Discover Awesome MCP Servers

ETH Price Current Server

A minimal Model Context Protocol (MCP) server that fetches the current Ethereum (ETH) price in USD. Data source: the public CoinGecko API (no API key required). This MCP is designed to simulate malicious behavior, specifically an attempt to mislead LLM to return incorrect results.

Teable MCP Server

Connects Teable no-code databases to LLMs, enabling AI agents to query records, explore schema structures, retrieve data history, and interact with spaces, bases, tables, and views using natural language.

MCP Demo Server

A minimal fastmcp demonstration server that provides a simple addition tool through the MCP protocol, supporting deployment via Docker with multiple transport modes.

mcp_server

Okay, I understand. You want me to describe how to implement a weather MCP (Message Client Protocol) server that can be called by a client IDE like Cursor. Here's a breakdown of the implementation, covering key aspects and considerations: **1. Understanding the Requirements** * **MCP (Message Client Protocol):** MCP is a relatively simple protocol, often used for communication between applications. It's typically text-based and uses delimiters to separate messages. We'll need to define the specific MCP format for our weather server. * **Weather Data Source:** You'll need a reliable source of weather data. This could be: * **A free weather API:** Examples include OpenWeatherMap, WeatherAPI.com (check their free tiers and usage limits). * **A paid weather API:** Provides more features, higher limits, and potentially better accuracy. * **A local weather data file:** For testing or if you have your own data source. * **Client IDE (Cursor):** The server needs to be designed so that Cursor can easily send requests and receive responses. This means using a standard communication method (like TCP sockets) and a well-defined MCP format. * **Functionality:** At a minimum, the server should be able to: * Get the current weather for a given location (e.g., city, zip code, latitude/longitude). * Potentially provide a forecast for a given location. **2. Defining the MCP Format** This is crucial. Let's define a simple MCP format for our weather server. We'll use a newline character (`\n`) as the delimiter between messages. * **Request Format (Client to Server):** ``` GET_WEATHER <location>\n GET_FORECAST <location>\n ``` * `GET_WEATHER`: Requests the current weather for the specified location. * `GET_FORECAST`: Requests a weather forecast for the specified location. * `<location>`: The location to get weather data for (e.g., "London", "90210", "40.7128,-74.0060"). You'll need to decide how you want to handle different location formats. * **Response Format (Server to Client):** ``` WEATHER <location> <temperature> <condition> <humidity>\n FORECAST <location> <day1_condition> <day1_high> <day1_low> <day2_condition> ...\n ERROR <message>\n ``` * `WEATHER`: Indicates the current weather data. * `<location>`: The location the weather data is for. * `<temperature>`: The current temperature (e.g., in Celsius or Fahrenheit). * `<condition>`: A brief description of the weather (e.g., "Sunny", "Cloudy", "Rainy"). * `<humidity>`: The current humidity (e.g., as a percentage). * `FORECAST`: Indicates the weather forecast data. * `<location>`: The location the forecast is for. * `<day1_condition>`, `<day1_high>`, `<day1_low>`, etc.: Forecast data for each day. You'll need to define how many days of forecast you'll provide and the format of the data. * `ERROR`: Indicates an error occurred. * `<message>`: A description of the error. **3. Server Implementation (Python Example)** Here's a basic Python example using the `socket` module. This is a simplified version; you'll need to add error handling, more robust data parsing, and potentially threading for handling multiple clients concurrently. ```python import socket import json # For parsing JSON responses from weather APIs import requests # For making HTTP requests to weather APIs # Configuration HOST = '127.0.0.1' # Standard loopback interface address (localhost) PORT = 65432 # Port to listen on (non-privileged ports are > 1023) WEATHER_API_KEY = "YOUR_OPENWEATHERMAP_API_KEY" # Replace with your API key WEATHER_API_URL = "http://api.openweathermap.org/data/2.5/weather?q={}&appid={}&units=metric" # Example URL def get_weather_data(location): """Fetches weather data from a weather API.""" try: url = WEATHER_API_URL.format(location, WEATHER_API_KEY) response = requests.get(url) response.raise_for_status() # Raise HTTPError for bad responses (4xx or 5xx) data = response.json() temperature = data['main']['temp'] condition = data['weather'][0]['description'] humidity = data['main']['humidity'] return temperature, condition, humidity except requests.exceptions.RequestException as e: print(f"Error fetching weather data: {e}") return None, None, None except (KeyError, IndexError) as e: print(f"Error parsing weather data: {e}") return None, None, None def handle_client(conn, addr): """Handles communication with a single client.""" print(f"Connected by {addr}") while True: data = conn.recv(1024) # Receive data from the client if not data: break # Client disconnected message = data.decode('utf-8').strip() print(f"Received: {message}") if message.startswith("GET_WEATHER"): try: location = message.split(" ")[1] temperature, condition, humidity = get_weather_data(location) if temperature is not None: response = f"WEATHER {location} {temperature} {condition} {humidity}\n" else: response = f"ERROR Could not retrieve weather for {location}\n" except IndexError: response = "ERROR Invalid request format. Use GET_WEATHER <location>\n" except Exception as e: response = f"ERROR An unexpected error occurred: {e}\n" elif message.startswith("GET_FORECAST"): # Implement forecast logic here (similar to GET_WEATHER) location = message.split(" ")[1] response = f"ERROR Forecast functionality not yet implemented for {location}\n" else: response = "ERROR Invalid command\n" conn.sendall(response.encode('utf-8')) # Send the response back to the client conn.close() print(f"Connection closed with {addr}") def main(): """Main server function.""" 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() # Accept incoming connections handle_client(conn, addr) # Handle the client in a separate function (or thread) if __name__ == "__main__": main() ``` Key improvements and explanations: * **Error Handling:** Includes `try...except` blocks to catch potential errors during API requests, JSON parsing, and other operations. This prevents the server from crashing and provides more informative error messages to the client. * **Weather API Integration:** Uses the `requests` library to make HTTP requests to a weather API (OpenWeatherMap in this example). You'll need to replace `"YOUR_OPENWEATHERMAP_API_KEY"` with your actual API key. The `get_weather_data` function handles the API call and parses the JSON response. * **JSON Parsing:** Uses the `json` library to parse the JSON response from the weather API. * **Clearer Response Messages:** The server now sends more informative error messages to the client when something goes wrong. * **`handle_client` Function:** The code that handles communication with a single client is now in a separate function, `handle_client`. This makes the code more organized and easier to read. * **UTF-8 Encoding:** Explicitly encodes and decodes messages using UTF-8 to handle a wider range of characters. * **`response.raise_for_status()`:** This line in `get_weather_data` is important. It raises an `HTTPError` if the API returns a bad status code (4xx or 5xx), which indicates a problem with the request or the API itself. * **`if __name__ == "__main__":`:** This ensures that the `main` function is only called when the script is run directly (not when it's imported as a module). **4. Client Implementation (Cursor IDE)** You'll need to write code within Cursor to connect to the server, send requests, and display the responses. Here's a conceptual outline (the exact code will depend on Cursor's API and capabilities): ```python import socket HOST = '127.0.0.1' PORT = 65432 def get_weather(location): """Gets weather data from the server.""" try: with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: s.connect((HOST, PORT)) request = f"GET_WEATHER {location}\n" s.sendall(request.encode('utf-8')) data = s.recv(1024) response = data.decode('utf-8').strip() return response except Exception as e: return f"Error: {e}" # Example usage (within Cursor): location = "London" # Or get the location from user input in Cursor weather_data = get_weather(location) print(weather_data) # Display the weather data in Cursor's output window ``` **5. Key Considerations and Enhancements** * **Threading/Asynchronous Handling:** For a production server, use threading or asynchronous I/O (e.g., `asyncio` in Python) to handle multiple client connections concurrently. The basic example above handles only one client at a time. * **Error Handling:** Implement robust error handling on both the client and server sides. Handle network errors, API errors, invalid input, etc. * **Input Validation:** Validate the location input to prevent errors and potential security issues. * **Security:** If you're exposing the server to the internet, consider security measures like authentication and authorization. For local development, this is less critical. * **Configuration:** Use a configuration file (e.g., a `.ini` or `.json` file) to store settings like the host, port, API key, and other parameters. This makes the server more flexible and easier to configure. * **Logging:** Implement logging to track server activity, errors, and other important events. This is helpful for debugging and monitoring. * **Data Caching:** Cache weather data to reduce the number of API calls and improve performance. Use a caching library like `cachetools` or `redis`. * **More Sophisticated MCP:** Consider a more structured MCP format using JSON or another data serialization format for more complex data structures. However, keep it relatively simple for ease of implementation. * **Unit Testing:** Write unit tests to ensure that the server is working correctly. **Example Interaction** 1. **Client (Cursor) sends:** `GET_WEATHER London\n` 2. **Server receives:** `GET_WEATHER London` 3. **Server fetches weather data for London from the API.** 4. **Server sends:** `WEATHER London 15 Cloudy 80\n` (if the temperature is 15 degrees, the condition is cloudy, and the humidity is 80%) 5. **Client (Cursor) receives:** `WEATHER London 15 Cloudy 80` 6. **Client (Cursor) displays:** "Weather in London: 15 degrees, Cloudy, Humidity: 80%" This detailed explanation and the Python example should give you a solid foundation for building your weather MCP server. Remember to adapt the code and the MCP format to your specific needs and the capabilities of Cursor. Good luck!

applemail-mcp-server

A local MCP server that lets Claude read macOS Mail.app via JXA, with tools for listing accounts/mailboxes, searching and reading messages. Read-only by default, with optional opt-in write tools for compose drafts and Apple Calendar events.

jira-minimal-mcp

Minimal MCP server for Jira with configurable tools to reduce token usage, starting from ~150 tokens for basic issue retrieval.

MCP-Odoo

A bridge that allows AI agents to access and manipulate Odoo ERP data through a standardized Model Context Protocol interface, supporting partner information, accounting data, financial records reconciliation, and invoice queries.

Google Workspace MCP Server

Enables management of Google Workspace apps (Docs, Sheets, Gmail, Calendar, Drive) from the command line via Gemini CLI.

Subwatch MCP

Enables reading public Reddit posts and comments on demand, with tools to search, get recent posts, post details, top comments, and server status. Runs on Cloudflare Workers for use with Claude and Open WebUI.

demandscope

MCP server that provides demand-signal scanning tools from GitHub, Hacker News, npm, and PyPI to validate digital-product and dev-tool ideas. Enables agents to query repo signals, trends, and download counts via stdio JSON-RPC.

flstudio-mcp-mac

Enables controlling FL Studio on macOS via MCP, including transport, mixer, channel, MIDI export, and Piano Roll note insertion.

Confluence MCP Server

Enables integration with Atlassian Confluence to browse spaces, search content using CQL, and manage pages directly from MCP-compatible applications. It automatically converts Confluence storage formats into markdown for seamless interaction with AI-driven editors and tools.

musfiraai-mcp-server

Exposes Musfiraai's company info, services, AI stack, FAQ, reviews, and contact details as MCP tools/resources/prompts, enabling any MCP-compatible AI client to query this data via natural language.

Score de Crédito

Enables credit score and risk analysis queries for Brazilian individuals (CPF) and companies (CNPJ) via MCP over HTTP, with a single read-only tool and prepaid usage.

VA-MCP

An MCP server for checking OWASP Top 10 vulnerabilities during API development testing. It analyzes API information and returns security assessment results to help developers identify potential security issues.

mcp-cli-catalog

An MCP server that publishes CLI tools on your machine for discoverability by LLMs

minesweeper-mcp

A stdio-based MCP server that enables users to play and manage Minesweeper games through a Rails 8 REST API. It provides tools to start games, track states, and perform actions like opening cells and flagging mines.

Dynatrace MCP Server

Enables interaction with Dynatrace Grail and API endpoints by executing DQL queries, retrieving logs, reporting active problems, and querying metrics.

context-ledger

Provides local, explicitly scoped memory for coding agents via MCP, storing durable project knowledge in a per-repository SQLite database with tools to record, search, and retrieve context.

samskriti-project

A local MCP server that enables multiple AI coding tools to share structured project state (decisions, tasks, bugs) so they coordinate without re-explaining.

CRM Agent Tools

Provides Claude agents with CRM contact lookup, action logging, and prompt cost auditing tools. Includes a Streamlit UI to demo the same tools without an MCP client.

Hyperion V2

Universal MCP server that enables any LLM agent (Claude, Cursor, Cline) to control a real Chrome browser with 5 perception engines, resilient heartbeat, and real-time vision streaming.

Gopher & Gemini MCP Server

Enables AI assistants to browse and interact with both Gopher and Gemini protocol resources safely and efficiently.

Image Process MCP Server

Um servidor MCP para processamento de imagens que utiliza a biblioteca Sharp para fornecer funcionalidades de manipulação de imagem.

pixso-mcp-server

Brings Pixso design context into AI coding workflows, enabling AI assistants to understand design layers, component variants, variables, and local styles for design-to-code tasks.

TickDB MCP

Unified real-time & historical market data API for Forex, stocks (US/HK/A-share), crypto, indices & precious metals. Tick, order book depth & K-line via REST + WebSocket. AI-native: MCP server, Skill & CLI.

herald

An MCP server that allows AI assistants to send formatted text, files, and updates to Telegram topics, and to capture incoming messages from work chats into a local inbox for retrieval.

withings-mcp

Read-only Model Context Protocol server for Withings weight and body-composition history, providing normalized measurements, weight history, and latest measurements via OAuth 2.0.

React USWDS MCP Server

Indexes the locally-installed @trussworks/react-uswds package and helps code assistants discover components, inspect props, generate correct imports and usage snippets, and suggest appropriate components for UI use cases.

temporal-mcp-server

Enables AI assistants to observe and manage Temporal workflows, including listing, starting, signaling, resetting, terminating, querying, and retrieving history, with support for wait_for_activity and webhook utilities.