← Back to Skills Marketplace
219
Downloads
0
Stars
0
Active Installs
6
Versions
Install in OpenClaw
/install bu-bu-wei-ying
Description
复杂APP开发统一技能 - 融合敏捷开发、CI/CD、DevOps的最佳实践,核心理念"每步必测、每层必验、环环相扣、层层守护
README (SKILL.md)
步步为营 - 复杂APP开发统一技能
当用户要求进行复杂应用开发、构建 CI/CD 流程、执行 DevOps 任务时,按以下步骤引导。
核心原则
修正原则(最高优先级)
只修正指定问题和相关联代码,不改变其他代码
- 只修改用户明确指出的问题代码
- 只修改与问题直接相关的关联代码
- 已正确运行的功能不得被修改
- 每次修改前确认变更范围
- 修改后验证其他功能未被影响
开发理念
每步必测 → 每次代码变更必须运行测试
每层必验 → 每个模块必须通过验证
环环相扣 → 各阶段必须有序衔接
层层守护 → 每层必须有保护机制
开发流程检查清单
1. 需求分析阶段
- 明确功能需求和非功能需求
- 拆分任务为可执行的子任务
- 预估时间和资源
- 确定验收标准
2. 代码开发阶段
- 创建功能分支
git checkout -b feature/xxx - 编写代码并遵循编码规范
- 编写单元测试(覆盖率 > 80%)
- 运行本地 lint/static analysis
- 进行代码审查
3. 构建阶段
- 本地构建成功
npm run build/docker build - 运行集成测试
- 安全扫描
npm audit/trivy image - 生成构建产物(Docker 镜像/二进制文件)
4. 测试阶段
- 部署到预发布环境
- 执行 E2E 测试
- 性能测试(响应时间、并发)
- 回归测试
- 获取测试报告
5. 发布阶段
- 创建 Release Tag
- 更新 CHANGELOG
- 执行灰度发布(5% → 25% → 100%)
- 监控系统指标
- 准备回滚方案
6. 运维监控阶段
- 确认监控告警正常
- 检查日志聚合
- 验证备份机制
- 文档更新
常用命令参考
包管理
npm install # 安装依赖
npm run build # 构建项目
npm test # 运行测试
npm run lint # 代码检查
npm run audit # 安全审计
Docker
docker build -t app:v1 . # 构建镜像
docker run -d app:v1 # 运行容器
docker-compose up -d # 启动服务栈
docker logs -f \x3Ccontainer> # 查看日志
Git 工作流
git checkout -b feature/xxx # 创建功能分支
git add . && git commit -m "feat: xxx"
git push origin feature/xxx # 推送分支
git merge main # 合并主干
git tag v1.0.0 && git push --tags # 打标签
CI/CD 检查
curl -I http://localhost:PORT/health # 健康检查
netstat -tlnp | grep PORT # 端口检查
tail -100 /var/log/app.log # 查看日志
使用示例
示例1:开发新功能
用户:帮我开发一个用户注册功能
Agent:
1. 创建分支 feature/user-registration
2. 编写 API、数据库模型、单元测试
3. 运行测试确保通过
4. 构建镜像并推送到仓库
5. 执行灰度发布
示例2:排查部署问题
用户:生产环境部署失败了
Agent:
1. 检查构建日志
2. 验证环境变量配置
3. 测试数据库连接
4. 检查网络/端口
5. 执行回滚(如需要)
自动化脚本
检查清单脚本位于 scripts/checklist.sh:
| 命令 | 说明 |
|---|---|
verify-fix \x3C文件> [哈希] |
验证修正范围(核心安全检查) |
dev-check |
完整开发检查(lint + test + build) |
docker-build [镜像] [标签] |
Docker 构建 + 安全扫描 |
health [主机] [端口] [端点] |
HTTP 健康检查 |
canary [镜像] [阶段] |
灰度发布(默认: 5,25,100%) |
rollback [命名空间] [部署] |
回滚到上一个稳定版本 |
coverage [阈值] |
测试覆盖率检查(默认: 80%) |
all |
执行全部检查 |
监控模板
从 templates/ 目录导入监控规则:
Grafana (grafana-alerts.json)
- 应用错误率告警
- 响应延迟告警
- 服务可用性告警
- CPU/内存/磁盘使用率
Datadog (datadog-alerts.json)
- 错误率告警
- 延迟告警
- 健康检查失败
- 基础设施指标
- 日志错误激增检测
验证步骤
每次任务完成后,必须执行:
- 自检清单 - 对照检查清单确认每项完成
- 结果报告 - 向用户汇报关键指标(测试覆盖率、构建状态等)
- 下一步建议 - 提出优化建议或后续行动
Usage Guidance
This skill is a DevOps checklist with a helper script — generally plausible but not fully self-describing. Before installing or running: 1) Inspect scripts (scripts/checklist.sh) yourself and confirm you trust them. 2) Note that the script will call docker, kubectl, trivy, curl, python3, node, docker-compose, netstat, bc, etc.; ensure those tools are present and you understand what credentials (kubeconfig, docker registry login, npm credentials) the script will use. 3) Run first in a safe environment (local sandbox or CI runner with restricted privileges) to avoid accidental access to production clusters or sensitive logs. 4) If you expect it to manage clusters/releases, require explicit documentation of needed permissions and consider least-privilege kubeconfig. 5) If you want to proceed, ask the author/maintainer to update the metadata to list all required binaries and to document credential expectations.
Capability Analysis
Type: OpenClaw Skill
Name: bu-bu-wei-ying
Version: 2.0.0
The 'bu-bu-wei-ying' skill bundle is a well-structured DevOps and CI/CD toolkit designed to guide an AI agent through a standardized software development lifecycle. The instructions in SKILL.md and README.md emphasize safety and precision, specifically through the 'Fix Scope Principle' which prevents the agent from making unintended changes to code. The automation script (scripts/checklist.sh) uses standard industry tools like git, npm, docker, and kubectl for legitimate tasks such as syntax validation, testing, and deployment without any signs of malicious intent, data exfiltration, or obfuscation.
Capability Assessment
Purpose & Capability
Name/description and included files (checklist script, templates) align with a DevOps/CI-CD checklist. However the declared required binaries list (bash, git, npm) is incomplete: the script and SKILL.md also call docker, kubectl, trivy, curl, python3, node, docker-compose, netstat, bc and possibly others. That mismatch is incoherent: a legitimate DevOps skill should declare necessary tooling or document optional features.
Instruction Scope
SKILL.md and scripts stay within DevOps scope: running builds, tests, lint, security scans, health checks, reading logs (/var/log/app.log), invoking kubectl/docker for deploy/rollback, and inspecting git state. These actions require access to local files, container runtime and cluster credentials but are expected for this purpose.
Install Mechanism
Instruction-only skill (no install spec) with shipped helper script. No remote downloads or archive extraction; low install risk from this metadata perspective.
Credentials
The skill declares no required environment variables or primary credential, yet its runtime actions rely on external credentials/config (e.g., docker registry auth, kubeconfig for kubectl, possibly npm registries). This is plausible for a DevOps tool but users should be aware it assumes existing credentials and access without declaring them explicitly.
Persistence & Privilege
always:false and no evidence the skill attempts to persist or modify other skills or system-wide agent settings. It operates as a local script invoked by the agent.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install bu-bu-wei-ying - After installation, invoke the skill by name or use
/bu-bu-wei-ying - Provide required inputs per the skill's parameter spec and get structured output
Version History
v2.0.0
步步为营 2.0.0 重大更新
- 全新中英文文档体系,新增英文文档(SKILL-en.md, README.md)
- 引入自动化脚本(scripts/checklist.sh),支持一键化开发与运维检查
- 新增通用监控模板(Datadog 与 Grafana,templates/ 目录)
- 大幅重构开发与修正规程:首要强调只修正指定问题、严格控制变更范围
- 优化流程化开发、测试、发布与监控清单,增强CI/CD、DevOps最佳实践覆盖
- 明确常用命令、典型场景示例,便于日常工作高效复用
v1.4.0
- 结构全面复查和优化,整体内容重构
- 增加快速参考部分,提升查阅效率
- 更新版本号至1.4.0,并同步“融合”方案说明
- 保留并完善上一版所有核心原则与操作流程
v1.3.0
**步步为营 v1.3.0 Changelog**
- 全面重构文档结构,章节重新编号与归类,逻辑更清晰
- 新增“快速参考”与相关速查表,便于查找修正与决策流程
- 优化修正流程描述,升级为“标准修正流程(7步)”并配备流程图
- 统一各项原则为“APP修正铁律十大原则”,精简表述,便于掌握
- 合理合并与删减重复内容,提升可读性与实操性
v1.2.0
- 扩展了APP开发修正铁律,从七大修正原则升级为“十大修正原则”,新增精准修复、功能守护、修复验证等内容
- 修正检查清单同步补充为十条,强化修复规范和功能验证
- 版本号升级为1.2.0,最终原则与发布元数据同步更新
v1.1.0
- 新增“APP开发修正铁律(核心七原则)”章节,提出适用于Bug修复和改动的七大原则
- 修正原则涵盖最小修改、逻辑简化、行为一致性、单次测试、回滚准备、问题记忆、自动检测等内容
- 更新元数据,增加 version 字段为 1.1.0
- 丰富质量保障与修正流程相关自检与检查清单
- 文档结构优化,进一步明确各阶段关键操作与最佳实践
v1.0.0
步步为营 1.0.0 初始发布
- 首次发布,融合敏捷开发、CI/CD、DevOps及步步为营核心铁律,适用于复杂APP开发。
- 提出“每步必测、每层必验、环环相扣、层层守护”开发守则。
- 收录四层模块化架构及各模块开发顺序。
- 提供敏捷开发流程、优先级管理规范及自检验证清单。
- 包含完整质量保障体系、API验证方法、DevOps工具链及关键监控指标。
- 制定团队协作、代码评审、版本发布与回滚规范及常用命令速查表。
Metadata
Frequently Asked Questions
What is 步步为营?
复杂APP开发统一技能 - 融合敏捷开发、CI/CD、DevOps的最佳实践,核心理念"每步必测、每层必验、环环相扣、层层守护. It is an AI Agent Skill for Claude Code / OpenClaw, with 219 downloads so far.
How do I install 步步为营?
Run "/install bu-bu-wei-ying" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is 步步为营 free?
Yes, 步步为营 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does 步步为营 support?
步步为营 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created 步步为营?
It is built and maintained by smxtx (@smxtx); the current version is v2.0.0.
More Skills