Build an AI Agent Connected to WeCom in 5 Minutes: MCP Protocol Practical Tutorial

Why WeCom + MCP?

WeCom (Enterprise WeChat) covers 80%+ of Chinese enterprises and is the de facto standard for internal communication. Connecting AI Agent to WeCom means:

Zero learning curve: Employees use AI in their familiar chat interface
Built-in permissions: Reuse WeCom's organizational structure and access control
Message delivery: AI can proactively push messages to employees and groups
Approval integration: AI can initiate approval workflows on behalf of employees

WeCom + MCP = the shortest path to enterprise AI Agent

Architecture Overview

┌──────────────┐     MCP Protocol    ┌──────────────────┐     HTTP API     ┌──────────────┐
│  AI Agent    │ ◄──────────────────► │   MCP Server     │ ◄──────────────► │  WeCom API   │
│  (Claude)    │    JSON-RPC         │  (Our impl)      │    REST calls    │              │
└──────────────┘                     └──────────────────┘                   └──────────────┘

MCP Server Tools:
├── send_message        Send message to user/group
├── search_contacts     Search address book
├── get_department      Get department info
├── create_approval     Create approval request
├── upload_file         Upload file to WeCom
└── get_chat_history    Get group chat history

Step 1: Create WeCom Application

1.1 Get WeCom Credentials

1. App Management → Create App
2. Record credentials:
   - corpId: ww1234567890abcdef
   - agentId: 1000002
   - secret: xxxxxxxxxxxxxxxxxxxxxxxxxx
3. Set callback URL: https://your-domain.com/wecom/callback

Step 2: Implement MCP Server

2.1 Project Structure

wecom-mcp-server/
├── src/main/java/com/example/wecom/
│   ├── WecomMcpServerApplication.java
│   ├── config/WecomConfig.java
│   ├── mcp/
│   │   ├── WecomMcpServer.java
│   │   └── tools/
│   │       ├── SendMessageTool.java
│   │       ├── SearchContactsTool.java
│   │       ├── GetDepartmentTool.java
│   │       └── CreateApprovalTool.java
│   └── wecom/
│       ├── WecomApiClient.java
│       └── WecomTokenManager.java
└── pom.xml

2.2 Core Dependencies

<dependencies>
    <dependency>
        <groupId>io.modelcontextprotocol</groupId>
        <artifactId>mcp-spring-boot-starter</artifactId>
        <version>0.7.0</version>
    </dependency>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-web</artifactId>
        <version>3.3.0</version>
    </dependency>
</dependencies>

2.3 MCP Server Implementation

@SpringBootApplication
public class WecomMcpServerApplication {
    public static void main(String[] args) {
        SpringApplication.run(WecomMcpServerApplication.class, args);
    }
}

@Configuration
@ConfigurationProperties(prefix = "wecom")
@Data
public class WecomConfig {
    private String corpId;
    private String agentId;
    private String secret;
    private String token;
    private String encodingAesKey;
}

2.4 Send Message Tool

@Component
public class SendMessageTool implements McpTool {

    private final WecomApiClient wecomClient;

    @Override
    public String name() {
        return "send_message";
    }

    @Override
    public String description() {
        return "Send a message to a specified user or group via WeCom";
    }

    @Override
    public JsonNode inputSchema() {
        return objectMapper.readTree("""
            {
              "type": "object",
              "properties": {
                "receiverType": {
                  "type": "string",
                  "enum": ["user", "group"],
                  "description": "Receiver type: user or group"
                },
                "receiverId": {
                  "type": "string",
                  "description": "Receiver ID"
                },
                "messageType": {
                  "type": "string",
                  "enum": ["text", "markdown", "file"],
                  "description": "Message type"
                },
                "content": {
                  "type": "string",
                  "description": "Message content"
                }
              },
              "required": ["receiverType", "receiverId", "messageType", "content"]
            }
            """);
    }

    @Override
    public McpToolResult execute(Map<String, Object> args) {
        String receiverType = (String) args.get("receiverType");
        String receiverId = (String) args.get("receiverId");
        String messageType = (String) args.get("messageType");
        String content = (String) args.get("content");

        try {
            String msgId = wecomClient.sendMessage(
                receiverType, receiverId, messageType, content
            );
            return McpToolResult.success(
                Map.of("msgId", msgId, "status", "sent")
            );
        } catch (WecomApiException e) {
            return McpToolResult.error("Send failed: " + e.getMessage());
        }
    }
}

2.5 Token Manager (Auto-refresh)

@Component
public class WecomTokenManager {

    private final WecomConfig config;
    private final RestTemplate restTemplate;
    private String accessToken;
    private long expiresAt;

    public synchronized String getAccessToken() {
        if (accessToken != null && System.currentTimeMillis() < expiresAt) {
            return accessToken;
        }
        refreshAccessToken();
        return accessToken;
    }

    private void refreshAccessToken() {
        String url = String.format(
            "https://qyapi.weixin.qq.com/cgi-bin/gettoken?corpid=%s&corpsecret=%s",
            config.getCorpId(), config.getSecret()
        );

        JsonNode json = restTemplate.getForObject(url, JsonNode.class);
        this.accessToken = json.path("access_token").asText();
        int expiresIn = json.path("expires_in").asInt();
        this.expiresAt = System.currentTimeMillis() + (expiresIn - 300) * 1000L;
    }
}

Step 3: Configuration

application.yml

server:
  port: 8081

wecom:
  corpId: ${WECOM_CORP_ID}
  agentId: ${WECOM_AGENT_ID}
  secret: ${WECOM_SECRET}
  token: ${WECOM_CALLBACK_TOKEN}
  encodingAesKey: ${WECOM_ENCODING_AES_KEY}

mcp:
  server:
    name: wecom-mcp-server
    version: 1.0.0
    transport: sse
    sseEndpoint: /mcp/sse

Step 4: Configure in Claude Desktop

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "wecom": {
      "url": "http://localhost:8081/mcp/sse",
      "transport": "sse"
    }
  }
}

Now you can use WeCom features directly in Claude:

User: Send a message to the product group notifying about tomorrow's 3pm review meeting

Claude: I'll send that message for you.
[Calling send_message tool]
→ receiverType: "group"
→ receiverId: "product_group_id"
→ messageType: "text"
→ content: "Notice: Review meeting tomorrow at 3pm. Please attend."

Message sent successfully! msgId: msg_20260611_001

Production Deployment

Docker

FROM eclipse-temurin:21-jre-alpine
WORKDIR /app
COPY target/wecom-mcp-server.jar app.jar
EXPOSE 8081
ENTRYPOINT ["java", "-jar", "app.jar"]

version: '3.8'
services:
  wecom-mcp:
    build: .
    ports:
      - "8081:8081"
    environment:
      - WECOM_CORP_ID=${WECOM_CORP_ID}
      - WECOM_AGENT_ID=${WECOM_AGENT_ID}
      - WECOM_SECRET=${WECOM_SECRET}
    restart: unless-stopped

Security Notes

Never hardcode secrets: Use environment variables or Vault
IP whitelist: Set API call IP whitelist in WeCom admin
Message audit: Log all sent messages for compliance
Rate limiting: WeCom API has rate limits — implement throttling

Summary

The "5-minute" setup is possible because MCP standardizes the integration — implement the MCP Server once, and all MCP-compatible AI tools can use it:

Implement once, use everywhere: Claude, Cursor, custom Agents all connect
Spring Boot ecosystem: WeCom SDK, auth, audit out of the box
Production-ready: Auto token refresh, rate limiting, audit logs, Docker deployment

This is the shortest path from "AI that chats" to "AI that works" — MCP makes it simple.