Discover Awesome MCP Servers

MCP Knowledge Base Server

Provides semantic search and data retrieval capabilities over a knowledge base with multiple tools including keyword search, category filtering, and ID-based lookup with in-memory caching.

DesignBot MCP

Forwards messages to the Designsystemet assistant endpoint, enabling access through MCP-compatible clients.

Aws Sample Gen Ai Mcp Server

Okay, here's a sample code snippet demonstrating how to use Gen-AI (Bedrock) with an MCP (Message Control Protocol) server. This example focuses on the core concepts and assumes you have the necessary libraries and configurations set up. It's a simplified illustration and will need adaptation based on your specific MCP server and Bedrock use case. **Conceptual Overview** 1. **MCP Server:** This acts as a central point for receiving requests. It could be a simple TCP server or a more sophisticated message queue system. The code below uses a basic TCP server for demonstration. 2. **Bedrock (Gen-AI):** This is where the AI model resides. You'll use the Bedrock API to send prompts and receive responses. 3. **Workflow:** * The MCP server receives a request (e.g., a text prompt). * The server forwards the prompt to Bedrock. * Bedrock processes the prompt and returns a response. * The server sends the response back to the client. **Python Example (using `socket` for MCP and `boto3` for Bedrock)** ```python import socket import boto3 import json # Configuration (replace with your actual values) MCP_HOST = 'localhost' # Or your MCP server's IP address MCP_PORT = 12345 # Or your MCP server's port BEDROCK_REGION = 'us-east-1' # Or your Bedrock region BEDROCK_MODEL_ID = 'anthropic.claude-v2' # Or your desired Bedrock model ID ACCEPTABLE_ORIGINS = ["localhost", "127.0.0.1"] # Add any other acceptable origins here # Initialize Bedrock client bedrock = boto3.client(service_name='bedrock-runtime', region_name=BEDROCK_REGION) def handle_request(client_socket, client_address): """Handles a single request from a client.""" try: data = client_socket.recv(1024).decode('utf-8') if not data: return # Client disconnected print(f"Received from {client_address}: {data}") # Check origin (very basic example - improve this for production!) try: request_json = json.loads(data) origin = request_json.get("origin", None) prompt = request_json.get("prompt", None) except json.JSONDecodeError: print("Invalid JSON received") client_socket.sendall("Invalid JSON".encode('utf-8')) return if origin not in ACCEPTABLE_ORIGINS: print(f"Request from unacceptable origin: {origin}") client_socket.sendall("Origin not allowed".encode('utf-8')) return if not prompt: print("No prompt provided") client_socket.sendall("No prompt provided".encode('utf-8')) return # Call Bedrock try: response = invoke_bedrock(prompt) client_socket.sendall(response.encode('utf-8')) except Exception as e: print(f"Bedrock error: {e}") client_socket.sendall(f"Bedrock error: {e}".encode('utf-8')) except Exception as e: print(f"Error handling request: {e}") finally: client_socket.close() def invoke_bedrock(prompt): """Invokes the Bedrock model with the given prompt.""" # Construct the request body (adjust based on the model) body = json.dumps({ "prompt": prompt, "max_tokens_to_sample": 200, # Adjust as needed "temperature": 0.5, # Adjust as needed "top_p": 0.9 # Adjust as needed }) try: response = bedrock.invoke_model( modelId=BEDROCK_MODEL_ID, contentType='application/json', accept='application/json', body=body ) response_body = json.loads(response['body'].read().decode('utf-8')) completion = response_body['completion'] # Adjust based on model's response format return completion except Exception as e: print(f"Error invoking Bedrock: {e}") return f"Error: {e}" 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) # Listen for up to 5 incoming connections print(f"MCP server listening on {MCP_HOST}:{MCP_PORT}") while True: client_socket, client_address = server_socket.accept() print(f"Accepted connection from {client_address}") handle_request(client_socket, client_address) # Handle the request in a separate function if __name__ == "__main__": start_mcp_server() ``` **Explanation:** * **Imports:** Imports necessary libraries (`socket`, `boto3`, `json`). * **Configuration:** Sets up configuration variables for the MCP server address, port, Bedrock region, and model ID. **Crucially, replace these with your actual values.** * **`handle_request()`:** * Receives data from the client socket. * Decodes the data (assuming UTF-8 encoding). * **Important:** Includes a very basic origin check. **This is a placeholder and needs to be significantly improved for any production environment.** You should implement robust authentication and authorization. The example expects a JSON payload with `origin` and `prompt` fields. * Calls `invoke_bedrock()` to send the prompt to Bedrock. * Sends the response back to the client. * Handles potential errors. * Closes the client socket. * **`invoke_bedrock()`:** * Constructs the request body for the Bedrock API. **This is highly model-dependent.** The example shows a basic structure for Anthropic Claude. You'll need to consult the Bedrock documentation for the specific model you're using to determine the correct request format. * Calls the `bedrock.invoke_model()` method. * Parses the response from Bedrock. **Again, the response format is model-dependent.** The example assumes a `completion` field in the response. * Handles potential errors. * **`start_mcp_server()`:** * Creates a TCP socket. * Binds the socket to the specified host and port. * Listens for incoming connections. * Accepts connections in a loop. * Calls `handle_request()` to process each connection. * **`if __name__ == "__main__":`:** Starts the MCP server when the script is run. **How to Run:** 1. **Install Libraries:** ```bash pip install boto3 ``` 2. **Configure AWS Credentials:** Make sure you have configured your AWS credentials (e.g., using `aws configure` or environment variables) so that `boto3` can access Bedrock. The IAM role or user you're using must have permissions to invoke the Bedrock model. 3. **Replace Placeholders:** Update the configuration variables at the top of the script with your actual values. 4. **Run the Script:** ```bash python your_script_name.py ``` 5. **Test with a Client:** You'll need a client application to send requests to the MCP server. Here's a simple Python client example: ```python import socket import json MCP_HOST = 'localhost' MCP_PORT = 12345 def send_request(prompt, origin): """Sends a request to the MCP server.""" with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.connect((MCP_HOST, MCP_PORT)) message = json.dumps({"prompt": prompt, "origin": origin}) s.sendall(message.encode('utf-8')) data = s.recv(1024) print(f"Received: {data.decode('utf-8')}") if __name__ == "__main__": prompt = "Write a short poem about the ocean." origin = "localhost" # Or "127.0.0.1" send_request(prompt, origin) ``` **Important Considerations:** * **Error Handling:** The error handling in the example is basic. You should implement more robust error handling, including logging and retries. * **Security:** The origin check is extremely basic. For production environments, you *must* implement proper authentication and authorization to prevent unauthorized access. Consider using TLS/SSL for secure communication. * **Scalability:** For high-volume traffic, consider using a more scalable MCP server architecture, such as a message queue (e.g., RabbitMQ, Kafka) or a load balancer. You might also need to scale your Bedrock usage. * **Bedrock Model Configuration:** The `invoke_bedrock()` function needs to be carefully configured based on the specific Bedrock model you're using. Refer to the Bedrock documentation for the model's input and output formats, available parameters, and best practices. * **Asynchronous Processing:** For better performance, consider using asynchronous programming (e.g., `asyncio`) to handle multiple requests concurrently. * **Rate Limiting:** Be aware of Bedrock's rate limits and implement appropriate rate limiting in your MCP server to avoid exceeding those limits. * **Data Validation:** Validate the data received from clients to prevent malicious input. * **Logging:** Implement comprehensive logging to track requests, responses, and errors. This example provides a starting point. You'll need to adapt it to your specific requirements and environment. Remember to prioritize security, error handling, and scalability as you develop your application.

Apple Mail Summary MCP

Enables AI agents to fetch emails from local Apple Mail accounts and mailboxes, and parse Google Scholar alert emails to extract paper titles and links.

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.

Context Engine

A task-aware context compression layer for Agent workflows, RAG pipelines, and AI Coding assistants, reducing noisy logs, retrieval chunks, and code context into high-signal LLM inputs via CLI, Python SDK, and MCP.

pipedrive-mcp

MCP server for Pipedrive CRM providing 88 tools for full CRUD on deals, persons, organizations, activities, and more, with custom field resolution and safety guards.

viraill-mcp

Enables MCP clients to audit AI search visibility, generate intent-aligned social content, and assess Agentic Commerce Readiness with remediation files.

AppFlowy MCP Server

Provides AI assistants with full read/write access to AppFlowy Cloud, enabling management of workspaces, pages, databases, trash, and favorites, plus conversion of Markdown into formatted AppFlowy document blocks.

Perplexity MCP Server

Enables web search and deep research using Perplexity Sonar models through MCP clients like Cursor or Claude Desktop.

wxauto-mcp

Windows微信自动化MCP服务，提供UIA探测、加好友、发送消息、读取消息、发布朋友圈等工具，通过单消费者队列串行执行真实动作。

LLMMO Game Server

Enables LLM-driven text game state management by exposing MCP tools for managing players, locations, items, entities, and abstract concepts.

Dummy MCP Server

A simple Meta-agent Communication Protocol server built with FastMCP framework that provides 'echo' and 'dummy' tools via Server-Sent Events for demonstration and testing purposes.

zulip-mcp

MCP server that exposes Zulip REST API tools via SSE, enabling message retrieval, stream/topic listing, draft management, and notifications.

magnolia-docs-mcp

Enables AI assistants to search and retrieve information from the official Magnolia CMS documentation through a set of MCP tools.

raalarcon-jira-mcp-server

Open source MCP Server for Jira & Atlassian — manage issues, sprints, comments & Confluence via Claude, Cursor, or any MCP client

Sanity MCP Server

A self-hosted Sanity MCP server with full CRUD, atomic transactions, reference tracking, and advanced tools for content management and operations.

GrabzIt MCP Server

Enables AI assistants to capture website screenshots, generate PDF/DOCX documents, and scrape rendered web data through the GrabzIt API.

mcp-workflowy

mcp-workflowy

Codex to OpenCode MCP Server

Enables Codex to delegate coding tasks to an OpenCode CLI locally, returning structured results such as exit codes, session summaries, tool calls, and git diffs.

mcp-imagenate

An MCP server for image generation using multiple providers including Google Gemini, OpenAI, and BFL FLUX. It supports various models, aspect ratios, and resolutions, with options for image and text output.

DailyBot MCP

Enables Littlebird to read and submit DailyBot check-ins through the DailyBot CLI.

PitchLink MCP

An MCP server that reads startup pitch drafts from Notion to provide comprehensive investor-style analysis and scoring. It evaluates key areas like market opportunity and team strength, delivering feedback through a visual dashboard.

面试鸭 MCP Server

Spring AI をベースにした面接アヒル検索問題の MCP Server サービスで、AI が企業の面接本番問題と回答を迅速に検索できるようにします。

Shopify MCP Server by CData

Shopify MCP Server by CData

Agentic Tools MCP Server

A Model Context Protocol server providing AI assistants with comprehensive project, task, and subtask management capabilities with project-specific storage.

sliverc2-mcp

sliverc2-mcp

Trello MCP Server

Enables AI assistants to retrieve Trello card information by ID or link, providing access to card details including labels, members, due dates, and attachments through a standardized interface.

ssh-client-mcp-server

A local MCP server that enables LLMs to execute shell commands on remote hosts over SSH with multiple authentication methods.

Grok Build Bridge

Connects OpenAI Codex and Claude Code to the local Grok Build CLI for second-opinion code reviews, bounded coding tasks, web research, media generation, and TTS.