发现优秀的 MCP 服务器

Llama 4 Maverick MCP Server

Bridges Llama models with Claude Desktop through Ollama, enabling privacy-first local AI operations with 10+ built-in tools for file operations, web search, calculations, and custom model deployment. Features streaming support, hybrid intelligence workflows, and extensive Python ecosystem integration for research, development, and enterprise applications.

Earth Class Mail MCP

Access and manage your Earth Class Mail virtual mailbox, including viewing mail pieces, requesting actions like scan or shred, and listing inboxes.

Sigma MCP Server

An MCP server that provides Sigma rule validation and configuration capabilities for AI assistants. It enables users to validate Sigma detection rules against various validators and manage validator configurations through MCP tools and resources.

Excel Reader Server

镜子 (jìng zi)

Neon MCP

Connects your Neon account to Claude, ChatGPT, and AI agents via Open Finance Brazil, enabling natural language queries about balances, statements, credit card bills, and investments — read-only and regulated by the Central Bank.

Construction ISO 19650 (UK BIM) MCP

Enables ISO 19650 and UK BIM Level 2 compliance for construction projects, validating EIR, BEP, MIDP, TIDP, and CDE structures.

Jason-MCP

Enables natural language interaction with JetBrains YouTrack, including issue management, sprint operations, team reporting, name-based assignments, and notification checks.

Gemini MCP Ultimate

Enables Claude to offload large codebase analysis to Google Gemini with 1M+ token context windows, persistent sessions, and dramatic token savings.

Spotify-Agent

Okay, this is a complex project, and I can't provide you with a complete, ready-to-run MCP (Minecraft Protocol) server implementation here. Building a full MCP server with Spotify, Last.fm, and internet integration for music recommendations requires significant coding effort and knowledge of several technologies. However, I can give you a detailed outline, code snippets (in Python, as it's commonly used for these tasks), and guidance to get you started. **I. Project Overview** The goal is to create an MCP server plugin (or a standalone server that interacts with a Minecraft server) that: 1. **Connects to Spotify and Last.fm APIs:** Retrieves user listening history, track information, artist information, and similar artist data. 2. **Interacts with the Internet (e.g., for Lyrics):** Fetches lyrics for songs. 3. **Stores and Processes Music Data:** Creates a database or data structure to store user preferences, track information, and recommendation data. 4. **Provides Minecraft Commands:** Allows players to request recommendations, view their listening history, and interact with the music system within the game. 5. **Delivers Recommendations:** Generates music recommendations based on user data and similarity algorithms. **II. Technology Stack** * **Minecraft Server:** You'll need a Minecraft server (e.g., Spigot, Paper, Fabric). Choose one you're comfortable with. * **MCP Server Library/Framework:** Choose a library that simplifies MCP server development. Popular options include: * **Python:** `mcstatus`, `nbt`, `minecraft-protocol` (for lower-level control) * **Java:** Spigot API, Paper API (if you're building a plugin) * **Programming Language:** Python or Java are good choices. Python is often preferred for scripting and API interactions. * **Spotify API:** Use the Spotify Web API (requires a Spotify Developer account). * **Last.fm API:** Use the Last.fm API (requires a Last.fm API key). * **Lyrics API (Optional):** Consider using a lyrics API like Lyrics.ovh or Musixmatch API. * **Database (Optional):** Consider using a database (e.g., SQLite, MySQL, PostgreSQL) to store user data and music information. SQLite is a good choice for smaller projects. * **Recommendation Algorithm:** Implement a recommendation algorithm (e.g., collaborative filtering, content-based filtering, hybrid approach). **III. Detailed Steps and Code Snippets (Python Example)** **1. Setting up the MCP Server (Python)** ```python import socket import struct import json import threading # Configuration HOST = 'localhost' PORT = 25565 # Default Minecraft port def handle_client(conn, addr): print(f"Connected by {addr}") try: # Handshake data = conn.recv(256) packet_length = struct.unpack('>i', data[:4])[0] packet_id = struct.unpack('>b', data[4:5])[0] if packet_id == 0x00: # Handshake packet handshake_data = json.loads(data[5:].decode('utf-8')) print(f"Handshake data: {handshake_data}") # Status Request data = conn.recv(256) packet_length = struct.unpack('>i', data[:4])[0] packet_id = struct.unpack('>b', data[4:5])[0] if packet_id == 0x00: # Status Request # Respond with server status status = { "version": { "name": "My Music Server", "protocol": 754 # Example protocol version }, "players": { "max": 100, "online": 0, "sample": [] }, "description": { "text": "A server for music recommendations!" } } status_json = json.dumps(status) status_bytes = status_json.encode('utf-8') status_length = len(status_bytes) response = struct.pack('>i', status_length + 1) + struct.pack('>b', 0x00) + status_bytes conn.sendall(response) # Ping Request data = conn.recv(256) packet_length = struct.unpack('>i', data[:4])[0] packet_id = struct.unpack('>b', data[4:5])[0] if packet_id == 0x01: # Ping Request ping_payload = data[5:] response = struct.pack('>i', len(ping_payload) + 1) + struct.pack('>b', 0x01) + ping_payload conn.sendall(response) except Exception as e: print(f"Error handling client: {e}") finally: conn.close() print(f"Connection closed with {addr}") def start_server(): server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) # Allow reuse of the address server_socket.bind((HOST, PORT)) server_socket.listen(5) print(f"Server listening on {HOST}:{PORT}") while True: conn, addr = server_socket.accept() client_thread = threading.Thread(target=handle_client, args=(conn, addr)) client_thread.start() if __name__ == "__main__": start_server() ``` **Explanation:** * **`socket`:** Creates a socket for network communication. * **`struct`:** Packs and unpacks data in binary format (required for the Minecraft protocol). * **`json`:** Handles JSON data for the server status. * **`threading`:** Allows handling multiple client connections concurrently. * **`handle_client`:** This function handles the communication with a single Minecraft client. It receives the handshake, status request, and ping request, and sends back the appropriate responses. * **`start_server`:** Sets up the server socket and listens for incoming connections. **Important:** This is a *very* basic MCP server. It only handles the handshake, status request, and ping request. You'll need to implement the full Minecraft protocol to handle player login, chat messages, commands, and other game events. **2. Integrating with Spotify and Last.fm APIs** ```python import spotipy from spotipy.oauth2 import SpotifyClientCredentials import pylast # Spotify API Credentials (replace with your own) SPOTIPY_CLIENT_ID = "YOUR_SPOTIFY_CLIENT_ID" SPOTIPY_CLIENT_SECRET = "YOUR_SPOTIFY_CLIENT_SECRET" # Last.fm API Credentials (replace with your own) LASTFM_API_KEY = "YOUR_LASTFM_API_KEY" LASTFM_API_SECRET = "YOUR_LASTFM_API_SECRET" # Initialize Spotify client client_credentials_manager = SpotifyClientCredentials(client_id=SPOTIPY_CLIENT_ID, client_secret=SPOTIPY_CLIENT_SECRET) sp = spotipy.Spotify(client_credentials_manager=client_credentials_manager) # Initialize Last.fm network network = pylast.LastFMNetwork( api_key=LASTFM_API_KEY, api_secret=LASTFM_API_SECRET, username="YOUR_LASTFM_USERNAME" # Replace with a Last.fm username for testing ) def get_spotify_track_info(track_name, artist_name): """Searches Spotify for a track and returns track information.""" query = f"track:{track_name} artist:{artist_name}" results = sp.search(q=query, type='track', limit=1) if results['tracks']['items']: track = results['tracks']['items'][0] return { 'id': track['id'], 'name': track['name'], 'artist': track['artists'][0]['name'], 'album': track['album']['name'], 'popularity': track['popularity'], 'uri': track['uri'] } else: return None def get_lastfm_recent_tracks(username): """Gets the recent tracks played by a Last.fm user.""" user = network.get_user(username) recent_tracks = user.get_recent_tracks(limit=10) # Get the last 10 tracks tracks = [] for item in recent_tracks: track = item.track tracks.append({ 'artist': track.artist.name, 'title': track.title }) return tracks # Example usage if __name__ == "__main__": # Spotify Example track_info = get_spotify_track_info("Bohemian Rhapsody", "Queen") if track_info: print("Spotify Track Info:", track_info) else: print("Track not found on Spotify.") # Last.fm Example recent_tracks = get_lastfm_recent_tracks("YOUR_LASTFM_USERNAME") # Replace with your Last.fm username print("\nLast.fm Recent Tracks:") for track in recent_tracks: print(f"- {track['artist']} - {track['title']}") ``` **Explanation:** * **`spotipy`:** The Spotify API library. You'll need to install it: `pip install spotipy` * **`pylast`:** The Last.fm API library. You'll need to install it: `pip install pylast` * **API Credentials:** You *must* replace the placeholder credentials with your own API keys and secrets from the Spotify Developer Dashboard and the Last.fm API page. * **`get_spotify_track_info`:** Searches Spotify for a track and returns information like ID, name, artist, album, and popularity. * **`get_lastfm_recent_tracks`:** Gets the recent tracks played by a Last.fm user. **3. Handling Minecraft Commands** You'll need to integrate your music functionality with Minecraft commands. The specific way to do this depends on the MCP server library you're using. Here's a general outline: * **Command Registration:** Register a new command (e.g., `/musicrecommend`, `/mylistenhistory`). * **Command Parsing:** Parse the arguments provided with the command. * **Command Execution:** Execute the appropriate music-related function based on the command and arguments. * **Response to Player:** Send a message back to the player in the Minecraft chat window with the results (e.g., the recommended songs, their listening history). **Example (Conceptual - using a hypothetical MCP library):** ```python # Hypothetical MCP library functions def register_command(command_name, command_handler): # Registers a command with the server pass def send_message_to_player(player_name, message): # Sends a message to a player in the game pass def handle_music_recommend_command(player_name, args): # Get the player's Last.fm username (from a database or configuration) lastfm_username = get_player_lastfm_username(player_name) if not lastfm_username: send_message_to_player(player_name, "Please link your Last.fm account first.") return # Get the player's recent tracks from Last.fm recent_tracks = get_lastfm_recent_tracks(lastfm_username) # Generate recommendations (implementation details omitted) recommendations = generate_recommendations(recent_tracks) # Format the recommendations into a message message = "Recommended Songs:\n" for track in recommendations: message += f"- {track['artist']} - {track['title']}\n" # Send the recommendations to the player send_message_to_player(player_name, message) def get_player_lastfm_username(player_name): # This function would retrieve the Last.fm username associated with the player # from a database or configuration file. # For example: # player_data = database.get_player_data(player_name) # return player_data.get('lastfm_username') return "YOUR_LASTFM_USERNAME" # Replace with your Last.fm username for testing # Register the command register_command("musicrecommend", handle_music_recommend_command) ``` **4. Recommendation Algorithm** This is the most complex part. Here are some approaches: * **Collaborative Filtering:** Find users with similar listening habits and recommend songs they like that the current user hasn't heard. Requires a database of user listening data. * **Content-Based Filtering:** Analyze the characteristics of the songs the user likes (e.g., genre, tempo, mood) and recommend songs with similar characteristics. You can use the Spotify API to get track features. * **Hybrid Approach:** Combine collaborative and content-based filtering for better results. **Example (Simplified Content-Based Filtering):** ```python def generate_recommendations(recent_tracks): """Generates recommendations based on recent tracks.""" recommendations = [] for track in recent_tracks: # Get Spotify track info track_info = get_spotify_track_info(track['title'], track['artist']) if track_info: # Get similar artists from Spotify artist_id = sp.artist(track_info['uri'].split(':')[2])['artists'][0]['id'] related_artists = sp.artist_related_artists(artist_id)['artists'] # Get top tracks from related artists for artist in related_artists[:3]: # Limit to 3 related artists top_tracks = sp.artist_top_tracks(artist['id'])['tracks'][:2] # Limit to 2 tracks for top_track in top_tracks: recommendations.append({ 'artist': top_track['artists'][0]['name'], 'title': top_track['name'] }) return recommendations ``` **5. Database (Optional)** If you want to store user data (e.g., Last.fm usernames, listening history), you'll need a database. SQLite is a good choice for smaller projects. ```python import sqlite3 # Database setup conn = sqlite3.connect('music_server.db') cursor = conn.cursor() # Create a table to store player data cursor.execute(''' CREATE TABLE IF NOT EXISTS players ( player_name TEXT PRIMARY KEY, lastfm_username TEXT ) ''') conn.commit() def set_player_lastfm_username(player_name, lastfm_username): """Sets the Last.fm username for a player in the database.""" cursor.execute("INSERT OR REPLACE INTO players (player_name, lastfm_username) VALUES (?, ?)", (player_name, lastfm_username)) conn.commit() def get_player_lastfm_username(player_name): """Gets the Last.fm username for a player from the database.""" cursor.execute("SELECT lastfm_username FROM players WHERE player_name = ?", (player_name,)) result = cursor.fetchone() if result: return result[0] else: return None # Example usage if __name__ == "__main__": set_player_lastfm_username("Player123", "YOUR_LASTFM_USERNAME") # Replace with your Last.fm username username = get_player_lastfm_username("Player123") print(f"Last.fm username for Player123: {username}") conn.close() ``` **IV. Key Considerations and Challenges** * **Minecraft Protocol:** Understanding the Minecraft protocol is crucial for building a functional MCP server. It's complex and constantly evolving. * **API Rate Limits:** The Spotify and Last.fm APIs have rate limits. You need to handle these limits gracefully to avoid being blocked. Implement caching and throttling. * **User Authentication:** You'll need a way for players to link their Spotify/Last.fm accounts to their Minecraft accounts. This typically involves OAuth 2.0. * **Data Storage:** Choosing the right data storage solution (database, file system, etc.) depends on the scale of your project. * **Performance:** Generating recommendations can be computationally expensive. Optimize your code and consider using background threads or asynchronous tasks. * **Error Handling:** Implement robust error handling to deal with API errors, network issues, and invalid user input. * **Security:** Protect API keys and user data. Don't store sensitive information in plain text. * **Scalability:** If you plan to support a large number of users, you'll need to design your system to be scalable. **V. Next Steps** 1. **Choose an MCP Server Library:** Select a library that suits your needs and programming language preference. 2. **Implement the Basic MCP Server:** Get a basic MCP server running that can handle the handshake, status request, and ping request. 3. **Integrate with the Spotify and Last.fm APIs:** Implement the functions to retrieve user listening history and track information. 4. **Implement Command Handling:** Add the ability to register and handle Minecraft commands. 5. **Implement a Recommendation Algorithm:** Start with a simple algorithm and gradually improve it. 6. **Add a Database (Optional):** If you need to store user data, set up a database. 7. **Test Thoroughly:** Test your server with multiple Minecraft clients and different scenarios. **VI. Chinese Translation of Key Terms** Here's a translation of some key terms that might be helpful: * **MCP Server:** MCP服务器 (MCP fúwùqì) * **Spotify API:** Spotify API (Spotify API) (Note: API is often used directly in Chinese) * **Last.fm API:** Last.fm API (Last.fm API) * **Recommendation Algorithm:** 推荐算法 (tuījiàn suànfǎ) * **Minecraft Command:** Minecraft 命令 (Minecraft mìnglìng) * **Listening History:** 听歌历史 (tīng gē lìshǐ) * **API Key:** API 密钥 (API mìyào) * **Database:** 数据库 (shùjùkù) * **User Authentication:** 用户认证 (yònghù rènzhèng) * **Rate Limit:** 速率限制 (sùlǜ xiànzhì) or 频率限制 (pínlǜ xiànzhì) * **Collaborative Filtering:** 协同过滤 (xiétóng guòlǜ) * **Content-Based Filtering:** 基于内容的过滤 (jī yú nèiróng de guòlǜ) This is a substantial project, but by breaking it down into smaller steps and using the resources and guidance provided, you can make progress towards building your music recommendation MCP server. Good luck!

Calendar MCP Server

A unified Model Context Protocol server for managing calendars across Google Calendar, Microsoft 365, and Exchange On-Premises. It provides tools for scheduling, conflict detection, and event management across multiple provider accounts using a consistent data format.

weather-mcp

MCP server that provides weather forecast tools including current conditions, daily forecasts, and day-specific recommendations like whether to carry an umbrella or wear a jacket, powered by Open-Meteo.

GitLab Code Review MCP Server

Enables AI assistants to interact with GitLab merge requests, review code changes, search projects, and create draft comments for code review workflows.

Scaleway MCP Server

A comprehensive MCP server for managing Scaleway cloud infrastructure, enabling control of instances, databases, storage, networking, and more through natural language.

Motion MCP Server

Bridges Motion's AI-powered calendar and task management API with LLMs via the Model Context Protocol, enabling natural language management of tasks, projects, schedules, comments, and more.

kalundborg-mcp

MCP server for Kalundborg Kommune, enabling search of content, news, popular pages, contact info, and services from kalundborg.dk.

Weather MCP Server

A Model Context Protocol (MCP) server that provides real-time weather alerts and forecasts from the US National Weather Service.

Snackbar MCP Server

A personal note store exposed as an MCP server. Enables any MCP-speaking assistant to create, search, list, and categorize notes, with per-client bearer tokens for author attribution.

ForIT Microsoft Graph

Provides direct access to Microsoft Graph API with multi-tenant account management, allowing users to interact with Microsoft 365 services across multiple tenants through a single flexible graph-request tool.

2d-assets-mcp

Generates mock 2D PNG assets for game prototypes, supporting gradients, patterns, text, spritesheets, and embedded metadata.

Sensory-Grounding MCP

Gives AI coding agents a closed-loop verification cycle for visual, audio, and video output, with enforcement hooks that make verification mandatory.

AQICN MCP Server

通过模型上下文协议 (MCP) 与世界空气质量指数 (World Air Quality Index) 交互，以获取全球城市和坐标的实时空气质量数据。

SAE4U Memory

Provides persistent memory for Claude with hierarchical categorization, cross-corpus recall, session journals, and customizable persona, enabling memory continuity across sessions.

repo-health

Provides read-only repository health scanning tools for drift detection, module reachability, prompt bloat, evidence calibration, and registration completeness, enabling agents to diagnose repositories via MCP.

notebooklm-api

Enables document Q&A with grounded citations using the official Gemini File Search API. Upload documents, create stores, and ask questions that return exact passage citations.

kairo

Autonomously analyzes projects, generates Playwright tests via OpenAI, runs them, classifies failures, and recommends fixes, iterating up to 3 times.

Remote MCP Server (Authless)

Enables deployment of MCP servers without authentication on Cloudflare Workers, allowing tools to be accessed remotely from clients like Claude Desktop or Cloudflare AI Playground via Server-Sent Events.

IssueBadge MCP Server

A Model Context Protocol server that enables AI assistants to create, issue, and manage digital badges and certificates using natural language.

talos-mcp-server

MCP server for Talos Linux cluster operations, providing typed actions for cluster configuration, bootstrap, kubeconfig, and more via talosctl.

Prem MCP Server

一个模型上下文协议（Model Context Protocol, MCP）服务器的实现，它能够与 Claude 和其他兼容 MCP 的客户端无缝集成，从而访问 Prem AI 的语言模型、检索增强生成（Retrieval-Augmented Generation, RAG）能力以及文档管理功能。