发布日期:2026-08-13 | 分类:AI工具教程
WAI Play(waiterve/wai-play)是GitHub新开源的AI Web游戏测试与质量评估平台(345星)。利用AI自动生成测试用例、模拟玩家操作、评估游戏质量,替代人工QA流程。
git clone https://github.com/waiterve/wai-play.git
cd wai-play
# 安装前端依赖
npm install
# 安装Python AI模块
pip install -r requirements.txt
# 安装Playwright浏览器
npx playwright install chromium
# config.yaml
ai:
provider: openai
api_key: ${OPENAI_API_KEY}
model: gpt-5.6
target:
url: "http://localhost:3000" # 被测游戏地址
type: "web_game" # 游戏类型
testing:
max_sessions: 5 # 并发测试会话数
session_duration: 300 # 每次测试时长(秒)
coverage:
- "ui_responsiveness" # UI响应性
- "game_logic" # 游戏逻辑
- "performance" # 性能
- "user_experience" # 用户体验
screenshots: true # 自动截图
video: true # 录制视频
# 启动测试
npx wai-play run --config config.yaml
# AI会自动:
# 1. 分析游戏页面结构
# 2. 生成测试用例(点击/输入/拖拽等)
# 3. 模拟玩家操作
# 4. 检测异常(崩溃/卡顿/逻辑错误)
# 5. 生成测试报告
from wai_play import GameTester
tester = GameTester(config="config.yaml")
# 运行单次测试
report = tester.test(
url="http://localhost:3000",
scenarios=["login", "gameplay", "payment"],
duration=60
)
# 查看报告
print(f"通过率: {report.pass_rate}%")
print(f"发现Bug: {report.bugs}")
for bug in report.bugs:
print(f" - {bug.severity}: {bug.description}")
print(f" 截图: {bug.screenshot}")
# 批量测试
results = tester.batch_test(
urls=["http://game1.com", "http://game2.com"],
runs=10
)
# scenarios.yaml
- name: "新手引导测试"
steps:
- action: "click"
target: "#start-button"
expect: "页面跳转到引导页"
- action: "wait"
duration: 2
- action: "click"
target: ".tutorial-next"
expect: "出现下一步引导"
ai_check: true # AI验证每步结果
- name: "压力测试"
steps:
- action: "rapid_click"
target: ".game-area"
count: 1000
interval: 0.01
ai_check: true
# 报告输出目录:./reports/
# 包含:
# - HTML报告(含截图和视频)
# - JSON数据(用于CI/CD集成)
# - 性能图表
# 在CI/CD中集成
# .github/workflows/test.yml
# - run: npx wai-play run --config config.yaml --fail-on-bug
WAI Play(waiterve/wai-play)是GitHub新开源的AI Web游戏测试与质量评估平台(345星)。利用AI自动生成测试用例、模拟玩家操作、评估游戏质量,替代人工QA流程。
Midscene.js UI自动化教程