← Back to Skills Marketplace
Agent Regression Testing
by
LuiciferYi
· GitHub ↗
· v1.0.0
· MIT-0
115
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install agent-regression-testing
Description
管理回归测试计划、执行回归测试用例、验证缺陷修复。 Use when: 用户说"开始回归测试"、"验证缺陷修复"、"执行回归验证"。
README (SKILL.md)
Agent Regression Testing - 智能体回归验证
飞书文档写入位置
知识库: 7616288931050507220
节点: 效果评测/回归测试报告
管理回归测试计划、执行回归测试用例、验证缺陷修复。
触发条件
当用户需要以下操作时触发:
- "开始回归测试"
- "验证缺陷修复"
- "执行回归验证"
- "生成回归报告"
- "回归测试计划"
功能
提供回归测试全流程管理:
- 回归测试范围确定
- 回归测试用例选择
- 回归测试计划制定
- 回归测试执行
- 缺陷修复验证
- 回归测试报告生成
使用方法
回归测试计划:
- 版本信息: [版本号]
- 变更内容: [变更描述]
- 影响范围: [影响模块]
- 回归范围: [测试范围]
- 回归用例数: [X]个
- 预计执行时间: [X]小时
回归用例选择:
- 核心功能用例: [X]个
- 缺陷相关用例: [X]个
- 关联模块用例: [X]个
- 基础功能用例: [X]个
缺陷修复验证:
- 缺陷ID: [BUG-XXX]
- 缺陷描述: [描述]
- 修复版本: [版本号]
- 验证结果: [通过/失败]
- 验证备注: [备注]
回归测试执行:
- 执行日期: [日期]
- 执行人: [执行人]
- 测试环境: [环境]
- 执行结果:
- 通过: [X]个
- 失败: [X]个
- 阻塞: [X]个
- 通过率: [X%]
回归结论:
- 是否通过: [是/否]
- 遗留问题: [问题列表]
- 发布建议: [建议]
回归测试策略
1. 全量回归
- 适用场景: 核心版本发布、架构变更、重大重构
- 优点: 覆盖全面,风险最低
- 缺点: 耗时较长,资源消耗大
- 触发条件:
- 版本号升级(如 v1.0 → v2.0)
- 架构级变更(如单体→微服务)
- 核心框架升级(如 Python 3.10 → 3.12)
- 数据库结构变更
2. 选择性回归
- 适用场景: 局部功能修改、缺陷修复、小版本迭代
- 优点: 效率高,针对性强
- 缺点: 可能遗漏关联问题
- 选择标准:
| 优先级 | 选择标准 | 说明 | 示例 |
|---|---|---|---|
| P0 | 修改模块的核心用例 | 直接修改的功能 | 修改了创建接口 → 执行创建相关用例 |
| P0 | 历史缺陷相关用例 | 该模块曾出现缺陷 | 智能体模块曾有问题 → 执行该模块用例 |
| P1 | 与修改相关的功能用例 | 上下游关联功能 | 修改了创建 → 执行列表、详情用例 |
| P1 | 基础功能用例 | 公共基础功能 | 登录、权限、日志等 |
| P2 | 其他模块抽样用例 | 非直接关联模块 | 抽样执行其他模块用例 |
3. 自动化回归
- 适用场景: 每日构建、持续集成、冒烟测试
- 优点: 快速、可重复、无人值守
- 触发时机:
- 每次代码提交后
- 每日定时(如每晚22:00)
- 发布前自动执行
- 手动触发
4. 回归策略选择矩阵
| 变更类型 | 建议策略 | 回归范围 | 预计耗时 |
|---|---|---|---|
| 缺陷修复(单点) | 选择性回归 | 缺陷模块 + 关联模块 | 2-4小时 |
| 功能优化(局部) | 选择性回归 | 修改模块 + 核心功能 | 4-8小时 |
| 新功能上线(模块级) | 选择性回归 | 新模块 + 关联模块 + 核心功能 | 1-2天 |
| 版本迭代(迭代级) | 选择性回归 | 全量P0 + 修改模块P1 | 2-3天 |
| 大版本发布(系统级) | 全量回归 | 所有用例 | 1-2周 |
| 架构重构(架构级) | 全量回归 | 所有用例 + 专项测试 | 2-4周 |
回归测试流程
阶段1:准备阶段
1. 分析变更影响
- 获取变更清单(代码变更、配置变更、数据变更)
- 评估影响范围(模块级、功能级、接口级)
- 识别风险点(高风险功能、历史问题区域)
↓
2. 确定回归范围
- 选择回归策略(全量/选择性/自动化)
- 确定回归用例集
- 评估回归工作量
↓
3. 制定回归计划
- 制定时间表
- 分配资源
- 准备环境
阶段2:执行阶段
4. 执行回归测试
- 冒烟测试(环境可用性)
- 核心功能回归(P0用例)
- 全量回归(根据策略)
- 实时记录结果
↓
5. 验证缺陷修复
- 获取待验证缺陷清单
- 执行缺陷验证用例
- 更新缺陷状态
- 记录验证结果
阶段3:评估阶段
6. 生成回归报告
- 汇总执行结果
- 统计通过率
- 分析失败原因
- 整理遗留问题
↓
7. 风险评估
- 功能风险评估
- 性能风险评估
- 安全风险评估
- 兼容性风险评估
↓
8. 发布决策
- 通过:可以发布
- 有条件通过:修复后发布
- 不通过:不能发布
回归测试准入准出
准入标准(开始回归前必须满足)
| 准入项 | 要求 | 检查方式 |
|---|---|---|
| 代码冻结 | 代码已冻结,不再变更 | 代码分支状态 |
| 冒烟通过 | 核心功能冒烟测试100%通过 | 冒烟测试报告 |
| 环境就绪 | 回归环境已搭建,配置正确 | 环境检查 |
| 用例准备 | 回归用例已确定并评审 | 用例清单 |
| 缺陷修复 | 计划修复的缺陷已修复 | 缺陷状态 |
准出标准(回归通过必须满足)
| 准出项 | 要求 | 说明 |
|---|---|---|
| P0用例通过率 | 100% | 核心功能必须全部通过 |
| P1用例通过率 | ≥95% | 重要功能允许少量失败 |
| P2用例通过率 | ≥80% | 一般功能允许失败 |
| 缺陷修复率 | 100% | 计划修复的缺陷必须验证通过 |
| 新增缺陷 | 无P0级缺陷 | 回归期间未发现严重问题 |
| 遗留问题 | 有明确规避方案 | 遗留问题不影响核心功能 |
发布风险评估
风险等级定义
| 风险等级 | 判定标准 | 处理建议 |
|---|---|---|
| 🟢 低风险 | P0通过率100%,无遗留问题 | 可以发布 |
| 🟡 中风险 | P0通过率100%,有低风险遗留问题 | 修复后发布 |
| 🟠 高风险 | P0通过率\x3C100%,或有P1遗留问题 | 修复后重新回归 |
| 🔴 极高风险 | 有P0级缺陷未修复 | 禁止发布 |
发布决策矩阵
| 回归结果 | 遗留问题 | 建议决策 |
|---|---|---|
| 全量通过 | 无 | ✅ 立即发布 |
| 全量通过 | 低风险问题 | ✅ 修复后发布 |
| 核心通过,一般失败 | 无 | ⚠️ 评估后发布 |
| 核心通过,一般失败 | 有风险问题 | ❌ 修复后重新回归 |
| 核心未通过 | - | ❌ 禁止发布 |
输出
- 回归测试计划
- 回归用例清单
- 回归测试执行记录
- 缺陷修复验证记录
- 回归测试报告
报告模板
# 回归测试报告
## 1. 基本信息
- 版本号: [版本]
- 回归日期: [日期]
- 执行人: [执行人]
- 测试环境: [环境]
## 2. 变更说明
- 变更内容: [描述]
- 影响范围: [范围]
## 3. 回归范围
- 回归用例总数: [X]个
- 核心功能用例: [X]个
- 缺陷相关用例: [X]个
- 关联模块用例: [X]个
## 4. 执行结果
- 通过: [X]个 ([Y]%)
- 失败: [X]个 ([Y]%)
- 阻塞: [X]个 ([Y]%)
- 未执行: [X]个 ([Y]%)
## 5. 缺陷修复验证
| 缺陷ID | 描述 | 验证结果 | 备注 |
|--------|------|----------|------|
| BUG-001 | [描述] | 通过 | - |
## 6. 遗留问题
- [问题列表]
## 7. 结论与建议
- 回归结论: [通过/不通过]
- 发布建议: [建议]
Usage Guidance
This skill is an instruction-only regression testing assistant and appears coherent with its purpose. Before installing, confirm: (1) you are comfortable the skill will create/update Feishu docs at the hardcoded knowledge-base/node (verify those IDs and scope), (2) your agent's Feishu connector permissions are appropriate (document creation/update and messaging), and (3) if you expect the agent to collect change lists or fetch PR/issue data, ensure any repository or issue-tracker connectors have the least-privilege access required. Because the SKILL.md is high-level, test it in a non-production environment first to verify where and how it will write reports and what external systems it will query.
Capability Analysis
Type: OpenClaw Skill
Name: agent-regression-testing
Version: 1.0.0
The skill bundle is a legitimate tool for managing software regression testing processes, including test planning, execution, and reporting. It utilizes Feishu (Lark) document tools (feishu_create_doc, feishu_update_doc) to store reports in a specified knowledge base, which aligns with its stated purpose in SKILL.md. No malicious code, data exfiltration, or harmful prompt injection attempts were found.
Capability Assessment
Purpose & Capability
Name, description, and instructions all describe regression test planning, execution, defect verification, and report generation. The allowed Feishu doc tools align with the documented behavior (writing reports to a specified knowledge base/node). No unrelated binaries, env vars, or config paths are requested.
Instruction Scope
SKILL.md contains detailed templates and step-by-step guidance for regression testing and explicitly specifies a Feishu knowledge base/node to write reports. The instructions do not direct reading arbitrary local files or exfiltration. One implicit item: steps like “获取变更清单” (get change list) assume the agent may fetch project artifacts (PRs, changelogs, issue lists) if connectors are available; the doc does not instruct how to obtain those, so actual data access depends on platform connectors and permissions.
Install Mechanism
No install spec and no code files beyond package.json and SKILL.md — it is instruction-only, so nothing is written to disk by an installer.
Credentials
The skill declares no required environment variables or credentials. It relies on platform-provided tools for Feishu operations (feishu_create_doc, feishu_update_doc, etc.). Those tool permissions are expected for a reporting/Doc-writing skill; nothing in the SKILL.md requests unrelated secrets or additional credentials.
Persistence & Privilege
always is false and there is no indication the skill modifies other skills, writes persistent agent-wide configuration, or requests elevated system privileges. Its scope is limited to generating and writing testing artifacts (e.g., Feishu docs).
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agent-regression-testing - After installation, invoke the skill by name or use
/agent-regression-testing - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
- Initial release of the agent-regression-testing skill.
- Supports end-to-end regression testing management: planning, test case execution, defect verification, and report generation.
- Provides detailed guidelines and templates for regression test plans, execution records, defect validation, and results reporting.
- Offers selection criteria and best practices for full, selective, and automated regression strategies.
- Includes clear entry/exit criteria, risk assessment, and publishing decision matrix for regression testing.
- Integrates with Feishu document tools for documentation and reporting.
Metadata
Frequently Asked Questions
What is Agent Regression Testing?
管理回归测试计划、执行回归测试用例、验证缺陷修复。 Use when: 用户说"开始回归测试"、"验证缺陷修复"、"执行回归验证"。 It is an AI Agent Skill for Claude Code / OpenClaw, with 115 downloads so far.
How do I install Agent Regression Testing?
Run "/install agent-regression-testing" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Regression Testing free?
Yes, Agent Regression Testing is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Agent Regression Testing support?
Agent Regression Testing is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Regression Testing?
It is built and maintained by LuiciferYi (@luiciferyi); the current version is v1.0.0.
More Skills