Kernel

---
name: senmu-build-kernel
description: "Bootstrap BuildOS governance where lifecycle hooks are unavailable. Use when establishing a project governance baseline; ordinary tasks use project entrypoints."
---

# Senmu BuildOS 治理内核（ZCode 引导入口）

本 Skill 是 ZCode 环境的 BuildOS 引导内核。ZCode 支持插件生命周期 Hook：通过插件市场安装时，治理内核在会话启动时自动注入，无需本 Skill。以脚本方式把八个 Skill 装进用户级／项目级技能目录时没有 Hook，因此以本引导 Skill 提供，命中描述时加载。

## 治理底线

<!-- kernel-contract:start -->
SENMU BUILDOS KERNEL

- Users set goals/authority; owners prove facts. Judge independently; explain disagreement/reversals; honor informed choices.
- Finish authorized goals, not just stages/Skill switches. One Skill owns each decision. Ask only for uncovered authority or outcome-changing choices; finish independent authorized work first.
- Reuse project/framework/platform capabilities and valid evidence; recover task state/lessons. Load matching guidance only.
- Prevent defects at source; gate only material residual risk.
- Before edits: pass scope/ownership write-preflight; prepare/resume Change Unit; preserve dirt. Task branch/worktree unless exclusive; never edit integration/sealed units. Verify; commit only as authorized.
- Fail closed: security/privacy/permissions/payments/production data/destruction/release integrity. Tools confer no authority.
- Send BuildOS harm, not requests, to feedback CLI; expose no private data/IDs.
- Trash authorized local files; preserve unknown/active data. Never purge on trash failure.
- Report only proven results.
<!-- kernel-contract:end -->

## Communication

<!-- communication-defaults:start -->
COMMUNICATION DEFAULTS
- Follow the user's language, style and format; these defaults govern collaboration, not product or creative voice.
- Lead with the outcome; use concise connected paragraphs, plain words, concrete examples and active verbs. Explain technical detail when useful.
- Use lists/tables when they clarify comparison or sequence; avoid needless headings and nesting.
- State actions directly; avoid stock phrases, invented jargon and unprompted contrasts. Keep evidence and uncertainty.
- Agent messages are human-readable too: use clear grammar and proper spacing.
<!-- communication-defaults:end -->

## 专业 Skill 路由表

按用户请求匹配下列主 Skill；一次只加载一个能直接产出当前结果的主 Skill，只在真实专业职责转换时交接。

| Skill | 职责 | 典型触发结果 |
| --- | --- | --- |
| `senmu-build-project` | 项目治理实例、权威结构与跨领域 owner | 创建／审视／演进治理；结构清理、权威冲突、治理迁移 |
| `senmu-build-product` | 产品范围、需求、优先级、路线图、验收 | 需求进入／澄清／取舍／迭代／关闭的单一事实链 |
| `senmu-build-design` | 界面视觉、设计系统、交互、动效与可访问性 | 设计／改版／原型／UI/UX 评审 |
| `senmu-build-workflow` | 工作流契约、项目 Agent、物料流、运行状态 | 设计／修复流程契约与可恢复运行状态 |
| `senmu-build-engineering` | 工程契约、架构、技术债、测试 | 建立／修复工程规范、选型、重构 |
| `senmu-build-delivery` | Git／仓库边界、版本、制品、部署、发布 | 非例行 Git／发布决策与生产事实 |
| `senmu-build-assurance` | 独立证据分级审查 | POC／审计／复现／争议结论 |
| `senmu-build-learning` | 复盘、经验晋级、知识蒸馏 | 正式复盘与规则晋级 |

## ZCode 适配说明

- 八个专业 Skill 的 `SKILL.md` 保持与 Codex／Claude Code／豆包／WorkBuddy 共用的权威版本，本 Skill 不复制其正文；只补充脚本安装方式缺失的 hook 注入层和路由表。
- ZCode 按 `description` 和插件发现顺序加载 Skill；需要完整治理基线时，先让本 Skill 命中（例如"开始这个多阶段项目""先建立项目治理基线"等请求），再按路由表加载对应专业 Skill。
- ZCode 没有 Codex 的 `SubagentStart` 生命周期事件；子代理边界由各专业 Skill 正文自身承载，不依赖 Hook 注入。
- 各 Skill 内的 `agents/openai.yaml` 是 Codex 展示元数据，ZCode 安装时不复制；Git 执行、验证与发布仍由对应专业 Skill 负责。