# self-test-skill **Repository Path**: jokingremarks/self-test-skill ## Basic Information - **Project Name**: self-test-skill - **Description**: self-test-skill - **Primary Language**: Unknown - **License**: MIT - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2026-06-28 - **Last Updated**: 2026-06-28 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # AIQA — AI-Driven Quality Assurance for Frontend > 多智能体协作的质量验收框架。Test Agent 找 Bug,Verify Agent 确认,Skill 固化流程。 > *A multi-agent quality assurance framework. Test Agent finds bugs, Verify Agent confirms them, Skill locks in the process.* --- ## 安装 / Installation ```bash # 一行命令安装 / One command to install npx aiqa init # 或安装到指定目录 / Or install to a target directory npx aiqa init --dir /path/to/project ``` 安装后自动生成 `.claude/` 目录,包含 Agent 定义、模板文件和状态追踪。 *Creates `.claude/` with agent definitions, templates, and state tracking.* ```bash # 手动安装 / Manual install cp -r agents/ templates/ .claude/ ``` --- ## 解决的问题 / What Problem It Solves 前端项目的基础 Bug 没人发现——搜索框输入后不筛选、状态筛选点了没反应、API 参数名写错。这些 Bug 不是因为逻辑复杂,而是因为**没有人逐项检查**。 *Basic frontend bugs go unnoticed — search doesn't filter, status filters don't respond, API parameter names are wrong. Not because the logic is complex, but because nobody checks systematically.* | Agent | 做什么 / Role | 中文 | |-------|-------------|------| | **Test Agent** | Executes scenarios, reports bugs | 逐场景执行测试,产出 Bug 报告 | | **Verify Agent** | Independently reproduces, confirms root cause | 独立复现 Bug,确认根因,评估定级 | | **Self-Test Skill** | Module complete hook, auto-trigger | 固化验收流程,新模块完成自动触发 | --- ## 快速开始 / Quick Start ### 1. 编写你的测试场景 / Write Test Scenarios 复制 `test-scenarios.template.md` 为 `test-scenarios.md`: *Copy `test-scenarios.template.md` to `test-scenarios.md` and fill in your scenarios:* ```markdown ## 3. 用户管理(核心 CRUD) | 3.1 | 列表分页加载 | 显示列表 | 检查表格 | | 3.2 | 🔴 搜索功能 | 输入关键词后列表筛选 | 重点检查 | ``` ### 2. 运行自检 / Run Self-Test ```bash /agent "Execute test-agent.md: test all scenarios, write report to test-report.md" ``` ### 3. 验证 Bug / Verify Bugs ```bash /agent "Execute verify-agent.md: independently reproduce each bug, confirm root cause" ``` --- ## 核心设计 / Core Design ### 停止规则 / Stop Rules | 条件 / Condition | 触发 / Trigger | 动作 / Action | |------|------|------| | ALL GREEN | All checks pass | ✅ 验收通过 / Approved | | P0 Bug | Core function broken | 🛑 阻塞合入 / Block merge | | 连续相同失败 | 2x same error | 🛑 Builder 在瞎猜 / Spinning | | 回归 / Regression | Fix broke something | 🛑 拆东墙补西墙 / Rollback | ### 三条红线 / Three Red Lines ``` 🚫 Never report success without checker output 永远不在没有 checker 输出的情况下报告成功 🚫 Never weaken/delete/skip checks to reach ALL GREEN 永远不弱化/删除/跳过检查来达到 ALL GREEN 🚫 Never modify the checker's tool whitelist 永远不修改 checker 的工具白名单 ``` --- ## 文件结构 / File Structure ``` .claude/ ├── agents/ │ ├── test-agent.md ← 测试 Agent / Tester │ ├── verify-agent.md ← 验证 Agent / Verifier │ └── self-test-skill.md ← 验收 Skill / Module Hook ├── test-scenarios.md ← 你的测试场景 / Your scenarios ├── test-report.md ← 测试报告 (auto-generated) ├── loop-config.md ← Loop 调度配置 └── state.md ← Bug 追踪表 / Bug tracker ``` --- ## 适用场景 / Use Cases - ✅ React / Vue / Angular 前端项目 - ✅ Spring Boot / Express / FastAPI 后端项目 - ✅ 任何有搜索+筛选+CRUD 的管理系统 - ✅ AI Agent 开发的项目的质量门禁 --- ## License MIT