超压缩 communication mode。Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman.
面向棘手 bug、性能回退和 BMad story execution failures 的纪律化 diagnosis loop。Reproduce → minimise → hypothesise → instrument → fix → regression-test. Use when user says "diagnose this" / "debug this", reports a bug, says something is broken/throwing/failing, describes a performance regression, or needs to diagnose a failure found during BMad dev-story or review.
设置 Claude Code hooks,在危险 git commands(push, reset --hard, clean, branch -D 等)执行前阻止它们。Use when user wants to prevent destructive git operations, add git safety hooks, or block git push/reset in Claude Code.
围绕 plan 或 design 持续 interview user,直到达成 shared understanding,并逐一解决 decision tree 的每个分支。Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Grilling session that challenges a plan against the existing domain model, feature docs, and documented decisions, then updates CONTEXT.md and ADRs as decisions crystallise. Use when user wants to stress-test a plan against project language, PRD/Spec assumptions, or durable technical decisions.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
AI Native Doc System 是一套面向 AI 辅助开发的 Markdown 文档系统和 skill 工作流。它基于 Matt Pocock skills 的中文本地化版本,结合 BMad Method 的复杂项目工作流,以及 AI-native 文档实践整理而成。
这个仓库不是从零开始写的一组 skills。更准确地说,它是一个整合与再设计项目:把已有 engineering skills 接入同一条文档链,并补充 PRD、Plan、Spec、Issue、BMad 融合和长期真相源规则。
AI 编码真正难的不是“让 AI 写代码”,而是让 AI 在每次新会话里读到正确上下文,并知道什么内容是正式约束。
这个仓库解决四个问题:
PRD -> PLAN -> SPEC -> issues 文档链。docs/ 作为长期真相源。核心原则:
docs/ 是长期真相源。
skills 负责生成、读取和维护文档产物。
BMad 负责展开、审计和执行;AI-native 文档负责裁决、沉淀和控制。
本仓库建立在以下来源之上:
mattpocock/skills:原始上游项目,提供多种 agent skills 的基础设计和实现。vinvcn/mattpocock-skills-zh-CN:Matt Pocock skills 的简体中文本地化版本,是本仓库中多数组件的直接上游来源。to-plan、to-spec、BMad-aware to-issues、setup 配置文档,以及面向长期真相源的 skill 改造。详细来源、派生目录和改动说明见 THIRD_PARTY_NOTICES.md。
setup-ai-document-system:在目标仓库中初始化 agent adapter 和 docs/agents/*.md 配置。grill-with-docs:围绕领域语言、边界和长期技术决策进行追问,并更新 CONTEXT.md / ADR。to-prd:把对话上下文整理为 docs/features/<feature>/PRD.md。to-plan:生成可选阶段计划,放在 docs/plans/ 或 docs/features/<feature>/PLAN.md。to-spec:把 PRD / Plan 转成带验证命令的 docs/features/<feature>/SPEC.md。to-issues:把 Spec / Plan 拆成可执行的 vertical-slice issues;在 BMad story cycle 中只生成控制面板 issues。diagnose:纪律化 bug / performance regression diagnosis loop。tdd:使用 red-green-refactor loop 做 test-driven development。triage:基于 issue tracker 和 triage labels 管理 incoming issues。zoom-out:在不熟悉代码区域时生成模块、调用方、文档和决策地图。improve-codebase-architecture:寻找 deepening opportunities,改善 testability 和 AI-navigability。prototype:构建 throwaway prototype 来验证状态模型、业务逻辑或 UI 方向。caveman:超压缩沟通模式。grill-me:围绕计划或设计持续追问,直到达成共同理解。handoff:把当前 conversation 压缩成可交接文档。write-a-skill:创建结构正确、支持 progressive disclosure 的新 agent skills。git-guardrails-claude-code:设置 Claude Code hooks,阻止危险 git commands。migrate-to-shoehorn:将 test files 从 as type assertions 迁移到 @total-typescript/shoehorn。scaffold-exercises:创建 exercises 目录结构。setup-pre-commit:设置 Husky、lint-staged、typecheck 和测试 hook。把需要的 skills 复制到目标仓库的 .agents/skills/ 目录,或使用支持 .claude-plugin/plugin.json 的 installer 安装本仓库 active skills。
最小推荐集合:
skills/setup-ai-document-system
skills/grill-with-docs
skills/to-prd
skills/to-plan
skills/to-spec
skills/to-issues
首次使用顺序:
setup-ai-document-system,建立目标仓库自己的 agent adapter 和 docs/agents/*.md 配置。grill-with-docs 收敛领域语言、边界和长期决策。to-prd -> to-plan(可选)-> to-spec -> to-issues。diagnose、tdd、triage、prototype、zoom-out 或 improve-codebase-architecture。BMAD_INTEGRATION.md 和 LEVEL3_BMAD_WORKFLOW.md。这是推荐结构,不是强制模板。项目已有 AGENTS.md、CLAUDE.md 或其他工具入口时,应把它们视为 agent adapter。
/
├── AGENTS.md / CLAUDE.md / other agent adapter
├── CONTEXT.md
├── README.md
├── docs/
│ ├── agents/
│ │ ├── document-system.md
│ │ ├── domain.md
│ │ ├── issue-tracker.md
│ │ └── triage-labels.md
│ ├── adr/
│ ├── plans/
│ └── features/
│ └── <feature>/
│ ├── PRD.md
│ ├── PLAN.md
│ ├── SPEC.md
│ └── issues/
└── .agents/
└── skills/
PLAN.md 是可选文档。只有当功能、版本或跨功能交付需要明确阶段和顺序时才创建。
修改 skill、manifest 或 README 后,运行:
node scripts/validate-skills.mjs
当前验证脚本会检查 .claude-plugin/plugin.json 中列出的 active skills 是否存在,并确认 README 中提及这些 skill。
后续可在 fixtures/minimal-target/ 增加一个最小目标仓库样例,用来验证这些 skills 假设的文档结构。这个 fixture 应作为 smoke test 输入,而不是新的工作流规范;真正的规范仍以 docs/AI_DOCUMENT_SYSTEM.md 为准。
本项目使用 MIT License。
本仓库包含来自 vinvcn/mattpocock-skills-zh-CN 的派生 skill 内容;该项目基于 Matt Pocock 的 MIT 授权项目 mattpocock/skills。BMad 相关内容来自对 BMad Method 工作流和本地 BMad skills 的融合整理。
完整来源、改动说明和署名信息见 THIRD_PARTY_NOTICES.md。
npx claudepluginhub pop1414/ai-native-doc-systemUltra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Frontend design skill for UI/UX implementation
Memory compression system for Claude Code - persist context across sessions
Marketing skills for AI agents — conversion optimization, copywriting, SEO, paid ads, ad creative, and growth
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.