitlines(): if line.startswith("STATUS:"): status = line[len("STATUS:"):] elif line.startswith("OUTPUT:"): output = line[len("OUTPUT:"):] if status == "OK": return output else: raise Exception(f"Pandoc error: {output}") except Exception as e: print(f"Error communicating with host service: {e}") return None if __name__ == "__main__": pandoc_command = "pandoc -s input.md -o output.pdf" # Example command try: result = send_pandoc_command(pandoc_command) if result: # Decode the base64 encoded PDF data pdf_data = base64.b64decode(result) # Save the PDF to a file with open("output_from_docker.pdf", "wb") as f: f.write(pdf_data) print("PDF saved to output_from_docker.pdf") else: print("No output received from Pandoc.") except Exception as e: print(f"Error: {e}") ``` Key changes in the Base64 example: * **Host Service:** The `execute_pandoc` function now checks if the output is binary. If it is, it encodes the output using base64 before sending it back. It attempts to decode the output as UTF-8; if that fails, it assumes it's binary. * **Docker Container:** The `app.py` script now decodes the base64 encoded data and saves it to a file. This revised example handles binary output (like PDFs) correctly. Remember to adjust the buffer size (`s.recv(4096)`) as needed to accommodate the size of the output. This comprehensive explanation and code should give you a solid foundation for building your MCP-based Pandoc controller. Remember to prioritize security and error handling in your implementation.

CoderSwap MCP Server

Enables AI agents to autonomously create and manage topic-specific vector knowledge bases with end-to-end functionality including project creation, content ingestion from URLs, semantic search, and progress tracking. Provides a complete research workflow without exposing low-level APIs.

Academic MCP

Enables users to search, download, and read academic papers from multiple platforms including arXiv, PubMed, bioRxiv, Google Scholar, Semantic Scholar, and CrossRef through a unified interface.

Netmiko MCP Server

Enables interaction with Cisco network devices through SSH using Netmiko. Supports retrieving device status, interface details, configurations, and resource monitoring through natural language commands in Claude Desktop.

mcp-zeroentropy

MCP Server for ZeroEntropy collections, top documents and rerankers

MCP Manager

An enterprise-level MCP gateway and proxy that sits between an organization's MCP servers and clients. MCP Manager mitigates security threats, enables fine-grained permissions, enforces policies and guardrails, and generates comprehensive, end-to-end logs.

Futuur API MCP Integration

Futuur API MCP Integration é um servidor poderoso baseado em TypeScript que implementa o Protocolo de Contexto de Modelo (MCP) para uma integração perfeita com a API Futuur. Este projeto fornece uma interface robusta para lidar com dados de mercado, categorias, informações de usuário e operações de apostas.

MCP Recipes Server

Um servidor que expõe ferramentas para consultar receitas usando o Protocolo de Contexto do Modelo (MCP).

NanoKVM MCP Server

Enables AI assistants to remotely control Sipeed NanoKVM hardware for BIOS-level management of servers and headless machines. It provides tools for power control, keyboard and mouse emulation, screen capture, and ISO image mounting via the Model Context Protocol.

Trade Surveillance Support MCP Server

Automates trade surveillance support workflows by parsing inquiry emails, searching SQL configs and Java code using keyword-based metadata annotations, executing reports, and generating comprehensive responses.

AI-Powered Server-Client Computer Use SDK

File containing information about the use of artificial intelligence software development kit (SDK) on servers and clients for computer systems.