MCP Server

What is MCP?

Core Concepts

As Agent-based applications rapidly evolve, developers are encountering common challenges during tool development and integration: the lack of unified standards hampers efficiency; constant adaptation across different platforms increases time and manpower costs; and open-source tools vary in quality, making maintenance and consolidation of high-quality capabilities expensive.

That’s where MCP (Model Context Protocol) comes in. Released by Anthropic, MCP is a leading open protocol designed to establish a secure and consistent two-way connection between large language models (LLMs) and external data sources. It addresses the fragmentation in tool implementation and enables cross-model compatibility.

Think of MCP as a universal plug or a USB port—the protocol defines a consistent specification for transmitting data from applications to LLMs. Any model that supports MCP can seamlessly interact with any MCP-compliant application.

Understanding Agent / MCP / API (Tool) Through the Story of a Smart Butler

Imagine having a smart butler at home. You simply describe what you need, and they take care of everything on your behalf. That’s the role of an Agent.

Role	Story Analogy	Technical Equivalent	Main Responsibilities
👤 User	Homeowner	End User	Submits a request and waits for the result.
🤖 Agent	Smart Butler	Intelligent Agent	Understands user intent, performs reasoning, and orchestrates tools. •Duties: Interpret user needs, decide execution steps and order. •Capability: Connect multiple tasks into a complete solution. •Analogy: The brain and decision-making center.
🔌 MCP	Plug Standard (Type-C)	Protocol Layer	Even the smartest butler struggles if every tool works differently. MCP solves the “tool invocation standardization” issue, allowing the butler to focus on task planning instead of interface compatibility. • Unified tool invocation specification • Agent doesn't need to understand the internal logic of each tool • Plug-and-play after standardization
🛠️ Tool	Tools (Weather / Flight / Email)	APIs or Services	Tools are the entities that perform the actual tasks: • Weather tool → returns weather info • Flight tool → provides flight data • Email tool → sends confirmation emails

Evolution of Three Development Approaches: API Docs → Function Call → MCP

From “Human Understanding” → “Model-Assisted Understanding” → “Model Execution via Protocol”

Dimension	API Documentation	Function Call	MCP Protocol
🧠 Understanding Entity	Human Developer	Model + Human (manual tool registration)	Model (auto protocol interpretation)
🎯 Target Audience	Human-readable docs	Model-readable function signatures	Machine-readable protocol description
🚀 Usage Flow	Read docs → Write request → Execute	Manually define function schema → Model invokes	MCP auto-exposes capability → Model discovers and invokes
🧩 Tool Setup	Requires separate wrappers per platform	Requires registering function per platform	Configure MCP service address once
🔍 Discoverability	Via documentation	Manually registered	Auto-discovered by model (dynamic capability detection)
🔄 Tool Update	Manual doc sync required	Each platform must update separately	MCP auto-syncs capability manifest
🧰 Integration Complexity	High: manual wrappers & prompt tuning	Medium: schema definition required	Low: auto-exposed via protocol, no duplicate config
🧠 Model Comprehension	Driven by prompt + doc interpretation	Depends on function schema clarity	Protocol-defined structured types enable native model understanding
🔒 Invocation Security	Weak (prone to parameter errors)	Moderate (schema constrains input format)	Strong (protocol-level validation + context control)
💬 Analogy	“Read the manual, do it yourself”	“Read function signature, follow the format”	“Plug in Type-C, auto-detect and run”
⚙️ Abstraction Layer	Application layer	SDK layer	Protocol layer
🌐 Extensibility	Low: custom wrappers needed	Medium: depends on platform support	High: cross-platform capability standardization
👩‍💻 Developer Experience	Manual-centric: write & maintain docs	Semi-automated: still need schema + setup	Fully automated: integrate once, reuse across LLMs

How to Use

APIfox MCP and HAP MCP serve different purposes depending on your integration needs:

• APIfox MCP allows large language models to read and understand the API documentation of HAP applications — ideal for “making the model API-aware.”
• HAP MCP enables direct access and manipulation of application data — designed for “executing business actions through the model.”

	APIfox MCP	HAP MCP
Purpose	Enable the model to understand HAP API documentation.	Allow the model to execute HAP application APIs.
Entry Point	Documentation URL: https://apifox.mingdao.com	Accessible from within the Application API Documentation panel.
Use Case	When the goal is to understand available API structures for generating API call strategies or orchestrating workflows. 📍Example: For a Customer Case Management application, if you want to create a real-time data dashboard, you can let the model (via a client like Cursor) access the API documentation using APIfox MCP (preferably V3 APIs).	When the goal is to perform actions or access real-time data directly within a production environment. 📍Example: In an orchestration flow, the model may directly retrieve application data, perform analysis, map fields, or update records using HAP MCP.
Characteristics	Exposes API definitions, documentation, and examples (read-only). Helps the model act as a knowledgeable API assistant. Best suited for: project development, API discovery, documentation automation.	Exposes executable tools with permission and security control. Helps the model act as a capable operational agent. Best suited for: business execution, production system workflows, governance.

MCP Configuration

Retrieve the MCP Configuration from HAP

• Open the Application API Documentation, navigate to the MCP tab, and copy the generated URL for configuration.

  

Cursor Configuration

• Go to [Settings] > [Tools & MCP] to add a new MCP server.

  

• Once saved, return to the list. A green indicator means the connection is successful.

  

Dify Integration

Using MCP in Agent Applications

• Click the Plugins button in the upper-right corner, explore the Marketplace, search for MCP, and locate “MCP SSE / StreamableHTTP”, then click Install.

  

• After installation, locate “MCP SSE / StreamableHTTP” under your installed plugins.

  

• Example Configuration:
  Enter your MCP config. You can replace "hap-mcp" with any alias name. The url should be the one generated from your application.

    {"hap-mcp":{"transport":"streamable_http","url":"your-mcp-url-here"}}

• Add the MCP tool and run a test to verify

  

Using MCP in Workflows

• Click the Plugins button in the upper-right corner. In the Marketplace, search for “Agent Strategy (Support MCP Tools)” and click Install.

  

  

• Example Configuration:
  Enter your MCP config. You can replace "hap-mcp" with any alias name. The url should be the one generated from your application.

    {"hap-mcp":{"transport":"streamable_http","url":"your-mcp-url-here"}}

VS Code + Cline Plugin

• Install the Cline Plugin

  

• Click "MCP Servers" in the bottom-right corner of VS Code to complete the configuration

  

• Example Configuration:

  "hap-mcp": {
  "url": "your-mcp-url-here", 
  "type": "streamableHttp",
  "disabled": false,
  "autoApprove": []
  }

---