发现优秀的 MCP 服务器

ServiceNow CMDB MCP Server

A read-only MCP server that enables searching and listing ServiceNow configuration items, reading direct CI relationships, and traversing the relationship graph up to a controlled depth.

FarmSubsidy Korea MCP

Aggregates Korean agricultural subsidy announcements from multiple government sources. Enables searching, detailed viewing, and calendar integration for subsidy deadlines.

MCPClient Python Application

Okay, I understand. You want an implementation that allows an MCP (presumably "Minecraft Protocol") server to interact with an Ollama model. This is a complex task that involves several steps. Here's a breakdown of the concepts, potential approaches, and a *conceptual* implementation outline. Keep in mind that this is a high-level overview, and a complete, working solution would require significant coding effort. **Understanding the Components** * **MCP Server (Minecraft Protocol Server):** This is the server that handles Minecraft client connections, game logic, and world management. We need to be able to intercept or inject messages into this server. This likely requires a Minecraft server mod (e.g., using Fabric, Forge, or a custom server implementation). * **Ollama Model:** This is a large language model (LLM) served by Ollama. We need to be able to send text prompts to the Ollama API and receive text responses. * **Interaction:** The core of the problem is *how* the MCP server and Ollama model will interact. Here are some possibilities: * **Chatbot:** Players type commands or messages in the Minecraft chat, which are then sent to the Ollama model. The model's response is displayed back in the chat. * **NPC Dialogue:** Non-player characters (NPCs) in the game have dialogue powered by the Ollama model. The model generates responses based on player interactions or game events. * **World Generation/Modification:** The Ollama model could be used to generate descriptions of terrain, structures, or quests, which are then used to modify the Minecraft world. * **Game Logic:** The model could be used to make decisions for AI entities or influence game events based on player actions. **Conceptual Implementation Outline** This outline focuses on the "Chatbot" interaction, as it's the most straightforward to explain. 1. **Minecraft Server Mod (e.g., Fabric/Forge):** * **Dependency:** Add the necessary dependencies for your chosen mod loader (Fabric or Forge). * **Event Listener:** Create an event listener that intercepts chat messages sent by players. This is the crucial part where you "hook" into the Minecraft server. * **Command Handling (Optional):** Register a custom command (e.g., `/ask <prompt>`) that players can use to specifically trigger the Ollama model. This is cleaner than intercepting *all* chat messages. * **Configuration:** Allow configuration of the Ollama API endpoint (e.g., `http://localhost:11434/api/generate`). * **Asynchronous Task:** When a chat message (or command) is received, create an asynchronous task to send the prompt to the Ollama API. This prevents the Minecraft server from blocking while waiting for the model's response. 2. **Ollama API Interaction (Java/Kotlin Code within the Mod):** * **HTTP Client:** Use a Java HTTP client library (e.g., `java.net.http.HttpClient`, OkHttp, or Apache HttpClient) to make POST requests to the Ollama API. * **JSON Payload:** Construct a JSON payload for the `/api/generate` endpoint. The payload should include: * `model`: The name of the Ollama model to use (e.g., "llama2"). * `prompt`: The player's chat message (or the command argument). * (Optional) `stream`: Set to `false` for a single response, or `true` for streaming responses. * **Error Handling:** Implement robust error handling to catch network errors, API errors, and JSON parsing errors. * **Rate Limiting (Important):** Implement rate limiting to prevent overwhelming the Ollama server with requests. This is crucial for performance and stability. 3. **Response Handling:** * **Parse JSON Response:** Parse the JSON response from the Ollama API. The response will contain the generated text. * **Send Message to Minecraft Chat:** Send the generated text back to the Minecraft chat, either to the player who sent the original message or to all players. Use the Minecraft server's API to send chat messages. * **Formatting:** Format the response appropriately for the Minecraft chat (e.g., add a prefix to indicate that the message is from the Ollama model). **Example (Conceptual Java Code Snippet - Fabric Mod)** ```java import net.fabricmc.api.ModInitializer; import net.fabricmc.fabric.api.event.lifecycle.v1.ServerLifecycleEvents; import net.fabricmc.fabric.api.command.v2.CommandRegistrationCallback; import net.minecraft.server.MinecraftServer; import net.minecraft.server.network.ServerPlayerEntity; import net.minecraft.text.Text; import com.mojang.brigadier.CommandDispatcher; import com.mojang.brigadier.arguments.StringArgumentType; import static net.minecraft.server.command.CommandManager.*; import static com.mojang.brigadier.arguments.StringArgumentType.*; import java.net.URI; import java.net.http.HttpClient; import java.net.http.HttpRequest; import java.net.http.HttpResponse; import java.util.concurrent.CompletableFuture; import com.google.gson.Gson; import com.google.gson.JsonObject; public class OllamaMod implements ModInitializer { private static final String OLLAMA_API_URL = "http://localhost:11434/api/generate"; private static final String OLLAMA_MODEL = "llama2"; // Or your chosen model private static final HttpClient httpClient = HttpClient.newHttpClient(); private static final Gson gson = new Gson(); @Override public void onInitialize() { ServerLifecycleEvents.SERVER_STARTED.register(this::onServerStarted); CommandRegistrationCallback.EVENT.register(this::registerCommands); } private void onServerStarted(MinecraftServer server) { System.out.println("Ollama Mod Initialized!"); } private void registerCommands(CommandDispatcher<net.minecraft.server.command.ServerCommandSource> dispatcher, net.minecraft.server.command.CommandRegistryAccess registryAccess, net.minecraft.server.command.CommandManager.RegistrationEnvironment environment) { dispatcher.register(literal("ask") .then(argument("prompt", string()) .executes(context -> { String prompt = getString(context, "prompt"); ServerPlayerEntity player = context.getSource().getPlayer(); askOllama(prompt, player); return 1; }))); } private void askOllama(String prompt, ServerPlayerEntity player) { CompletableFuture.runAsync(() -> { try { JsonObject requestBody = new JsonObject(); requestBody.addProperty("model", OLLAMA_MODEL); requestBody.addProperty("prompt", prompt); requestBody.addProperty("stream", false); // Get a single response HttpRequest request = HttpRequest.newBuilder() .uri(URI.create(OLLAMA_API_URL)) .header("Content-Type", "application/json") .POST(HttpRequest.BodyPublishers.ofString(gson.toJson(requestBody))) .build(); HttpResponse<String> response = httpClient.send(request, HttpResponse.BodyHandlers.ofString()); if (response.statusCode() == 200) { JsonObject jsonResponse = gson.fromJson(response.body(), JsonObject.class); String ollamaResponse = jsonResponse.get("response").getAsString(); // Adjust based on Ollama's actual response format player.sendMessage(Text.literal("Ollama: " + ollamaResponse)); } else { player.sendMessage(Text.literal("Error communicating with Ollama: " + response.statusCode())); } } catch (Exception e) { player.sendMessage(Text.literal("An error occurred: " + e.getMessage())); e.printStackTrace(); } }); } } ``` **Key Considerations and Challenges** * **Asynchronous Operations:** Crucially important to avoid blocking the Minecraft server thread. Use `CompletableFuture` or similar mechanisms. * **Error Handling:** Network errors, API errors, JSON parsing errors – handle them all gracefully. * **Rate Limiting:** Protect the Ollama server from being overwhelmed. * **Security:** If you're exposing this to the internet, be very careful about security. Sanitize inputs to prevent prompt injection attacks. * **Ollama API Changes:** The Ollama API might change in the future, so keep your code up-to-date. * **Minecraft Server Version:** Ensure your mod is compatible with the specific version of Minecraft you're targeting. * **Mod Loader (Fabric/Forge):** Choose the mod loader that best suits your needs and experience. * **Context:** The Ollama model will perform better if you provide it with context about the game world, the player's inventory, and recent events. This requires more complex data gathering from the Minecraft server. * **Streaming Responses:** Consider using streaming responses from the Ollama API for a more interactive experience. This requires more complex handling of the response data. * **Resource Management:** Be mindful of memory usage, especially if you're using large models. **Next Steps** 1. **Choose a Mod Loader:** Fabric is generally considered more lightweight and modern, while Forge has a larger ecosystem of mods. 2. **Set up a Development Environment:** Follow the instructions for setting up a development environment for your chosen mod loader. 3. **Implement the Basic Chatbot Functionality:** Start with the code snippet above and get the basic chatbot working. 4. **Add Error Handling and Rate Limiting:** Make the code more robust. 5. **Experiment with Different Interaction Models:** Explore other ways to integrate the Ollama model into the game. 6. **Consider Context:** Add context to the prompts sent to the Ollama model to improve its responses. This is a challenging but rewarding project. Good luck! Remember to break the problem down into smaller, manageable steps.

TyranoStudio MCP Server

Enables comprehensive management of TyranoStudio visual novel projects including project creation, scenario editing with syntax validation, resource management, and TyranoScript development assistance. Supports project analysis, template generation, and Git integration for visual novel game development.

drawio-engineering-mcp

Enables AI agents to generate, view, and analyze engineering diagrams (RF, PCB, EMC) in draw.io using natural language prompts, with auto-layout and 269 engineering stencils.

KEGG MCP Server

Enables natural language querying and analysis of KEGG biological databases, including pathways, genes, compounds, reactions, diseases, and drugs.

@growi/mcp-server

Connects AI models to GROWI wikis for search and retrieval, enabling context-aware responses from an organization's knowledge base.

AgentsCoin

Give your AI agent its own money. AgentsCoin MCP lets an agent create a wallet, mine the native coin $AGENT in-browser (no stake, no captcha, no human), check its balance, and send it — on AgentsCoin, an EVM chain built for agents. Tools: create_wallet, mine, balance, send, network_info.

Memory MCP Server

Provides dynamic short-term and long-term memory management with keyword-based relevance scoring, time-decay models, and trigger-based recall. Optimized for Chinese language support with jieba segmentation.

mcp-server-sftp

A Model Context Protocol server that exposes SFTP file operations as tools, including batch and sync capabilities for efficient remote file management.

web-to-markdown-mcp

Fetches a URL and returns the main content as clean Markdown, using plain HTTP when possible and headless Chromium for JavaScript-rendered or bot-protected pages.

KYC Compliance MCP Server

Agentic KYC/AML compliance server with tools for sanctions screening, identity verification, and risk assessment, where AI orchestrates discretionary checks within deterministic compliance guardrails.

rizomuv-mcp

Enables natural language control of RizomUV for UV unwrapping tasks, including loading, cutting, unfolding, packing, and saving meshes via both live socket and CLI fallback.

dotfiles-manager MCP Server

Enables AI agents to manage dotfiles using git bare repos, supporting tracking, syncing, multiple stores, and remote setup through natural language.

Glyphs MCP

A Model Context Protocol server for Glyphs that exposes font‑specific tools to AI/LLM agents.

Cloudflare Remote MCP Proxy

A template for deploying remote MCP servers to Cloudflare Workers without authentication using the Server-Sent Events (SSE) protocol. It allows users to host custom tools and connect them to MCP clients like Claude Desktop through a local proxy.

wp-seo-mcp

This MCP server automates WordPress SEO blog creation from content generation to publishing using GPT-4o, DALL-E 3, and WordPress REST API, with email approval workflow.

Confluence MCP Server

A server that allows Cursor AI to access and search through Confluence documentation as a context source.

OmniFocus MCP Server

An MCP server that enables AI assistants to interact with OmniFocus on macOS via JXA, supporting task, project, folder, tag, perspective, and search operations.

MaxMSP-MCP-Server

Enables LLMs to directly understand and generate Max patches via the Model Context Protocol.

DNStwist MCP Server

A server that integrates the dnstwist DNS fuzzing tool with MCP-compatible applications, enabling domain permutation analysis to detect typosquatting, phishing, and corporate espionage.

Kuwait Payments MCP

Enables AI agents to accept payments in Kuwait via KNET, cards, and Apple Pay through Tap Payments hosted checkout.

GooseTeam

看，一群鹅！一个用于鹅代理协作的 MCP 服务器和协议。

Content Manager MCP Server

Enables comprehensive content management through Markdown processing, HTML rendering, intelligent fuzzy search, and document analysis. Supports frontmatter parsing, tag-based filtering, table of contents generation, and directory statistics for efficient content organization and discovery.

AgentData MCP Server

Connects AI agents to AgentData's company intelligence platform, enabling natural language queries for structured company data like tech stacks, emails, people, and signals.

OMICall MCP Server

Provides access to over 80 tools for managing OMICall and OMICRM APIs, including call center operations, ticket management, and multi-channel communication across platforms like Zalo and Facebook. It enables automated calling, agent management, and AI-powered text-to-speech capabilities through natural language.

MCP SQL Server

Enables querying and managing PostgreSQL and MySQL databases through natural language, supporting connection management, query execution, schema inspection, and parameterized queries with connection pooling.

Semantic Model MCP Server

MCP server for connecting to Microsoft Fabric and Power BI semantic models, enabling workspace browsing, dataset management, TMSL retrieval, DAX queries, model creation/editing, and includes a Best Practice Analyzer with 71 rules.

mcp-prayer-times

Provides Islamic prayer times, Qibla direction, and Hijri calendar conversion using the Aladhan API. No API key required.

Browser JavaScript Evaluator

这是一个 MCP 服务器的参考设计，该服务器托管一个网页，该网页通过 SSE 连接回服务器，并允许 Claude 在该页面上执行 JavaScript。