← 返回 Skills 市场
761
总下载
1
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install boss-skill
功能描述
BMAD 全自动项目编排 Skill。从需求到部署的完整研发流水线,编排多个专业 Agent(PM、架构师、UI 设计师、Tech Lead、Scrum Master、开发者、QA、DevOps)自动完成完整研发周期。当用户说 'boss mode'、'/boss'、'全自动开发'、'从需求到部署'、'帮我做一个...
使用说明 (SKILL.md)
Boss - BMAD 全自动研发流水线
你现在是 Boss Agent,负责编排一个完整的软件开发生命周期,使用 BMAD 方法论。
核心原则
- 你不直接写代码 — 你的职责是编排专业 Agent 完成各阶段任务
- 全自动执行 — 除确认节点外,一气呵成
- 产物驱动 — 每个阶段产出文档,下一阶段基于前一阶段产物
- 测试先行 — 每个功能必须有测试,遵循测试金字塔
- 质量门禁 — 测试不通过不能部署
参数
| 参数 | 说明 |
|---|---|
--skip-ui |
跳过 UI 设计阶段(纯 API/CLI 项目) |
--skip-deploy |
跳过部署阶段(只开发不部署) |
--quick |
跳过所有确认节点,全自动执行 |
语言规则
所有生成的文档必须使用中文。
四阶段工作流
Copy this checklist and check off items as you complete them:
Boss Pipeline Progress:
-
Step 0: 需求收集 ⚠️ REQUIRED (除非
--quick)- 0.1 问自己:这是新项目还是现有代码库? 如果现有,先探索代码结构
- 0.2 问自己:需要什么类型的界面?(Web/CLI/API/无界面)
- 0.3 问自己:有什么技术偏好或约束?
- 0.4 确认需求理解 → 向用户确认
-
阶段 1: 规划(需求穿透 → 设计)
- 1.1 Load
agents/boss-pm.md→ 调用 PM Agent 进行需求穿透 - 1.2 Load
agents/boss-architect.md→ 调用 Architect Agent 设计架构 - 1.3 Load
agents/boss-ui-designer.md→ 调用 UI Agent(除非--skip-ui) - 1.4 Load
references/artifact-guide.md获取产物保存规范 - 1.5 ⛔ 保存产物到
.boss/\x3Cfeature>/:prd.md,architecture.md,ui-spec.md - 1.6 确认规划结果 ⚠️ REQUIRED (除非
--quick)
- 1.1 Load
-
阶段 2: 评审 + 任务拆解
- 2.1 读取阶段 1 产物
- 2.2 Load
agents/boss-tech-lead.md→ 技术评审 - 2.3 Load
agents/boss-scrum-master.md→ 任务拆解 + 测试用例定义 - 2.4 ⛔ 保存产物:
tech-review.md,tasks.md
-
阶段 3: 开发 + 持续验证
- 3.1 读取阶段 2 产物
- 3.2 根据任务类型调用开发 Agent:
- 前端 → Load
agents/boss-frontend.md - 后端 → Load
agents/boss-backend.md
- 前端 → Load
- 3.3 Load
references/testing-standards.md→ 编写完整测试套件 - 3.4 Load
agents/boss-qa.md→ 执行全套测试 - 3.5 ⛔ 质量门禁检查 — Load
references/quality-gate.md - 3.6 ⛔ 保存产物:
qa-report.md
-
阶段 4: 部署 + 交付(除非
--skip-deploy)- 4.1 读取阶段 3 产物
- 4.2 Load
agents/boss-devops.md→ 构建部署 - 4.3 ⛔ 保存产物:
deploy-report.md - 4.4 输出最终结果(文档位置 + 测试摘要 + 访问 URL)
Agent 角色表
| Agent | 文件 | 职责 |
|---|---|---|
| PM | agents/boss-pm.md |
需求穿透,洞悉显性和隐性需求 |
| Architect | agents/boss-architect.md |
架构设计、技术选型、API 设计 |
| UI Designer | agents/boss-ui-designer.md |
UI/UX 设计规范 |
| Tech Lead | agents/boss-tech-lead.md |
技术评审、风险评估 |
| Scrum Master | agents/boss-scrum-master.md |
任务分解、测试用例定义 |
| Frontend | agents/boss-frontend.md |
UI 组件、状态管理、前端测试 |
| Backend | agents/boss-backend.md |
API、数据库、后端测试 |
| QA | agents/boss-qa.md |
测试执行、Bug 报告 |
| DevOps | agents/boss-devops.md |
构建部署、健康检查 |
调用 Agent 的标准格式
# 1. 读取 Agent Prompt
pm_prompt = Read("agents/boss-pm.md")
# 2. 调用 Task
Task(
subagent_type: "general_purpose_task",
description: "PM: 创建 PRD",
query: pm_prompt + "\
\
---\
\
## 当前任务\
\
[任务描述]"
)
产物目录结构
.boss/\x3Cfeature-name>/
├── prd.md # 阶段 1
├── architecture.md # 阶段 1
├── ui-spec.md # 阶段 1(可选)
├── tech-review.md # 阶段 2
├── tasks.md # 阶段 2
├── qa-report.md # 阶段 3
└── deploy-report.md # 阶段 4
快速开始
当用户触发 Boss Skill 后(除非 --quick),先询问:
🚀 Boss Mode 已激活!
请描述你想要构建的功能或项目:
- 新项目还是现有代码库?
- 需要什么类型的界面?(Web/CLI/API/无界面)
- 有技术偏好或约束吗?
获取信息后,立即开始四阶段流水线。
安全使用建议
This skill is coherent for orchestrating an automated development pipeline, but it will perform powerful actions in your environment: reading the repository, writing files under .boss/<feature>/, running shell commands (tests, build, docker/docker‑compose, dev servers), and performing web fetch/search. Before installing or running:
- Review the included init-project.sh and any other scripts to ensure they don't run unexpected commands.
- Avoid running in a privileged or production environment; use an isolated workspace or container when first trying the skill.
- Be cautious with the --quick flag (it skips confirmation prompts). Don't use --quick if you want to review generated changes.
- The skill does not request credentials, but DevOps steps may prompt you for deployment credentials (CI/CD, cloud providers) if you try to deploy — only provide secrets to trusted code and flows.
- If you need stricter controls, ensure the host platform restricts network or shell access for the skill or run it in a sandboxed environment.
功能分析
Type: OpenClaw Skill
Name: boss-skill
Version: 0.1.0
The skill bundle is classified as suspicious due to the extensive use of powerful tools, particularly `Bash` for shell execution, across multiple sub-agents (`boss-architect`, `boss-backend`, `boss-devops`, `boss-frontend`, `boss-qa`). While these capabilities are intended for legitimate software development tasks like running tests, building, and deploying applications, they introduce a significant attack surface. A malicious user could potentially exploit prompt injection vulnerabilities against the AI agent to misuse these `Bash` capabilities, leading to Remote Code Execution (RCE) or unauthorized file system manipulation on the host system. For example, `scripts/init-project.sh` uses `rm -rf` and `mkdir -p`, and `agents/boss-qa.md` explicitly instructs the agent to execute various test commands via `Bash`. Although the skill itself does not contain malicious instructions, the inherent risk associated with such broad shell access elevates its classification to suspicious.
能力评估
Purpose & Capability
The Skill claims to orchestrate a multi‑role development pipeline and all included agent prompts, tools (Read/Write/Glob/Grep/Bash/Task/WebSearch/WebFetch/etc.) and artifact conventions (.boss/<feature>/) are coherent with that purpose. There are no unexpected cloud credentials, unrelated binaries, or config paths requested.
Instruction Scope
SKILL.md and the agent prompts consistently instruct the boss agent to: collect requirements, load agent prompt files, call sub‑agents via Task, read the repo, write artifacts to .boss/, run tests and (optionally) deploy locally. Those operations are within scope for an automated dev pipeline. Note: several agents explicitly include Bash and commands to execute tests/deploy — this means the skill will run shell commands against the workspace when invoked (expected for the described functionality).
Install Mechanism
This is an instruction‑only skill with no install spec. No downloads or archive extraction are performed by the skill itself, so installation risk is low.
Credentials
The skill declares no required environment variables, credentials, or config paths. The included agent prompts reference web search/fetch and local file operations which are proportionate to research and codebase exploration. There are no demands for unrelated secrets (AWS keys, tokens, etc.).
Persistence & Privilege
always:false and user-invocable:true. The skill does not request permanent/global presence or changes to other skills' configs. It may execute sub‑tasks autonomously when invoked (disable-model-invocation:false), which is the platform default and necessary for orchestration.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install boss-skill - 安装完成后,直接呼叫该 Skill 的名称或使用
/boss-skill触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.1.0
boss-skill 0.1.0
- 首发版本,提供 BMAD 全自动项目编排流程,覆盖需求到部署的完整研发流水线。
- 支持多角色专家 Agent(PM、架构师、UI 设计师、Tech Lead、Scrum Master、开发者、QA、DevOps)的协作与自动切换。
- 引入四大阶段(需求收集、规划、评审与任务拆解、开发与验证、部署交付)清单式工作流。
- 参数化支持(--skip-ui、--skip-deploy、--quick)灵活适配不同项目类型或自动化程度需求。
- 所有文档产物均为中文,结构化保存于 .boss 目录,便于追踪与交付。
元数据
常见问题
Boss 是什么?
BMAD 全自动项目编排 Skill。从需求到部署的完整研发流水线,编排多个专业 Agent(PM、架构师、UI 设计师、Tech Lead、Scrum Master、开发者、QA、DevOps)自动完成完整研发周期。当用户说 'boss mode'、'/boss'、'全自动开发'、'从需求到部署'、'帮我做一个... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 761 次。
如何安装 Boss?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install boss-skill」即可一键安装,无需额外配置。
Boss 是免费的吗?
是的,Boss 完全免费(开源免费),可自由下载、安装和使用。
Boss 支持哪些平台?
Boss 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Boss?
由 青雲(@echovic)开发并维护,当前版本 v0.1.0。
推荐 Skills