Auto‑Generate Microservice Architecture Diagrams with Qwen3‑Coder and PlantUML

Purpose

Demonstrate using Lingma (LLM‑powered tool) to analyze a full‑stack microservice project, generate a PlantUML system architecture diagram and API documentation.

Prerequisites

Clone the sample repository:

git clone https://github.com/duffqiu/bookinfo.git

Install Visual Studio Code (download from https://code.visualstudio.com/).

Install the Lingma extension from the VS Code marketplace.

Install the PlantUML extension (or use the online PlantUML server at https://www.plantuml.com/plantuml/).

Workflow

Open the cloned bookinfo folder in VS Code.

Start Lingma in agent mode and select the model qwen3‑code .

Prompt the model to analyse the whole project and output a PlantUML description of the system architecture:

请详细分析下工程，并用plantuml给出系统架构图的描述

Lingma returns a .puml file.

Render the .puml file with the PlantUML extension to view the diagram.

Refine the diagram to keep only microservices and the data layer: 请只保留微服务和数据层 Generate a sequence diagram that shows interactions between microservices:

请用plantuml画出各个微服务交互的时序图，输出到文件action.uml

If the initial output omits version‑specific details for the reviews service, add them with:

需要增加描述不同的reviews版本的处理过程

Select a Java source file (e.g., ReviewsResource.java) as context and ask for an API specification:

请给这个文件创建接口文档

Sample API Documentation Produced for the reviews Service

# Reviews服务API文档
## 基础URL
http://reviews:9080

### 1. 健康检查
GET /health
响应: {"status":"Reviews is healthy"}   200 OK

### 2. 获取图书评论
GET /reviews/{productId}
路径参数: productId (int)
请求头: x-request-id, x-b3-traceid, x-b3-spanid, x-b3-parentspanid, x-b3-sampled, x-b3-flags 等
响应示例:
{
  "id":"0",
  "podname":"reviews-v1-7b4f5bd5c-2xk9m",
  "clustername":"cluster-1",
  "reviews":[
    {"reviewer":"Reviewer1","text":"...","rating":{"stars":5,"color":"black"}},
    {"reviewer":"Reviewer2","text":"...","rating":{"stars":4,"color":"black"}}
  ]
}
说明:
- v1 不返回 rating
- v2 返回 rating，color 为 "black"
- v3 返回 rating，color 为 "red"（需设置 STAR_COLOR="red"、ENABLE_RATINGS="true"、SERVICE_VERSION="v3"）

### 环境变量
ENABLE_RATINGS (bool)
STAR_COLOR (string, default "black")
SERVICES_DOMAIN, RATINGS_HOSTNAME, RATINGS_SERVICE_PORT, HOSTNAME, CLUSTER_NAME

### 方法概览
- getJsonResponse(productId, starsReviewer1, starsReviewer2) → String
- getRatings(productId, requestHeaders) → JsonObject
- health() → Response
- bookReviewsById(productId, requestHeaders) → Response

Key Points

The process is fully automated: Lingma reads the codebase, produces PlantUML files, and can generate Markdown API specs.

Model qwen3‑code understands Chinese prompts and returns syntactically correct PlantUML and Markdown.

Version‑specific behaviour of the reviews microservice is controlled by environment variables and can be captured by additional prompts.

All generated artifacts (architecture diagram, sequence diagram, API docs) can be rendered directly in VS Code using the PlantUML extension.