发现优秀的 MCP 服务器

mail-mcp

An Apple Mail MCP server that enables reading emails, listing mailboxes, searching messages, handling attachments, and creating drafts, with no ability to send mail.

starghlist

MCP server to manage your GitHub starred repositories and lists

MCP Server with Local LLM

Integrates local language models (like Qwen3-8B) with MCP clients, providing tools for chat, code analysis, text generation, translation, and content summarization using your own hardware.

elvanto-mcp

Self-hosted MCP server for Elvanto church management. Enables querying upcoming rosters, service songs, and submitting unavailability via Claude or ChatGPT.

Notion MCP Server for Antigravity

Enables AI models to search and read Notion pages by keyword or ID, allowing users to retrieve workspace content like meeting notes through natural language. It supports integration with Antigravity and Gemini to provide seamless access to Notion document data.

memvid-mcp

An MCP server that provides a persistent memory layer for AI agents by wrapping the memvid CLI. It enables users to manage memory files, perform hybrid searches, and conduct RAG-based question answering through 40 specialized tools.

Querit MCP Server

Connects AI assistants to Querit for live web search and clean webpage content retrieval, supporting both anonymous access and bring-your-own-key authentication.

Tally MCP

Manage Tally.so forms through AI conversation. Create, update, and organize forms using natural language instead of web interfaces. Features safe bulk operations and works with all major MCP clients.

Obsidian Index MCP Server

提供对 Obsidian 库的语义搜索能力，并通过 MCP 协议将最近的笔记作为资源暴露给 Claude。

White Hat Thumbnails

MCP server that lets AI assistants manage Roblox thumbnails end to end, including performance tracking, generation, publishing, auto-rotation, and Discord team notifications.

Steddion Energy MCP Server

Provides tools for Dutch energy market data: day-ahead and imbalance prices, weather forecasts, and battery storage business case calculations.

web3-jobs-mcp

web3-jobs-mcp

China Stock MCP Server

一个多重调用协议服务器，通过统一的API提供对中国股票市场数据的全面访问，并利用AKShare从各个中国交易所获取实时和历史数据。

3CX MCP Server

An MCP server that connects Claude to a 3CX Phone System (V20+), enabling user management, call monitoring, contact search, and forwarding configuration.

Firebase DAM MCP Server

Enables secure, read-only access to Firebase Firestore collections and Storage buckets for Digital Asset Management systems. It allows users to query assets, versions, and comments, and search storage files using flexible filters through standard MCP tools.

VAT Validation MCP Server

Enables real-time validation of EU VAT numbers using the official VIES service. Supports all 27 EU member states with automatic country code detection and provides company information for valid VAT numbers.

ask262

MCP server for exploring the ECMAScript specification and its implementation, enabling semantic search of spec sections, JavaScript execution in engine262, and access to a knowledge graph linking spec sections to implementation functions.

FastMCP 🚀

构建模型上下文协议服务器的快速、Pythonic 方法 🚀

figma-mcp-write-server

An MCP server that provides write access to Figma through the Plugin API, enabling AI agents to create, modify, and manage Figma designs programmatically.

Shellagent MCP Server

A Model Context Protocol (MCP) server that can be deployed locally via stdio or remotely via SSE/HTTP endpoints, supporting multiple MCP clients including VS Code, Cursor, Windsurf, and Claude Desktop.

wikimemory

Personal, remotely hosted memory service for Claude, Codex, and other MCP clients that preserves research, project state, and decisions in an auditable revision store with a web UI.

GitHub MCP Server Practice Repository

镜子 (jìng zi)

planist-mcp

Enables AI clients to inspect, read, and propose page edits in Planist workspaces over MCP. Supports listing pages, fetching structured content, and submitting edit proposals for review.

Process Mining MCP Server

A tool that integrates process mining capabilities with Claude AI, allowing users to analyze event log data from PostgreSQL databases through natural language interfaces in Claude Desktop or Cursor.

gw1-mcp

An MCP server that gives LLM clients reliable, deterministic knowledge of Guild Wars 1 builds: skill data lookup, official template code encoding/decoding, and build validation.

ModelContextProtocol (MCP) Java SDK v0.8.0 Specification

Okay, here's a breakdown of instructions for an AI on how to create a Java-based MCP (Minecraft Coder Pack) server and client. This is a complex task, so the instructions are broken down into manageable steps. The AI will need to understand Java, networking concepts (sockets, TCP), and the basics of Minecraft's internal structure (though we'll abstract away the really complex parts). **High-Level Goal:** The AI should create a simple Java application consisting of two parts: 1. **MCP Server:** A server application that listens for connections from MCP clients. It will receive commands from the client, process them (in a very basic way), and send responses back. For simplicity, the server will primarily focus on handling commands related to Minecraft block and item IDs. 2. **MCP Client:** A client application that connects to the MCP server, sends commands (e.g., "get block ID for dirt", "get item name for 260"), and displays the server's responses. **Phase 1: Project Setup and Basic Networking (Foundation)** 1. **Project Creation:** * Create a new Java project in a suitable IDE (IntelliJ IDEA, Eclipse, etc.). * Create two main packages: `server` and `client`. * Within each package, create a main class: `server.MCPServer` and `client.MCPClient`. 2. **Basic Server Socket Setup (server.MCPServer):** * **Import necessary classes:** `java.net.ServerSocket`, `java.net.Socket`, `java.io.BufferedReader`, `java.io.PrintWriter`, `java.io.InputStreamReader`. * **Create a `ServerSocket`:** The server should listen on a specific port (e.g., 12345). Handle `IOException` appropriately (try-catch). * **Accept Connections:** Use `serverSocket.accept()` to listen for incoming client connections. This will return a `Socket` object representing the connection to the client. * **Input/Output Streams:** For each connected client: * Create a `BufferedReader` to read data from the client (using `InputStreamReader` wrapped around the `Socket`'s input stream). * Create a `PrintWriter` to send data back to the client (using the `Socket`'s output stream). Set `autoFlush` to `true` for immediate sending. * **Basic Echo:** For now, the server should simply read a line of text from the client and send it back (echo). This verifies basic connectivity. * **Multi-threading (Important):** The server needs to handle multiple clients concurrently. Create a new `Thread` for each client connection. The `Runnable` for the thread should contain the input/output stream handling and the echo logic. This prevents one client from blocking the entire server. * **Error Handling:** Implement `try-catch` blocks to handle potential `IOExceptions` during socket operations. Log errors to the console. * **Resource Cleanup:** Ensure that sockets, input streams, and output streams are closed properly in a `finally` block to prevent resource leaks. 3. **Basic Client Socket Setup (client.MCPClient):** * **Import necessary classes:** `java.net.Socket`, `java.io.BufferedReader`, `java.io.PrintWriter`, `java.io.InputStreamReader`. * **Create a `Socket`:** Connect to the server's IP address and port (e.g., "localhost", 12345). Handle `IOException`. * **Input/Output Streams:** Similar to the server, create a `BufferedReader` and `PrintWriter` for communication. * **Send a Message:** Send a simple message to the server (e.g., "Hello from the client!"). * **Receive Response:** Read the server's response and print it to the console. * **Resource Cleanup:** Close the socket, input stream, and output stream in a `finally` block. 4. **Testing:** * Run the server. * Run the client. * Verify that the client connects to the server, sends a message, and receives the echoed response. **Phase 2: Command Handling and Data (Core Functionality)** 1. **Command Protocol:** * Define a simple command protocol. Commands will be strings. Examples: * `GET_BLOCK_ID <block_name>` (e.g., `GET_BLOCK_ID dirt`) * `GET_ITEM_NAME <item_id>` (e.g., `GET_ITEM_NAME 260`) * `LIST_BLOCKS` (Lists all known blocks) * `LIST_ITEMS` (Lists all known items) * `EXIT` (Client disconnects) 2. **Server-Side Command Parsing (server.MCPServer):** * In the server's client-handling thread, read the command from the client. * Use `String.split()` or regular expressions to parse the command and its arguments. * Implement a `switch` statement or `if-else` chain to handle different commands. 3. **Data Storage (Server-Side):** * Create a simple data structure to store block and item information. A `HashMap` is suitable: * `HashMap<String, Integer> blockNameToId`: Maps block names (String) to IDs (Integer). * `HashMap<Integer, String> itemIdToName`: Maps item IDs (Integer) to names (String). * **Populate with Sample Data:** Add a few entries to the `HashMap`s for testing. Example: * `blockNameToId.put("dirt", 3);` * `blockNameToId.put("stone", 1);` * `itemIdToName.put(260, "apple");` * `itemIdToName.put(276, "diamond_sword");` 4. **Command Implementation (Server-Side):** * **`GET_BLOCK_ID <block_name>`:** * Look up the block name in `blockNameToId`. * If found, send the ID back to the client. * If not found, send an error message (e.g., "Block not found"). * **`GET_ITEM_NAME <item_id>`:** * Parse the item ID as an integer. * Look up the ID in `itemIdToName`. * If found, send the name back to the client. * If not found, send an error message (e.g., "Item not found"). * **`LIST_BLOCKS`:** * Iterate through the `blockNameToId` map and send each block name and ID to the client, separated by a delimiter (e.g., "dirt:3\nstone:1\n"). * **`LIST_ITEMS`:** * Iterate through the `itemIdToName` map and send each item ID and name to the client, separated by a delimiter (e.g., "260:apple\n276:diamond_sword\n"). * **`EXIT`:** * Close the client socket and terminate the client-handling thread. Send a confirmation message to the client before closing. 5. **Client-Side Command Input and Output (client.MCPClient):** * Prompt the user to enter a command. * Send the command to the server. * Read the server's response and display it to the user. * Implement a loop that continues until the user enters the `EXIT` command. 6. **Testing:** * Run the server. * Run the client. * Test each command to ensure it works correctly. Test error cases (e.g., requesting a non-existent block). **Phase 3: Refinement and Error Handling (Polishing)** 1. **Error Handling:** * **Server-Side:** Add more robust error handling to the server. Catch potential `NumberFormatExceptions` when parsing item IDs. Log errors to a file or the console. Send informative error messages to the client. * **Client-Side:** Handle potential `IOExceptions` when reading from or writing to the socket. Display user-friendly error messages. 2. **Input Validation:** * **Server-Side:** Validate the input from the client to prevent potential security vulnerabilities (e.g., prevent command injection). Sanitize input before using it in lookups. * **Client-Side:** Provide basic input validation to the user (e.g., ensure that the item ID is a number). 3. **Code Style and Readability:** * Use meaningful variable names. * Add comments to explain the code. * Format the code consistently. * Break down long methods into smaller, more manageable methods. 4. **Configuration:** * Allow the server port to be configured via a command-line argument or a configuration file. * Allow the client to specify the server IP address and port via command-line arguments. 5. **Logging:** * Implement basic logging to record server activity and errors. Use a logging framework like `java.util.logging` or Log4j. **Example Code Snippets (Illustrative - Not Complete):** **Server (Simplified):** ```java // Inside the client-handling thread's Runnable String command = reader.readLine(); if (command != null) { String[] parts = command.split(" "); String action = parts[0]; switch (action) { case "GET_BLOCK_ID": if (parts.length > 1) { String blockName = parts[1]; Integer blockId = blockNameToId.get(blockName); if (blockId != null) { writer.println("Block ID: " + blockId); } else { writer.println("Block not found."); } } else { writer.println("Invalid command format."); } break; case "EXIT": writer.println("Goodbye!"); socket.close(); return; // Exit the thread's run() method default: writer.println("Unknown command."); } } ``` **Client (Simplified):** ```java Scanner scanner = new Scanner(System.in); String command; while (true) { System.out.print("Enter command: "); command = scanner.nextLine(); writer.println(command); writer.flush(); // Ensure the command is sent immediately String response = reader.readLine(); System.out.println("Server response: " + response); if (command.equals("EXIT")) { break; } } ``` **Important Considerations for the AI:** * **Security:** This is a simplified example. In a real-world application, security would be a major concern. The AI should be aware of potential vulnerabilities (e.g., command injection, denial-of-service attacks) and take steps to mitigate them. However, for this exercise, focus on the core functionality first. * **Scalability:** This example is not designed for high scalability. For a large number of clients, more advanced techniques (e.g., asynchronous I/O, thread pools) would be necessary. * **Minecraft Data:** The AI will need access to a data source for Minecraft block and item IDs. This could be a simple text file, a JSON file, or a database. For this exercise, a hardcoded `HashMap` is sufficient for testing. The AI should be able to load data from a file if instructed. * **Error Handling:** Robust error handling is crucial. The AI should anticipate potential errors and handle them gracefully. * **Modularity:** The AI should strive to create modular code that is easy to understand and maintain. **AI Task Breakdown:** The AI should perform the following tasks: 1. **Code Generation:** Generate the Java code for the server and client applications, following the instructions above. 2. **Data Population:** Populate the `HashMap`s with sample Minecraft block and item data. 3. **Testing:** Provide instructions on how to test the application. 4. **Documentation:** Generate basic documentation for the code. This is a substantial project. Start with Phase 1 and get the basic networking working before moving on to Phase 2 and Phase 3. Good luck!

Coach AI

A productivity coaching assistant designed for ADHD support that provides intelligent task management, goal tracking, and personalized recommendations. It utilizes a persistent memory system to learn user patterns and preferences, helping to reduce decision paralysis through actionable suggestions.

tableau-graphql-mcp

Enables querying Tableau lineage through the Metadata API via MCP, supporting impact analysis, where-used lookups, and content search. Provides read-only tools for answering natural-language questions about workbooks, datasources, fields, and dependencies.

pyFit

Enables 2D irregular polygon nesting (bin-packing) with tools to design, preview, get reports, and export DXF files for laser cutting or CNC routing.