Discover Awesome MCP Servers

Brosh Browser Screenshot

Captures comprehensive webpage screenshots with intelligent scrolling, text extraction, and HTML analysis, enabling AI tools to visually inspect and understand web content through the Model Context Protocol.

jsx-notation

Compresses React/Next.js files, HTML, and SVG into a compact JSXN notation optimized for AI assistants, reducing token consumption by ~40%. Provides an MCP server with tools to read, encode, and decode JSXN.

ncbi-mcp

Máy chủ MCP của Trung tâm Thông tin Sinh học Quốc gia (NCBI) thuộc NIH

booking_chest

MCP server for Cal.com scheduling, providing ~70 tools to manage schedules, event types, bookings, calendars, webhooks, and teams. Enables natural language control of Cal.com from Claude or any MCP-compatible client.

MCP Neurolora

An intelligent MCP server that provides tools for code analysis using OpenAI API, code collection, and documentation generation.

OfficeRnD MCP Server

A read-only MCP server that connects AI assistants to the OfficeRnD coworking and flex-space management platform. It enables natural language queries for community members, space bookings, billing records, and office resources.

PDFSizeAnalyzer-MCP

Enables comprehensive PDF analysis and manipulation including page size analysis, chapter extraction, splitting, compression, merging, and conversion to images. Provides both MCP server interface for AI assistants and Streamlit web interface for direct user interaction.

kilo-dev-mcp-server

An MCP server for internationalization (i18n) tasks, providing tools to translate, move, list, and remove translation keys in JSON files for the Kilo Code extension.

llama-mcp-server

MCP server bridging Claude Code to local llama.cpp. Run local LLMs alongside Claude for experimentation, testing, and cost-effective inference.

anthilll-inside

MCP server that reproduces SSE response-size hangs with Anthropic MCP clients by providing a configurable 'generate_lipsum' tool.

GitHub Integration Hub

Enables AI agents to interact with GitHub through OAuth-authenticated operations including starting authorization flows, listing repositories, and creating issues using stored access tokens.

DALL-E MCP Server

Enables AI assistants to generate high-quality images using OpenAI's DALL-E 3 model with configurable parameters like size, quality, and style. Generated images are automatically saved to the local filesystem with comprehensive error handling.

OpenSearch MCP Server

Enables LLMs to interact with OpenSearch clusters to monitor cluster health, manage indices, and perform data searches. It provides a standardized interface for real-time OpenSearch operations within MCP-compatible environments like Open WebUI.

notepm-mcp-server

An unofficial MCP server that enables AI assistants to search, read, create, and update pages in your NotePM workspace.

Spica MCP Server

MCP server that enables AI agents to interact with Spica servers, managing databases, serverless functions, storage, authentication, auditing, debugging, and version control.

MCP Unity Bridge Asset

Asset to be imported into Unity to host a WebSocket server for MCP Conmmunciation with LLMs

Terragrunt MCP Server

Enables analysis and validation of Terragrunt projects, including dependency mapping, stack structure, and optimization suggestions, directly in IDEs that support MCP.

Elasticsearch MCP Server AI

Transforms Elasticsearch into an AI-powered observability engine for analyzing logs, APM traces, and system metrics through natural language.

GraphSense MCP Server

Enables natural language queries about JavaScript/TypeScript codebases using graph and vector search. Allows finding functions by semantic description, callers, callees, and function details.

niscope-mcp

Enables AI assistants to directly control NI oscilloscopes (e.g., PXIe-5160/5164/5110) through the Model Context Protocol, including waveform acquisition, measurement, and configuration.

WikiJS MCP Server

Enables AI assistants to search and retrieve content from WikiJS knowledge bases, allowing integration with your Wiki through simple search and retrieval tools.

Arcane MCP

Enables AI assistants to manage Docker containers, images, volumes, networks, and compose projects through Arcane's API. Provides comprehensive tools for environment, container, image, volume, network, project, and system operations.

cockpit-mcp

Accounting automation MCP server for Japanese tax firms. Two-stage classifier (keyword + Claude AI) with TaxRuleEngine for consumption tax, withholding, invoice system, and confidence-based routing.

Openfort MCP Server

Enables AI assistants to interact with Openfort's wallet infrastructure, allowing them to create projects, manage configurations, generate wallets and users, and query documentation through 42 integrated tools.

MCP Server Proxy

Aggregates multiple MCP servers into a single endpoint, enabling LLM clients to access tools, resources, and prompts from various backends through one connection.

literature-agent-mcp

Exposes a local biomedical literature pipeline as MCP tools for automated research workflows. Enables literature search, open-access paper retrieval, and draft generation for biomedical and pathology domains through standard MCP clients.

Aws Sample Gen Ai Mcp Server

Okay, I can provide you with a sample code structure and explanation for using Gen-AI (specifically, Bedrock) with an MCP (Message Control Protocol) server in Python. This will be a conceptual outline, as a fully functional example would require specific details about your MCP server implementation and the exact Bedrock models you intend to use. **Conceptual Overview** The basic idea is: 1. **MCP Server:** This server receives requests (likely text prompts) from clients. 2. **Bedrock Integration:** The server takes the prompt, sends it to Amazon Bedrock, and receives a generated response. 3. **Response to Client:** The server sends the generated response back to the client via the MCP protocol. **Python Code Structure (Illustrative)** ```python import socket import json import boto3 # AWS SDK for Python (Boto3) # Configuration (Replace with your actual values) MCP_HOST = 'localhost' # Or your server's IP address MCP_PORT = 12345 # Or your server's port AWS_REGION = 'us-east-1' # Or your AWS region BEDROCK_MODEL_ID = 'anthropic.claude-v2' # Example model ID (replace) # Initialize Bedrock client bedrock = boto3.client(service_name='bedrock-runtime', region_name=AWS_REGION) def generate_text_with_bedrock(prompt, model_id=BEDROCK_MODEL_ID): """ Sends a prompt to Amazon Bedrock and returns the generated text. """ try: # Construct the request body (format depends on the model) if model_id.startswith("anthropic"): body = json.dumps({ "prompt": prompt, "max_tokens_to_sample": 200, # Adjust as needed "temperature": 0.5, # Adjust as needed "top_p": 0.9 }) content_type = 'application/json' accept = 'application/json' elif model_id.startswith("ai21"): body = json.dumps({ "prompt": prompt, "maxTokens": 200, "temperature": 0.7, "topP": 1 }) content_type = 'application/json' accept = 'application/json' else: raise ValueError(f"Unsupported model ID: {model_id}") response = bedrock.invoke_model( modelId=model_id, contentType=content_type, accept=accept, body=body ) response_body = json.loads(response['body'].read().decode('utf-8')) # Extract the generated text (depends on the model's response format) if model_id.startswith("anthropic"): generated_text = response_body['completion'] elif model_id.startswith("ai21"): generated_text = response_body['completions'][0]['data']['text'] else: raise ValueError(f"Unsupported model ID: {model_id}") return generated_text except Exception as e: print(f"Error generating text: {e}") return None def handle_client(client_socket): """ Handles communication with a single client. """ try: # Receive data from the client (assuming MCP format) data = client_socket.recv(1024).decode('utf-8') # Adjust buffer size as needed if not data: print("Client disconnected.") return # Parse the MCP message (assuming a simple JSON format for the prompt) try: message = json.loads(data) prompt = message.get('prompt') # Assuming the prompt is in a 'prompt' field model_id = message.get('model_id', BEDROCK_MODEL_ID) # Allow client to specify model except json.JSONDecodeError: print("Invalid JSON received.") client_socket.sendall("ERROR: Invalid JSON".encode('utf-8')) return if not prompt: print("No prompt received.") client_socket.sendall("ERROR: No prompt provided".encode('utf-8')) return # Generate text using Bedrock generated_text = generate_text_with_bedrock(prompt, model_id) if generated_text: # Construct the response message (MCP format) response_message = json.dumps({'response': generated_text}) client_socket.sendall(response_message.encode('utf-8')) else: client_socket.sendall("ERROR: Text generation failed".encode('utf-8')) except Exception as e: print(f"Error handling client: {e}") client_socket.sendall("ERROR: Internal server error".encode('utf-8')) finally: client_socket.close() def start_mcp_server(): """ Starts the MCP server. """ server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) server_socket.bind((MCP_HOST, MCP_PORT)) server_socket.listen(5) # Allow up to 5 pending connections print(f"MCP server listening on {MCP_HOST}:{MCP_PORT}") try: while True: client_socket, client_address = server_socket.accept() print(f"Accepted connection from {client_address}") # Handle the client in a separate thread or process (recommended for concurrency) # For simplicity, I'm calling it directly here, but threading/multiprocessing is better handle_client(client_socket) except KeyboardInterrupt: print("Server shutting down.") finally: server_socket.close() if __name__ == "__main__": start_mcp_server() ``` **Explanation and Vietnamese Translation** ```python import socket # Thư viện cho giao tiếp mạng socket import json # Thư viện để làm việc với dữ liệu JSON import boto3 # AWS SDK cho Python (Boto3) để tương tác với Bedrock # Cấu hình (Thay thế bằng giá trị thực tế của bạn) MCP_HOST = 'localhost' # Hoặc địa chỉ IP của máy chủ MCP của bạn MCP_PORT = 12345 # Hoặc cổng của máy chủ MCP của bạn AWS_REGION = 'us-east-1' # Hoặc khu vực AWS của bạn BEDROCK_MODEL_ID = 'anthropic.claude-v2' # ID của mô hình Bedrock (ví dụ, thay thế bằng mô hình bạn muốn dùng) # Khởi tạo client Bedrock bedrock = boto3.client(service_name='bedrock-runtime', region_name=AWS_REGION) def generate_text_with_bedrock(prompt, model_id=BEDROCK_MODEL_ID): """ Gửi một prompt đến Amazon Bedrock và trả về văn bản được tạo ra. """ try: # Xây dựng phần thân của yêu cầu (định dạng phụ thuộc vào mô hình) if model_id.startswith("anthropic"): body = json.dumps({ "prompt": prompt, "max_tokens_to_sample": 200, # Điều chỉnh nếu cần "temperature": 0.5, # Điều chỉnh nếu cần (độ ngẫu nhiên) "top_p": 0.9 # Điều chỉnh nếu cần (xác suất) }) content_type = 'application/json' accept = 'application/json' elif model_id.startswith("ai21"): body = json.dumps({ "prompt": prompt, "maxTokens": 200, "temperature": 0.7, "topP": 1 }) content_type = 'application/json' accept = 'application/json' else: raise ValueError(f"Unsupported model ID: {model_id}") response = bedrock.invoke_model( modelId=model_id, contentType=content_type, accept=accept, body=body ) response_body = json.loads(response['body'].read().decode('utf-8')) # Trích xuất văn bản được tạo ra (phụ thuộc vào định dạng phản hồi của mô hình) if model_id.startswith("anthropic"): generated_text = response_body['completion'] elif model_id.startswith("ai21"): generated_text = response_body['completions'][0]['data']['text'] else: raise ValueError(f"Unsupported model ID: {model_id}") return generated_text except Exception as e: print(f"Lỗi khi tạo văn bản: {e}") return None def handle_client(client_socket): """ Xử lý giao tiếp với một client duy nhất. """ try: # Nhận dữ liệu từ client (giả sử định dạng MCP) data = client_socket.recv(1024).decode('utf-8') # Điều chỉnh kích thước buffer nếu cần if not data: print("Client đã ngắt kết nối.") return # Phân tích cú pháp thông điệp MCP (giả sử định dạng JSON đơn giản cho prompt) try: message = json.loads(data) prompt = message.get('prompt') # Giả sử prompt nằm trong trường 'prompt' model_id = message.get('model_id', BEDROCK_MODEL_ID) # Cho phép client chỉ định model except json.JSONDecodeError: print("JSON không hợp lệ đã nhận.") client_socket.sendall("ERROR: Invalid JSON".encode('utf-8')) return if not prompt: print("Không nhận được prompt.") client_socket.sendall("ERROR: No prompt provided".encode('utf-8')) return # Tạo văn bản bằng Bedrock generated_text = generate_text_with_bedrock(prompt, model_id) if generated_text: # Xây dựng thông điệp phản hồi (định dạng MCP) response_message = json.dumps({'response': generated_text}) client_socket.sendall(response_message.encode('utf-8')) else: client_socket.sendall("ERROR: Text generation failed".encode('utf-8')) except Exception as e: print(f"Lỗi khi xử lý client: {e}") client_socket.sendall("ERROR: Internal server error".encode('utf-8')) finally: client_socket.close() def start_mcp_server(): """ Khởi động máy chủ MCP. """ server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) server_socket.bind((MCP_HOST, MCP_PORT)) server_socket.listen(5) # Cho phép tối đa 5 kết nối đang chờ print(f"Máy chủ MCP đang lắng nghe trên {MCP_HOST}:{MCP_PORT}") try: while True: client_socket, client_address = server_socket.accept() print(f"Đã chấp nhận kết nối từ {client_address}") # Xử lý client trong một thread hoặc process riêng biệt (khuyến nghị để đồng thời) # Để đơn giản, tôi gọi trực tiếp ở đây, nhưng threading/multiprocessing thì tốt hơn handle_client(client_socket) except KeyboardInterrupt: print("Máy chủ đang tắt.") finally: server_socket.close() if __name__ == "__main__": start_mcp_server() ``` **Key Points and Considerations** * **Error Handling:** The code includes basic error handling, but you should expand it to be more robust. Consider logging errors to a file. * **Concurrency:** The `handle_client` function is called directly in the `start_mcp_server` loop. This means the server can only handle one client at a time. Use threads or processes to handle multiple clients concurrently. The `threading` or `multiprocessing` modules in Python are suitable for this. * **MCP Protocol:** This example assumes a very simple JSON-based MCP protocol. You'll need to adapt the `handle_client` function to correctly parse and format messages according to your actual MCP protocol. Consider using a dedicated MCP library if one exists. * **Bedrock Model Parameters:** The `generate_text_with_bedrock` function includes parameters like `max_tokens_to_sample`, `temperature`, and `top_p`. These parameters control the generation process. Experiment with different values to get the desired results from your chosen Bedrock model. The specific parameters available and their meanings vary depending on the model. Refer to the Bedrock documentation for the model you are using. * **Authentication and Authorization:** This example doesn't include any authentication or authorization. In a real-world application, you'll need to implement security measures to protect your Bedrock API key and prevent unauthorized access to your server. Consider using IAM roles and policies for Bedrock access. * **Model ID:** Make sure the `BEDROCK_MODEL_ID` is correct and that you have access to that model in your AWS account and region. * **Region:** Ensure the `AWS_REGION` is the region where you have access to Bedrock. * **Dependencies:** You'll need to install the `boto3` library: `pip install boto3` **How to Use** 1. **Install Boto3:** `pip install boto3` 2. **Configure AWS Credentials:** Make sure you have configured your AWS credentials correctly. The easiest way is to configure the AWS CLI: `aws configure`. You'll need an AWS account and an IAM user with permissions to access Bedrock. 3. **Replace Placeholders:** Update the `MCP_HOST`, `MCP_PORT`, `AWS_REGION`, and `BEDROCK_MODEL_ID` variables with your actual values. 4. **Run the Server:** Execute the Python script. 5. **Create a Client:** Write a client application that connects to the MCP server and sends JSON messages with a "prompt" field. **Example Client (Python)** ```python import socket import json MCP_HOST = 'localhost' MCP_PORT = 12345 def send_prompt(prompt): client_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) client_socket.connect((MCP_HOST, MCP_PORT)) message = json.dumps({'prompt': prompt}) client_socket.sendall(message.encode('utf-8')) response = client_socket.recv(1024).decode('utf-8') client_socket.close() return response if __name__ == "__main__": user_prompt = "Write a short story about a cat who goes on an adventure." response = send_prompt(user_prompt) print(f"Response from server: {response}") ``` **Important Disclaimer:** This is a simplified example. Building a production-ready system requires careful consideration of security, scalability, error handling, and the specific requirements of your application. Always consult the official documentation for Amazon Bedrock and the MCP protocol you are using.

mcp-altegio

MCP server for Altegio API — appointments, clients, services, staff schedules

Stripe MCP Server

Enables Claude to integrate with Stripe for creating payment links, processing payments, and managing products and customers.

database-updater MCP Server

Gương của