---
title: "Evidence Collector"
type: source
tags: [testing, qa, evidence, agency]
date: 2026-04-21
---

## Source File
- [[raw/Agent/agency-agents/testing/testing-evidence-collector.md]]

## Summary
- 核心主题：QA 证据收集智能体 EvidenceQA 的角色定义与测试方法论
- 问题域：AI Agent 开发中的质量保证流程，避免无证据的"幻想式"报告
- 方法/机制：基于 Playwright 截图、可复现命令、事实检查的证据驱动 QA
- 结论/价值：建立现实的质量评估标准，默认发现 3-5 个问题，要求视觉证据

## Key Claims
- 视觉证据是唯一真相：无法截图证明的功能视为不存在
- 默认发现问题：首次实现总有 3-5+ 个问题，"零问题"是危险信号
- 一切需证明：每个声明都需要截图证据支撑
- 诚实质量评估：Basic/Good/Excellent 级别，不接受虚假的 A+ 评分

## Key Quotes
> "Screenshots Don't Lie" — 视觉证据是唯一真相
> "Default to Finding Issues" — 首次实现总有 3-5+ 个问题
> "Prove Everything" — 每个声明都需要截图证据

## Key Concepts
- [[证据驱动 QA（Evidence-Driven QA）]]：要求所有声明都有视觉证据支撑的 QA 方法论
- [[幻想式报告（Fantasy Reporting）]]：无证据支撑的乐观声明，如"零问题"、"完美评分"
- [[Reality Checker]]：通过实际命令和截图验证功能真实状态
- [[Playwright 截图]]：自动化捕获界面截图作为 QA 证据

## Key Entities
- [[EvidenceQA]]：截图驱动的 QA 专家智能体，厌恶幻想式报告
- [[Reality Checker]]：与 EvidenceQA 协同的质量检查智能体
- [[Test Results Analyzer]]：测试结果分析与缺陷预测智能体

## Connections
- [[Reality Checker (Agent)]] ← complements ← [[Evidence Collector]]
- [[Test Results Analyzer]] ← extends ← [[Evidence Collector]]
- [[Evidence Collector]] ← part_of ← [[The Agency]]
- [[The Agency]] ← contains ← [[Testing Agents]]

## QA Report Template

```markdown
# QA Evidence-Based Report

## 🔍 Reality Check Results
**Commands Executed**: [List actual commands run]
**Screenshot Evidence**: [List all screenshots reviewed]
**Specification Quote**: "[Exact text from original spec]"

## 📸 Visual Evidence Analysis
**Comprehensive Playwright Screenshots**: responsive-desktop.png, responsive-tablet.png, responsive-mobile.png, dark-mode-*.png
**What I Actually See**:
- [Honest description of visual appearance]

**Specification Compliance**:
- ✅ Spec says: "[quote]" → Screenshot shows: "[matches]"
- ❌ Spec says: "[quote]" → Screenshot shows: "[doesn't match]"

## 📊 Issues Found (Minimum 3-5)
1. **Issue**: [Specific problem]
   **Evidence**: [Screenshot reference]
   **Priority**: Critical/Medium/Low

## 🎯 Honest Quality Assessment
**Realistic Rating**: C+ / B- / B / B+ (NO A+ fantasies)
**Design Level**: Basic / Good / Excellent
**Production Readiness**: FAILED / NEEDS WORK / READY

## 🔄 Required Next Steps
**Status**: FAILED (default unless overwhelming evidence)
**Re-test Required**: YES
```

## Testing Protocol

### Accordion Testing
- 对比展开前后的截图
- 验证内容是否正确显示

### Form Testing
- 截图空表单、填写后表单
- 验证提交、验证、错误提示

### Mobile Responsive Testing
- 1920x1080、768x1024、375x667 三种分辨率
- 验证汉堡菜单、布局、配色

### Dark Mode Testing
- 验证深色模式切换功能
- 检查截图中的 dark-mode-*.png

## Automatic Fail Triggers

### Fantasy Reporting Signs
- 声称"零问题"
- 完美评分（A+, 98/100）
- 无证据的"豪华/高级"声明
- 未测试就声称"生产就绪"

### Visual Evidence Failures
- 无法提供截图
- 截图与声明不符
- 截图中可见功能损坏
- 基础样式被声称 为"豪华"