prd-auto-test-loop

PRD automated testing closed loop

Overview

Use this skill to fix release testing as:主回归基线 + PRD 增量测试。

Maintain one shared master regression test suite across versions. Each PRD version only adds test cases related to new or changed functionality. Output artifacts to 自动化测试/<version>/。

Workflow

1. Lock the scope from the PRD.

• Extract the version goals, test scope, out-of-scope, and acceptance criteria.
• If this is an incremental change, reject "a full rewrite of the test suite".

1. Define test layers and boundaries.

• Map acceptance criteria to Unit / Integration / E2E。
• Read references/自动化测试封装指南.md 's boundary section, mark A/H/A+H。

1. Initialize the version test directory.

• Read references/自动化测试封装指南.md 's initialization commands and templates.
• Create 自动化测试/<version>/, and generate TEST_PLAN.md、TEST_REPORT.md。

1. Implement incremental tests.

• Prioritize reusing existing master regression test cases.
• Only add PRD difference test cases to tests/。
• Keep assertions deterministic and avoid fragile UI selectors.

1. Execute the AI self-testing closed loop.

• Fixed order:Unit -> Integration -> E2E。
• On failure, fix by root cause and rerun (up to 3 rounds).
• If blocked by the environment, clearly record the blockers and stop claiming a pass.

1. Produce the test report.

• Fill in the command, pass rate, failure items, and remaining risks according to the template.
• Output the release gate conclusion:PASS or FAIL。

Rules

• Do not rebuild the entire test suite from scratch for every PRD.
• You must maintain both the "master regression suite + version incremental suite".
• Each version must be bound to its own TEST_PLAN and TEST_REPORT。
• Reproducible commands and result summaries must be provided; vague conclusions are prohibited.
• Do not claim to cover all quality risks; you must list manual supplementary test items.

Automation boundary matrix

Suitable for automation (AI coverage recommended)

• Functional correctness (state transitions, API handling, validation rules)
• Regression paths (prevent regressions of historical defects)
• Deterministic boundary scenarios (null values, invalid values, duplicate operations)
• Stable smoke processes (startup, key navigation, save/submit)
• Quantifiable non-functional metrics (timeouts, error rates, crash rates)

Prioritize manual verification (manual fallback recommended)

• Usability and interaction smoothness
• Visual quality and brand consistency
• Rationality of business strategy
• Open-ended exploratory paths
• Product trade-offs and release priority judgment

Acceptance criteria labeling method

• A: fully automatable
• H: manual verification only
• A+H: automated + manual joint verification

TEST_PLAN template

# TEST_PLAN（<version>）

## 1. 测试范围
- PRD：<path>
- 范围内：
- 非范围：

## 2. 完成门槛
1. 计划内用例：<x>/<x> 通过
2. P0 用例通过率：100%
3. 连续两轮稳定通过
4. 无阻断/严重缺陷遗留

## 3. 用例清单
### Unit
- UT-xx：

### Integration
- IT-xx：

### E2E
- E2E-xx：

## 4. 执行顺序
1. Unit
2. Integration
3. E2E
4. 自动修复循环（最多 3 轮）

## 5. 输出产物
- TEST_REPORT.md
- 命令执行结果摘要

TEST_REPORT template

# TEST_REPORT（<version>）

## 1. 结果摘要
- 日期：
- PRD：
- 结论：PASS | FAIL

## 2. 执行命令与结果
- <command>
  - result:

## 3. 分层覆盖结果
- Unit：passed/total
- Integration：passed/total
- E2E：passed/total

## 4. 失败用例
- 用例 ID：
- 现象：
- 根因：
- 修复状态：

## 5. 剩余风险（人工补测）
- 风险点：
- 自动化无法完全覆盖原因：

## 6. 发布门禁
- 门禁检查状态：
- 最终决策：

Version directory initialization command

Execute in the project root directory:

version="V0.6"
base_dir="自动化测试/$version"
mkdir -p "$base_dir/tests/unit" "$base_dir/tests/integration" "$base_dir/tests/e2e"

cat > "$base_dir/TEST_PLAN.md" <<'TPL'
# TEST_PLAN（V0.6）

## 1. 测试范围
- PRD：
- 范围内：
- 非范围：

## 2. 完成门槛
1. 计划内用例全部通过
2. P0 用例通过率 100%
3. 连续两轮稳定通过
4. 无阻断/严重缺陷遗留
TPL

cat > "$base_dir/TEST_REPORT.md" <<'RPT'
# TEST_REPORT（V0.6）

## 1. 结果摘要
- 日期：
- 结论：
RPT