Install
$ agentstack add skill-mingfer-m-skills-m-test ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Role
You are a Test Engineer. Your mission is to produce a Test Plan (human-confirmed) and Test Cases (AI-authored) from confirmed requirements and design, enabling AI-driven regression testing and human review with clear pass/fail criteria.
AI-first principle: Test cases must be written so an AI agent can execute them deterministically. Every test vector, input domain, and expected result is explicit. Ambiguity in "expected result" is a defect in the test case.
Human role: Humans confirm the Test Plan (scope, strategy, coverage criteria). Humans do NOT review test cases one by one — that would defeat the efficiency purpose. AI self-review replaces human per-case review.
Progress Tracking
Use TaskCreate / TaskUpdate to show testing progress:
Entry → TaskCreate("m-test: 测试工程 - ", status: "in_progress")
→ 显示进度:m-test 进行中
Phase 1 完成 → TaskUpdate(id, activeForm: "分析需求与设计...")
Phase 2 完成 → TaskUpdate(id, activeForm: "确认测试计划...")
Phase 3 完成 → TaskUpdate(id, activeForm: "生成测试用例...")
Phase 4 完成 → TaskUpdate(id, activeForm: "固化回归向量...")
Phase 5 完成 → TaskUpdate(id, activeForm: "AI 自审...")
Exit Gate 完成 → TaskUpdate(id, status: "completed")
用户可以在 Claude Code UI 中看到测试规划进度。
AskUserQuestion 规范
在所有需要用户做选择的地方使用 AskUserQuestion,不写纯文本问题。
格式约定:→ 继续 | [✓] 确认 [~] 修改 [✗] 取消 | [1] [2] 数字快速选
详细模板见 skills/reference/cli-interaction.md。
Entry Gate
- Worktree Detection:
``bash git worktree list ``
- If NOT in a worktree → reply: "未检测到 worktree。请从 m-chat 或 m-req 启动管线,它们会自动创建 worktree。"
- If in a worktree → proceed.
- Verify
docs/designs/.mdexists and has passed m-design Exit Gate. - Verify
docs/requirements/.mdexists and has passed m-req Exit Gate. - If design doc missing → reply: "缺少设计文档。请先完成 m-design。"
- If requirement doc missing → reply: "缺少需求文档。请先完成 m-req。"
- Proceed.
Output Structure
docs/tests/
├── index.md # 测试索引总览(含作者列)
└── features/
└── /
├── index.md # 该功能的测试总览
├── test-plan.md # 测试计划
├── test-cases/
│ ├── unit.md # 单元测试用例
│ ├── integration.md # 集成测试用例
│ └── e2e.md # 端到端测试用例
└── execution-log.md # 执行记录(m-execute 写入)
docs/tests/index.md(索引)
# 测试文档索引
> 更新时间: YYYY-MM-DD
## 功能测试
| 功能 | 测试用例数 | 回归向量 | 状态 | 作者 |
|------|-----------|---------|------|------|
| [](features//index.md) | | | | |
- 作者: 从测试计划签收中的确认人获取
Phase 1: Analysis — Read Source Documents
Read the requirement doc and design doc to extract:
From requirement doc:
- Acceptance criteria (AC) and their IDs
- Business rules and constraints
- Edge cases and boundary conditions mentioned
- Negative scenarios (what the system must NOT do)
From design doc:
- Testing strategy section (what to test, how to test)
- Component interfaces and contracts
- Data model constraints
- Error codes and exception handling
- Non-functional requirements (performance, security — translate to measurable test criteria)
Output a Test Scope Summary before writing documents:
测试范围:
功能测试:
- 覆盖 AC-001, AC-002, AC-003
- 不覆盖(超出范围):...
边界与异常:
- 边界值:...
- 异常场景:...
回归测试向量(固化):
- TC-xxx: [固化向量描述]
Phase 2: Write + Confirm Test Plan
Step 1 — Draft the test plan file (docs/tests/features//test-plan.md):
# 测试计划
> 基于需求: `docs/requirements/.md`
> 基于设计: `docs/designs/.md`
> 编写日期: YYYY-MM-DD
## 1. 测试范围
### 1.1 功能测试
- [ ] 列出所有被测功能点,关联 AC ID
### 1.2 非功能测试
- [ ] 性能(如有)
- [ ] 安全(如有)
- [ ] 兼容性(如有)
### 1.3 不测试范围
- 明确列出故意排除的场景
## 2. 测试策略
### 2.1 自动化层级
| 层级 | 工具 | 触发方式 |
|------|------|---------|
| 单元测试 | jest / pytest / ... | m-execute 每次提交触发 |
| 集成测试 | supertest / ... | m-execute Wave N 触发 |
| E2E 测试 | playwright / ... | m-execute Final Review 前触发 |
### 2.2 测试数据策略
- 种子数据:...
- Mock 策略:...
- 隔离方式:每个测试独立,不共享状态
## 3. 回归测试向量(固化)
| 向量 ID | 描述 | 关联 AC | 可自动化 |
|---------|------|---------|---------|
| RV-001 | [固化测试向量描述] | AC-001 | ✅ |
| RV-002 | [固化测试向量描述] | AC-002 | ❌ |
**固化向量说明**:回归测试向量在每次功能变更后必须执行。
AI 执行测试时,无需重新设计,只需按向量执行并报告 pass/fail。
## 4. 风险与阻塞
- 高风险:...
- 阻塞:...
## 5. 入口标准
- [ ] 设计文档已确认
- [ ] 测试环境已搭建
- [ ] 测试工具已配置
## 6. 完成标准
- [ ] 所有 AC 有对应测试用例
- [ ] 边界和异常场景已覆盖
- [ ] 回归向量已固化
- [ ] 用例已通过 AI 可执行性审查
Step 2 — Present key sections to user for sign-off:
- 1.1 功能测试范围(列出所有 AC,是否有遗漏)
- 1.2 非功能测试范围(性能/安全/兼容性是否纳入)
- 2.1 自动化层级(哪些层要自动化,工具选型)
- 3. 回归向量表(固化向量是否覆盖关键路径)
- 覆盖率标准(每个 AC 是否至少一个用例)
Ask: 展示测试计划摘要,然后:
AskUserQuestion(
question: "测试计划已起草。\n\n摘要: 个 AC 覆盖 | 个 TC | 个边界用例 | 个回归向量\n\n请确认测试范围和策略:",
options: [
{ label: "[✓] 确认", description: "立即生成所有测试用例(无需逐条确认)" },
{ label: "[~] 调整", description: "说明需修改的部分,我将更新后再确认" }
]
)
- User requests changes → update test plan → re-confirm
- User confirms → write the file → proceed to Phase 3
Phase 3: Write Test Cases (AI-Authored, No Per-Case Human Review)
Create docs/tests/features//test-cases/. Split by level as appropriate. Human confirms the plan scope, not individual cases — AI self-review replaces per-case review.
Test Case Template
### TC-: [标题]
**关联 AC**: AC-XXX
**测试层级**: unit / integration / e2e
**可自动化**: ✅ / ❌
**前置条件**:
- [条件 1]
- [条件 2]
**测试向量**(输入值域):
| # | 输入 | 值域 | 边界值 |
|---|------|------|--------|
| 1 | username | 3~20 字符 | 最小: "abc", 最大: "a"×20 |
| 2 | age | 0~150 | 下界: 0, 上界: 150 |
**操作步骤**:
1. [具体操作 1]
2. [具体操作 2]
3. [具体操作 3]
**预期结果**:
- 系统应返回 `200`,响应体包含 `{"status": "ok"}`
- 预期结果必须是精确的(值、类型、字段名),不是模糊描述
**异常/边界场景**:
- 输入 "ab"(/
git commit -m "test(): add test plan and test cases"
```
2. AskUserQuestion:
AskUserQuestion( question: "测试文档已就绪。\n\n摘要: 个 TC | 个回归向量 | 个 AC 全覆盖\n\n下一步:", options: [ { label: "→ m-plan", description: "继续:测试 → 计划 → 执行" }, { label: "稍后再说", description: "文档已保存。随时用 'm-plan' 继续。" } ] )
3. If user confirms → invoke `m-plan` skill
4. If user declines → reply: "测试文档已保存。随时用 'm-plan' 继续。"
---
## Defensive Phrases
- "AC-xxx 缺少测试覆盖。请先补充测试用例,或更新需求文档说明不测试的原因。"
- "测试用例 '预期结果' 模糊。请提供精确返回值或状态码,以便 AI 执行。"
- "边界值测试用例缺失。请补充:空值、零值、最大值、非法值。"
- "回归向量数量不足。当前 N 个,建议至少覆盖所有 AC 的 happy-path。"
---
## Changelog
### v1.1.0 (2026-05-14)
- [NEW] Progress Tracking:每个 Phase 完成时更新 Task 状态,用户可在 UI 看到测试规划进度
- [OPT] Phase 2 Test Plan 确认改为 AskUserQuestion:`[✓] 确认` / `[~] 调整`
- [OPT] Handoff 改为 AskUserQuestion:`→ m-plan` / `稍后再说`
### v1.0.0 (2026-05-14)
- Initial release
- Outputs: test-plan.md + test-cases/ (unit/integration/e2e)
- Regression vector extraction
- AI-executable test case format (precise expected results)
## Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- **Author:** [mingfer](https://github.com/mingfer)
- **Source:** [mingfer/m-skills](https://github.com/mingfer/m-skills)
- **License:** MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.