testcase-generator

Test Case Generator (testcase-generator)

1. Objective

Generate structured test cases based on test points, covering positive, negative, and boundary scenarios.

2. Input and Output

Input：test-case/plan.md、clarified-requirements/index.md、CLAUDE.md
Output：test-case/{ITEM}/{POINT}.md、test-case/all_cases.md

3. Core Principles

Strategy Guidance: Provide judgment criteria, not a mandatory quantity
Scenario-Driven: Generate test cases based on actual scenario complexity
Specific Data: Test data must be specific, do not use placeholders
Verifiability: Expected results must be clear and verifiable

Trigger Conditions

User executes the /testcase-gen command
test-case/plan.md exists

4. Core Theories

4.1 Equivalence Class Partitioning

Divide input data into several equivalence classes, and select representative data from each class for testing.

Application Example:

POINT: 用户名密码登录

输入项: 用户名、密码

用户名等价类:
  有效: [6-20字符,字母数字下划线] → "test_user123"
  无效: [<6] "ab" | [>20] "verylongusername12345" | [特殊字符] "user@name" | [空] ""

密码等价类:
  有效: [8-20字符,含大小写+数字] → "Test1234"
  无效: [<8] "Test12" | [缺大写] "test1234" | [空] ""

生成原则: 一次测试一个等价类,其他保持有效

4.2 Boundary Value Analysis

Boundaries are the most error-prone places.

Boundary Three-Point Method: off point (min-1) | on point (min) | inner point (middle value) | on point (max) | off point (max+1)

Application Example:

用户名长度 6-20 字符:
- 离点-下: 5字符 "abcde" → 失败
- 上点-下: 6字符 "abcdef" → 成功
- 内点: 13字符 "test_user_123" → 成功
- 上点-上: 20字符 "test_user_1234567890" → 成功
- 离点-上: 21字符 "test_user_12345678901" → 失败

4.3 Judgment of Number of Test Cases

There is no mandatory quantity; adjust adaptively according to scenario complexity.

Judgment Criteria：

Number of input items: the more input items, the more test cases
Business rule complexity: the more complex the rules, the more test cases
Risk level: Critical and High require more test cases
Test concerns: the more concerns, the more test cases

5. Test Case Format Specification

5.1 Markdown Text Protocol Format (v0.2)

Generate one .md file for each test point, with the following format:

## [P1] 用例标题
[测试类型] 功能
[前置条件] 前置条件描述
[测试步骤] 1. 步骤1。2. 步骤2
[预期结果] 1. 预期1。2. 预期2

## [P3][反向] 用例标题
[测试类型] 功能
[前置条件] 前置条件描述
[测试步骤] 1. 步骤1。2. 步骤2
[预期结果] 1. 预期1。2. 预期2

5.2 Field Description

Priority: P1-P5 (required)
- P1: Core function positive flow
- P2: Basic function positive flow
- P3: Core function exception scenarios
- P4: Boundary conditions
- P5: Low-frequency scenarios
[Negative]: Negative test case marker (optional)
Test Type: Function, Interface, Performance, Security, Compatibility, Usability, Installation and Deployment, Reliability, Localization, Maintainability, Scalability, Configuration (choose 1 of 12)
Preconditions: Preparation work before testing
Test Steps: Operation steps, separated by periods
Expected Results: Expected results, corresponding one-to-one with the steps

5.3 Test Type Enumeration (12 types)

功能
兼容性
易用性
性能
稳定性
安全性
可靠性
效果(AI类、资源类)
效果(硬件器件类)
可维护性
可移植性
埋点

Note:

Do not write "functional testing"; it must be "Function"
Parentheses must be Chinese parentheses
Case-sensitive

5.4 Step Format Specification

Recommended Format: Use Chinese periods as separators, with continuous numbering

[测试步骤] 1. 打开登录页面。2. 输入用户名test_user。3. 输入密码Test1234。4. 点击登录按钮
[预期结果] 1. 页面正常显示。2. 用户名输入框显示内容。3. 密码输入框显示密码。4. 登录成功,跳转到首页

Important: Numbering must start from 1 and increase continuously; the number of test steps and expected results must be consistent.

5.5 Test Data Requirements

Must be specific, do not use placeholders:

Correct example：

Username: test_user
Password: Test1234
Phone number: 13800138000
Amount: 99.99

Incorrect example：

Username: {valid_username}
Password: {password}
Phone number: {phone}
Amount: {amount}

5.6 Expected Result Requirements

Must be clearly verifiable:

Correct example：

Jump to the /home page, with "Welcome, test_user" displayed at the top
Prompt "Password length is insufficient, please enter a password of 8-20 characters"
The order status changes to "Paid", and inventory decreases by 1

Incorrect example：

Login successful
Prompt an error message
The system processes normally

5.7 Directory Structure

test-case/
├── plan.md
├── {ITEM}/
│   ├── {POINT1}.md
│   ├── {POINT2}.md
│   └── ...
└── all_cases.md

5.8 Merge Rules

Merge all {ITEM}/{POINT}.md into all_cases.md：

Group by ITEM
Sort by POINT under each ITEM
Add separator markers:# ITEM 名称 and ## POINT 名称

6. Test Case Generation Process

Step 1: Prepare context

cat test-case/plan.md
cat clarified-requirements/index.md
cat CLAUDE.md

Step 2: Understand test points

For each POINT:

Read the POINT name, risk level, and test focus
Read the relevant descriptions in the requirements document

Step 3: Identify input items

List all input items (fields, parameters, operations, etc.)
Identify the constraints of the input items

Step 4: Divide equivalence classes and boundary values

Valid equivalence class: input that conforms to the rules
Invalid equivalence class: input that violates the rules
Boundary values: critical points (minimum value, maximum value, critical value)

Step 5: Design test cases

Design based on the following strategy:

Core thinking framework:

Input analysis
- Which input items does this scenario involve?
- What constraints does each input have (length, format, range)?
Equivalence class identification
- Valid equivalence class: typical values under normal conditions
- Invalid equivalence class: various abnormal situations (empty values, out of range, format errors)
Test value evaluation
- If this equivalence class fails, will it expose different defects?
- If the test logic is the same as other equivalence classes, can it be merged?
- How can the scenarios mentioned in the test focus be reflected in the test cases?
Coverage judgment
- There is at least one test case for the valid flow
- Core exception scenarios are covered
- Boundary conditions are selected according to the actual situation
Test case design
- Core positive test case (P1): the most common success scenario
- Basic positive test case (P2): other valid equivalence classes
- Core exception test case (P3): key failure scenarios
- Boundary condition test case (P4): boundary value testing
- Low-frequency scenario test case (P5): uncommon but must-cover scenarios

Test case quality standards:

Specific data ("test123", not "correct username")
Expected result is verifiable ("displays error message 'Username cannot be empty'", not "shows error")
Naming convention (starts with "Verify")

Step 6: Unified validation after generation is complete

After each test item (ITEM) is generated, call the validation script to check the format
After all test items are generated, call the validation script again for an overall check
Validation script path:.claude/skills/testcase-generator/scripts/validate.py

7. Examples

Example 1: Correct test cases (username and password login)

## [P1] 验证有效用户名和密码登录成功
[测试类型] 功能
[前置条件] 已注册用户test_user，密码Test1234
[测试步骤] 1. 输入用户名test_user，输入密码Test1234，点击登录。2. 验证跳转到首页并显示用户名
[预期结果] 1. 登录请求成功。2. 跳转到/home页面，顶部显示'欢迎，test_user'

## [P3][反向] 验证用户名为空时登录失败
[测试类型] 功能
[前置条件] 进入登录页
[测试步骤] 1. 不输入用户名，输入密码Test1234，点击登录。2. 验证提示'用户名不能为空'
[预期结果] 1. 登录请求被拒绝。2. 停留在登录页，用户名输入框下方显示红色提示'用户名不能为空'

## [P4] 验证密码长度边界值（8位）
[测试类型] 功能
[前置条件] 已注册用户test_user，密码Test1234
[测试步骤] 1. 输入用户名test_user，输入8位密码Test1234，点击登录。2. 验证登录成功
[预期结果] 1. 登录请求成功。2. 跳转到/home页面

Example 2: Comprehensive scenario (multiple input items)

POINT: Ad account list - filter function

Input item analysis:

Time range: [Today|Yesterday|Last 7 days|Custom]
Account name: [text fuzzy search]
Account status: [Normal|Banned|Under review]

Generate test cases (6):

## [P1] 验证单一条件筛选成功
[测试类型] 功能
[前置条件] 广告账户列表页面，有10个账户
[测试步骤] 1. 选择时间"近7天"，点击查询。2. 验证返回符合条件的账户
[预期结果] 1. 列表刷新。2. 仅显示近7天有数据的账户

## [P1] 验证多条件组合筛选成功
[测试类型] 功能
[前置条件] 同上
[测试步骤] 1. 选择时间"近7天"，输入名称"test"，选择状态"正常"，点击查询
[预期结果] 1. 列表刷新。2. 仅显示同时满足3个条件的账户

## [P3][反向] 验证时间超过365天筛选失败
[测试类型] 功能
[前置条件] 同上
[测试步骤] 1. 选择自定义时间，开始日期2023-01-01，结束日期2024-01-02，点击查询
[预期结果] 1. 显示提示"时间范围不能超过365天"，列表不刷新

## [P3][反向] 验证不存在的账户名筛选返回空
[测试类型] 功能
[前置条件] 同上
[测试步骤] 1. 输入不存在的账户名"nonexistent123"，点击查询
[预期结果] 1. 列表刷新。2. 显示"暂无数据"

## [P2] 验证筛选条件清空功能
[测试类型] 功能
[前置条件] 已应用筛选条件
[测试步骤] 1. 点击"重置"按钮。2. 验证所有筛选条件清空，列表恢复默认
[预期结果] 1. 所有筛选项恢复默认值。2. 列表显示全部账户

## [P4][反向] 验证结束时间早于开始时间提示错误
[测试类型] 功能
[前置条件] 同上
[测试步骤] 1. 选择自定义时间，开始日期2024-01-02，结束日期2024-01-01，点击查询
[预期结果] 1. 显示提示"结束时间不能早于开始时间"

Description:

Medium-complexity POINT (multiple input items) generates 6 test cases
Coverage: valid combinations, invalid boundaries, clear operations
Selectrepresentative test cases, covering the main scenarios

Example 3: Incorrect test cases

## [P1] 验证登录成功
[测试类型] 功能
[前置条件] 用户已注册
[测试步骤] 1. 输入用户名{username}，输入密码{password}，点击登录
[预期结果] 1. 登录成功

Why it is incorrect：

Test data uses placeholders ({username}, {password})
Expected result is not verifiable ("Login successful" is too vague)

8. Validation process

Step 1: After each test item is generated

Call the validation script to check the format:

uv run .claude/skills/testcase-generator/scripts/validate.py \
  --single "test-case/{模块}/{测试点}.md"

The script outputs logs indicating format issues (such as incorrect priority, missing fields, etc.). The AI analyzes the logs and decides whether modifications are needed.

Step 2: After all generation is complete

Call the validation script again for an overall check:

uv run .claude/skills/testcase-generator/scripts/validate.py "test-case/" --check-duplicates

The script checks for duplicate test cases, format consistency, etc. The script only provides logs and does not directly modify files. The AI analyzes the logs and decides whether adjustments are needed.

Step 3: Generate quality report

Count test cases (by ITEM, by priority)
Count equivalence class coverage
Count boundary value coverage
List duplicate test cases (if any)

9. Checklist

Pre-generation check

Read test-case/plan.md
Read clarified-requirements/index.md
Read CLAUDE.md business context
Understand the hierarchical relationship between test items (ITEM) and test points (POINT)

Check when generating each POINT

A test case file has been generated for each POINT
The number of test cases is reasonable (based on scenario complexity)
All test cases have a priority (P1-P5)
All test cases have a test type (one of 12)
Test data is specific (no placeholders)
Expected results are verifiable (no vague descriptions)
The number of test steps matches the number of expected results
Test case titles start with "Verify"

Governance phase check

The global validation script was executed
Duplicate test cases were checked
The format complies with the specification (already checked by validate.py)
all_cases.md has been generated
No duplicate test cases (already checked by validate.py --check-duplicates)
Quality report generated

10. Script Interface

validate.py

# 单文件校验
uv run .claude/skills/testcase-generator/scripts/validate.py \
  --single "test-case/{模块}/{测试点}.md"

# 全局校验
uv run .claude/skills/testcase-generator/scripts/validate.py "test-case/"

# 全局校验 + 重复检测
uv run .claude/skills/testcase-generator/scripts/validate.py "test-case/" --check-duplicates

# 仅重复检测
uv run .claude/skills/testcase-generator/scripts/validate.py "test-case/" --duplicates-only

Validation Item:

Schema format
Required fields (priority, test type, test steps, expected results)
Test type enumeration (12 types)
Step number continuity
Consistency between number of steps and expected results
Equivalence class coverage check (--check-equivalence)
Boundary value coverage check (--check-boundary)
Duplicate detection (--check-duplicates)

to_excel.py

# 从目录导出
uv run .claude/skills/testcase-generator/scripts/to_excel.py "test-case/" -o "test-case/export.xlsx"

# 从单个文件导出
uv run .claude/skills/testcase-generator/scripts/to_excel.py "test-case/{模块}/{测试点}.md" -o "output.xlsx"

11. Exception Handling

Error	Handling Method
Single file validation failed	Retry 3 times; if still failed, skip and log
Too many test cases (>15)	Prompt to check for redundant equivalence classes
Missing boundary test cases	Warn and recommend adding
Missing invalid equivalence classes	Warn and recommend adding exception scenarios
Step numbers not continuous	Report error and require correction
Test type not in enumeration	Report error and require correction
Inconsistent number of steps and expected results	Report error and require correction
Export failed	Check openpyxl dependency

12. Stopping Point

✅ Test case generation completed

Artifacts:

test-case/{ITEM}/{POINT}.md - Test case files for each test point
test-case/all_cases.md - Summary of all test cases

Quality Assurance:

Generated based on equivalence class theory
Full boundary value coverage
Reasonable number of test cases (based on scenario complexity)

Statistics Example:

📊 生成统计:
  - 测试项: 8个
  - 测试点: 20个
  - 测试用例: 75个

  覆盖率:
  - 有效等价类: 100% (每POINT至少1个)
  - 无效等价类: 80% (主要异常全覆盖)
  - 边界值: 90% (所有范围型输入)

  用例分布:
  - P1(核心正向): 20个(27%)
  - P2(基本正向): 15个(20%)
  - P3(核心异常): 25个(33%)
  - P4(边界条件): 12个(16%)
  - P5(低频场景): 3个(4%)

Next Steps:

Manually review test cases
Export to Excel format (optional)
Adjust test cases based on business understanding