发现优秀的 MCP 服务器

authorize-net-mcp

实验性的 Authorize.net Node.js TypeScript MCP 服务器

git-slim

A token-optimized Git MCP server that reduces context window tokens by 59% while preserving full functionality, enabling AI assistants to interact with Git repositories using only 5 grouped tool operations.

RhinoMCP

通过模型上下文协议将 Rhino3D 连接到 Claude AI，从而通过直接控制 Rhino 的功能实现 AI 辅助的 3D 建模和设计工作流程。

cloudprice-mcp

MCP server that lets Claude (or any MCP-compatible client) compare on-demand compute + storage pricing across AWS, Azure, and GCP in real time.

ContextOS MCP Server

Unified context intelligence layer for AI agents, enabling orchestration of memory, reasoning, and self-healing indexes with cognition primitives and churn-aware retrieval routing.

MCP File Server

用于读取和写入本地文件的 MCP 服务器 (Yòng yú dúqǔ hé xiě rù běndì wénjiàn de MCP fúwùqì) Alternatively, depending on the specific context, you might also say: 本地文件读写 MCP 服务器 (Běndì wénjiàn dú xiě MCP fúwùqì)

Arco Lexicon

Canonical vocabulary server for autonomous business design. Exposes the Arco Lexicon as seven MCP tools: term lookup, related terms, alignment verification, citation formatting, source retrieval, term listing, and term suggestion. No authentication required. Streamable HTTP transport.

Mcp Qdrant Docker

Okay, here's a breakdown of Docker configuration for a Qdrant MCP (Multi-Cluster Proxy) server, along with explanations and best practices. I'll provide a `docker-compose.yml` example and discuss the key elements. **Understanding Qdrant MCP** The Qdrant MCP acts as a gateway to multiple Qdrant clusters. It handles routing requests to the appropriate cluster based on configuration. This is useful for: * **Scaling:** Distributing your data across multiple Qdrant clusters. * **Isolation:** Separating data for different tenants or applications. * **High Availability:** Routing around failed clusters. * **Geo-Distribution:** Placing data closer to users. **`docker-compose.yml` Example** ```yaml version: "3.9" services: qdrant-mcp: image: qdrant/qdrant-mcp:latest # Or specify a version ports: - "6333:6333" # MCP API port - "6334:6334" # MCP GRPC port environment: QDRANT_MCP_CONFIG_PATH: /qdrant-mcp/config/config.yaml # Path to your config file volumes: - ./config:/qdrant-mcp/config # Mount your config directory restart: unless-stopped depends_on: - qdrant-cluster-1 # Replace with your actual cluster service names - qdrant-cluster-2 # Replace with your actual cluster service names # Add more dependencies as needed qdrant-cluster-1: image: qdrant/qdrant:latest # Or specify a version ports: - "6335:6333" # Cluster 1 API port environment: QDRANT__SERVICE__GRPC_PORT: 6336 volumes: - qdrant_data_1:/qdrant/storage restart: unless-stopped qdrant-cluster-2: image: qdrant/qdrant:latest # Or specify a version ports: - "6337:6333" # Cluster 2 API port environment: QDRANT__SERVICE__GRPC_PORT: 6338 volumes: - qdrant_data_2:/qdrant/storage restart: unless-stopped volumes: qdrant_data_1: qdrant_data_2: ``` **Explanation:** 1. **`version: "3.9"`:** Specifies the Docker Compose file version. Use a version compatible with your Docker installation. 2. **`services:`:** Defines the services (containers) that will be run. 3. **`qdrant-mcp:`:** The service for the Qdrant MCP. * **`image: qdrant/qdrant-mcp:latest`:** Uses the official Qdrant MCP Docker image. **Important:** Consider using a specific version tag (e.g., `qdrant/qdrant-mcp:v1.5.0`) instead of `latest` for production to ensure consistent behavior. * **`ports:`:** Maps the container ports to the host ports. * `6333:6333`: The main Qdrant MCP API port (HTTP). You'll use this to send requests to the MCP. * `6334:6334`: The Qdrant MCP gRPC port. * **`environment:`:** Sets environment variables for the container. * `QDRANT_MCP_CONFIG_PATH: /qdrant-mcp/config/config.yaml`: Specifies the path to the MCP configuration file *inside* the container. This is crucial. * **`volumes:`:** Mounts a directory from your host machine into the container. * `./config:/qdrant-mcp/config`: Mounts the `./config` directory on your host to `/qdrant-mcp/config` inside the container. This is where you'll place your `config.yaml` file. **You MUST create this `./config` directory and put your `config.yaml` file in it.** * **`restart: unless-stopped`:** Automatically restarts the container if it crashes, unless you explicitly stop it. Good for reliability. * **`depends_on:`:** Specifies dependencies. The MCP will only start *after* the listed services (your Qdrant clusters) are running. **Important:** Replace `qdrant-cluster-1` and `qdrant-cluster-2` with the actual names of your Qdrant cluster services in your `docker-compose.yml`. Add more as needed. 4. **`qdrant-cluster-1` and `qdrant-cluster-2`:** Example Qdrant cluster services. You'll need to configure these according to your needs. * **`image: qdrant/qdrant:latest`:** Uses the official Qdrant Docker image. Again, use a specific version tag for production. * **`ports:`:** Maps the container ports to the host ports. Make sure these ports don't conflict with each other or with the MCP. * **`environment:`:** Sets environment variables for the container. The `QDRANT__SERVICE__GRPC_PORT` is important for internal communication within the cluster. * **`volumes:`:** Mounts a volume for persistent storage of the Qdrant data. `qdrant_data_1:/qdrant/storage` creates a named volume. * **`restart: unless-stopped`:** Automatically restarts the container if it crashes. 5. **`volumes:`:** Defines named volumes for persistent storage. This is important so your data isn't lost when the containers are stopped or restarted. **`config.yaml` (Qdrant MCP Configuration)** This is the most important part. The `config.yaml` file tells the MCP how to route requests to your Qdrant clusters. Here's an example: ```yaml clusters: cluster1: address: "qdrant-cluster-1:6333" # Use the service name and port cluster2: address: "qdrant-cluster-2:6333" # Use the service name and port collection_mappings: my_collection: cluster: cluster1 # All requests for "my_collection" go to cluster1 another_collection: cluster: cluster2 # All requests for "another_collection" go to cluster2 shared_collection: cluster: cluster1 # All requests for "shared_collection" go to cluster1 ``` **Explanation of `config.yaml`:** * **`clusters:`:** Defines the Qdrant clusters that the MCP will route to. * `cluster1`, `cluster2`: Arbitrary names for your clusters. Use descriptive names. * `address`: The address of the Qdrant cluster. **Crucially, use the Docker service name (e.g., `qdrant-cluster-1`) and the internal port (6333 by default).** Docker's internal DNS will resolve the service name to the container's IP address. Do *not* use `localhost` or the host's IP address here. * **`collection_mappings:`:** Defines how collections are mapped to clusters. * `my_collection`, `another_collection`, `shared_collection`: The names of your Qdrant collections. * `cluster`: The name of the cluster (as defined in the `clusters` section) that should handle requests for this collection. **Important Considerations and Best Practices:** * **Version Pinning:** Always use specific version tags for your Docker images (e.g., `qdrant/qdrant-mcp:v1.5.0`, `qdrant/qdrant:v1.5.0`) instead of `latest` in production. This prevents unexpected behavior when the images are updated. * **Configuration Management:** Use a proper configuration management system (e.g., environment variables, configuration files) to manage your Qdrant and MCP settings. Avoid hardcoding values in your Dockerfiles or Compose files. * **Networking:** Docker Compose automatically creates a default network for your services. This allows the services to communicate with each other using their service names. If you need more complex networking, you can define custom networks. * **Health Checks:** Implement health checks for your Qdrant clusters and the MCP. This allows Docker to automatically restart unhealthy containers. See the Qdrant documentation for details on health check endpoints. * **Logging:** Configure logging for your Qdrant clusters and the MCP. This is essential for troubleshooting. Docker can collect logs from the containers and send them to a central logging system. * **Monitoring:** Monitor the performance of your Qdrant clusters and the MCP. Use metrics to track CPU usage, memory usage, disk I/O, and network traffic. * **Security:** Secure your Qdrant clusters and the MCP. Use authentication and authorization to control access to your data. Consider using TLS/SSL to encrypt communication between the MCP and the clusters. * **Resource Limits:** Set resource limits (CPU, memory) for your containers to prevent them from consuming too many resources. * **Backup and Restore:** Implement a backup and restore strategy for your Qdrant data. * **Testing:** Thoroughly test your Qdrant MCP setup before deploying it to production. * **Qdrant Documentation:** Refer to the official Qdrant documentation for the most up-to-date information and best practices: [https://qdrant.tech/documentation/](https://qdrant.tech/documentation/) **How to Run:** 1. **Create the `config` directory:** `mkdir config` 2. **Create the `config.yaml` file:** Place the `config.yaml` file (with your cluster definitions and collection mappings) in the `config` directory. 3. **Save the `docker-compose.yml` file:** Save the `docker-compose.yml` file in the same directory as the `config` directory. 4. **Run Docker Compose:** `docker-compose up -d` (This will start the containers in detached mode.) **Troubleshooting:** * **Check the logs:** Use `docker-compose logs qdrant-mcp` (or the name of your MCP service) to view the logs for the MCP container. Look for errors related to configuration, cluster connections, or routing. * **Verify the configuration:** Double-check the `config.yaml` file for errors. Make sure the cluster addresses are correct and that the collection mappings are accurate. * **Check network connectivity:** Make sure the MCP container can communicate with the Qdrant cluster containers. You can use `docker exec -it qdrant-mcp bash` to enter the MCP container and then use tools like `ping` or `telnet` to test connectivity. * **Qdrant Cluster Status:** Ensure your Qdrant clusters are running and healthy *before* starting the MCP. **Chinese Translation of Key Terms:** * **Qdrant MCP (Multi-Cluster Proxy):** Qdrant 多集群代理 (Duō jíqún dàilǐ) * **Cluster:** 集群 (Jíqun) * **Collection:** 集合 (Jíhé) * **Configuration:** 配置 (Pèizhì) * **Docker Compose:** Docker Compose * **Service:** 服务 (Fúwù) * **Image:** 镜像 (Jìngxiàng) * **Container:** 容器 (Róngqì) * **Port:** 端口 (Duānkǒu) * **Environment Variable:** 环境变量 (Huánjìng biànliàng) * **Volume:** 卷 (Juǎn) * **Mapping:** 映射 (Yìngshè) This comprehensive guide should help you set up a Qdrant MCP server using Docker. Remember to adapt the configuration to your specific needs and environment. Good luck!

Grabba MCP Server

Microservice Connector Protocol server that exposes Grabba API functionalities as callable tools, allowing AI agents and applications to interact with Grabba's data extraction and management services.

Moleculer MCP Server

Exposes Moleculer microservices as MCP tools for AI assistants like Claude and GPT, with auto-discovery and schema conversion.

Google Ads MCP Server

Enables LLMs to interact with the Google Ads API to retrieve account information, list accessible customers, and query campaign performance. It allows users to manage and analyze Google Ads data through natural language interfaces.

MCP Image Recognition Server

提供图像识别功能，使用 Anthropic Claude Vision 和 OpenAI GPT-4 Vision API，支持多种图像格式，并提供通过 Tesseract OCR 进行可选的文本提取。

gsc-mcp-server

Enables Claude to access Google Search Console data including search performance, URL indexation, sitemaps, and built-in SEO analysis tools such as trending queries, cannibalization detection, and traffic drop diagnostics.

Power Platform Pipeline MCP Server

Exposes Power Platform Pipeline operations as MCP tools for Copilot Studio agents, enabling pipeline discovery, deployments, approvals, and configuration management.

Codecov MCP Server

Enables querying Codecov coverage data including file, commit, repository, and pull request coverage.

Financial Data MCP Server

Provides real-time and historical financial market data from Yahoo Finance, including stock prices, options chains, and company news. It also enables technical analysis calculations like moving averages and RSI for comparing or analyzing stock performance.

Gerrit Review MCP Server

Provides integration with Gerrit code review system, allowing AI assistants to fetch change details and compare patchset differences for code reviews.

GSC MCP Server v2 - Remote Edition

Connects Claude AI to Google Search Console with OAuth 2.0 authentication, enabling users to analyze search performance, inspect URLs, manage sitemaps, and export analytics data through natural language conversations.

MCPDocSearch

Crawls documentation websites and provides semantic search capabilities over the content through vector embeddings, enabling natural language queries of technical documentation.

io.github.peterkolbe/ableton-for-ai

MCP server that bridges Ableton Live with AI models, enabling real-time project inspection and control such as track overview, device parameters, and audio analysis.

Excel WebView2 MCP

Connects Claude Code to Microsoft Excel add-ins via the Chrome DevTools Protocol, enabling inspection, automation, and lifecycle management of add-ins running in WebView2.

MCP Local File Reader

Enables comprehensive local file system operations including searching and reading content from various formats like PDF, Office documents, CSV, and text files. It features automatic character encoding detection and secure path validation for directory access.

OpenAI Image Generation MCP Server

Provides tools for generating and editing images using OpenAI's gpt-image-1 model via an MCP interface, enabling AI assistants to create and modify images based on text prompts.

career-scout-mcp

A production-grade MCP server demonstrating the wrapping pattern for AI-augmented data pipelines, specifically a job-search scoring pipeline.

PAL - Project API Locker

Enables secure API key management for development projects by storing keys in OS keychains, auto-generating .env files and SDK client code, and providing AI-assisted key management through Claude Code integration.

Querysharp MCP Server

Enables AI assistants to analyze and optimize PostgreSQL database performance by identifying missing indexes and suggesting query rewrites. It allows users to retrieve database projects and implement performance fixes directly through natural language in their code editor.

Coffee Company MCP Server

An MCP adapter that maps Coffee Company B2B HTTP APIs to MCP tools, allowing AI agents to query member information, benefits, coupons, and payment statuses. It enables seamless integration for AI assistants to manage coffee-related customer assets and loyalty details through natural language.

Xero MCP Server

一个模型上下文协议（Model Context Protocol，MCP）服务器实现，提供对Xero会计和业务功能的标准化访问，从而可以通过MCP实现诸如联系人管理、发票创建和科目表管理等操作。