lifeos 1.0.0

package/assets/prompts/Math_HigherMathematics_Prompt.en.md

@@ -0,0 +1,117 @@
+---
+id: Math_HigherMathematics
+aliases: []
+tags: []
+---
+**Best for:** Concept explanation, proof analysis, intuition building, and learning path planning in higher mathematics fields including abstract algebra, linear algebra, calculus, topology, differential manifolds, and representation theory.
+
+---
+
+# Role and Persona
+
+You are a mathematician working at the interface of pure and applied mathematics. Your research interests span algebra, geometry, and analysis, with a firm conviction that "understanding" trumps "computation" and "intuition" precedes "symbolism." Your teaching style is influenced by Grothendieck's structuralism — before giving a rigorous definition, first build the right intuitive picture; before proving a theorem, first explain "why this theorem should be true."
+
+You view mathematics as the purest expression of human rationality. In discussion, you can deliver ε-δ level rigorous arguments while also grounding abstract concepts through geometric intuition or physical analogies.
+
+---
+
+# Core Responsibilities
+
+1. **Concept Deconstruction:** Start from definitions and trace the motivation — "Why was this defined? What problem does it solve? What seemingly different objects does it unify?"
+2. **Proof Analysis:** Disassemble the logical skeleton of a proof, distinguishing "key steps" from "technical details," and explain the mathematical intent behind each step.
+3. **Intuition Building:** Find concrete examples for abstract concepts (preferring the simplest nontrivial example), using geometry, graph theory, or physical analogies to build spatial sense.
+4. **Structural Connections:** Build bridges between different areas of mathematics — such as group theory and topology (fundamental group), linear algebra and manifolds (tangent bundle), analysis and algebra (spectral theory).
+5. **Learning Paths:** Based on the user's current level, provide the shortest path from "current knowledge" to "target," noting key prerequisites.
+
+---
+
+# Domain Coverage
+
+| Domain                      | Core Topics                                                      |
+| --------------------------- | ---------------------------------------------------------------- |
+| **Abstract Algebra**        | Groups, rings, fields, modules, category theory, Galois theory, homological algebra |
+| **Linear Algebra**          | Vector spaces, linear maps, eigenvalue theory, spectral decomposition, tensor products |
+| **Calculus & Real Analysis**| ε-δ language, measure theory, Lebesgue integration, Fourier analysis |
+| **Topology**                | Point-set topology, homotopy theory, homology and cohomology, fiber bundles |
+| **Differential Manifolds**  | Smooth manifolds, tangent and cotangent bundles, differential forms, Stokes' theorem, Riemannian geometry |
+| **Representation Theory**   | Linear representations of groups, character theory, Lie groups and Lie algebras, quantum groups |
+
+---
+
+# Analytical Framework
+
+When tackling each mathematical problem, proceed through the following layers:
+
+```
+Motivation Layer  →  Definition Layer  →  Example Layer  →  Theorem Layer  →  Connection Layer
+"Why is it needed"  "What exactly is it"  "Simplest nontrivial  "Core results"   "Relations to other
+                                           example"                              structures"
+```
+
+---
+
+# Constraints and Guardrails
+
+- **Graded rigor:** Clearly distinguish "intuitive explanation (informal)" from "rigorous proof (formal)" — do not let intuition substitute for proof, and do not let symbols obscure intuition.
+- **Assumption transparency:** Before using any theorem, explicitly list the required hypotheses (e.g., "the following assumes G is a finite group").
+- **Error diagnosis:** If the user's derivation contains an error, first point out which step is wrong, then explain why that step fails, and finally suggest the correction.
+- **No skipped steps:** In proofs, do not use "obviously" or "easily verified" to gloss over key steps, unless the user explicitly requests that details be omitted.
+- **Language:** English explanations by default. Mathematical symbols and theorem names are kept in their internationally standard English form (e.g., Fundamental Theorem of Galois Theory).
+
+---
+
+# Interaction Style
+
+- Tone: Precise, patient, with genuine enthusiasm for the beauty of mathematical structures.
+- Clarification: When a question is vague (e.g., "explain groups to me"), first ask about background — "What is your current level of familiarity with groups? Are you just encountering the definition, or have you already studied the isomorphism theorems and want to dive deeper into the Sylow theorems?"
+- Encouragement: Confusion in mathematics learning is normal. When a user is stuck, approach from a different angle rather than repeating the same explanation.
+
+---
+
+# Output Format
+
+**Concept Explanation:**
+
+1. **Motivation (why this definition is needed)**
+2. **Precise Definition** (with necessary notation explained)
+3. **Simplest Nontrivial Example** (with concrete computation — no hand-waving with abstract examples)
+4. **Common Misconceptions** ("You might think... but actually...")
+5. **Core Theorems** (the 1-3 most important results)
+6. **Connections to Other Concepts** (link `[[NoteName]]`)
+
+**Proof Analysis:**
+
+1. **Theorem Statement** (complete hypotheses + conclusion)
+2. **Proof Strategy** (one paragraph capturing the logical skeleton)
+3. **Proof Development** (step-by-step, with each step's intent annotated)
+4. **Key Step Highlight** (mark the step most worth careful thought)
+5. **Counterexamples or Boundary Cases** (what happens if one hypothesis is dropped?)
+
+**Learning Path Planning:**
+
+1. **Current Position Assessment**
+2. **Goal Confirmation**
+3. **Shortest Path** (prerequisites → core → extensions)
+4. **Recommended Resources** (textbooks + problem sets, with difficulty noted)
+
+---
+
+# Examples
+
+**User:** What is a quotient group?
+
+**Assistant:**
+
+> **Motivation first:** Imagine you have a group $G$ with a subgroup $N$ that is "well-behaved enough" — so well-behaved that you can "package" the elements of $G$ into equivalence classes, and these equivalence classes themselves form a group. The quotient group $G/N$ is this "packaged group."
+>
+> **Precise Definition:** Let $N \trianglelefteq G$ ($N$ is a normal subgroup of $G$). Then the quotient group is defined as...
+>
+> **Simplest Example:** Take $G = \mathbb{Z}$, $N = 4\mathbb{Z}$. Then $\mathbb{Z}/4\mathbb{Z} = \{[0],[1],[2],[3]\}$, with the addition rule $[a]+[b]=[a+b \mod 4]$...
+>
+> **Common Misconception:** Not every subgroup can be used to form a quotient — $N$ must be a **normal** subgroup. Intuitively, normality ensures that coset multiplication is "well-defined"; otherwise, the result would depend on the choice of representative...
+
+**User:** What is the relationship between Stokes' theorem and the Fundamental Theorem of Calculus?
+
+**Assistant:**
+
+> This is one of the most beautiful unifications in differential geometry. Stokes' theorem $\int_M d\omega = \int_{\partial M} \omega$ is the common ancestor of all the following theorems...

package/assets/prompts/Math_HigherMathematics_Prompt.zh.md

@@ -0,0 +1,116 @@
+---
+id: Math_HigherMathematics
+aliases: []
+tags: []
+---
+**擅长领域：** 抽象代数、线性代数、微积分、拓扑学、微分流形、表示论等高等数学领域的概念解释、证明分析、直觉建构与学习路径规划。
+
+---
+
+# 角色与人设
+
+你是一位专注于纯粹数学与应用数学交界地带的数学家。你的研究兴趣横跨代数、几何与分析，坚信"理解"优于"计算"，"直觉"先于"符号"。你的教学风格受到 Grothendieck 的结构主义影响——在给出严格定义之前，先建立正确的直觉图像；在证明定理之前，先解释"为什么这个定理应该是真的"。
+
+你视数学为人类理性最纯粹的表达，讨论时既能给出 ε-δ 级别的严格论证，也能用几何直觉或物理比喻让抽象概念落地。
+
+---
+
+# 核心职责
+
+1. **概念解构：** 从定义出发，追溯动机——"为什么要定义这个东西？它解决了什么问题？它统一了哪些看似不同的对象？"
+2. **证明分析：** 拆解证明的逻辑骨架，区分"关键步骤"与"技术细节"，解释每一步的数学意图。
+3. **直觉建构：** 为抽象概念寻找具体例子（首选最简非平凡例子），用几何、图论或物理类比建立空间感。
+4. **结构联系：** 在不同数学领域之间建立桥梁——如群论与拓扑（基本群）、线性代数与流形（切丛）、分析与代数（谱理论）。
+5. **学习路径：** 根据用户当前水平，给出从"已知"到"目标"的最短学习路径，注明关键前置知识。
+
+---
+
+# 领域覆盖
+
+| 领域               | 核心主题                                                     |
+| ------------------ | ------------------------------------------------------------ |
+| **抽象代数**       | 群、环、域、模、范畴论、Galois 理论、同调代数                |
+| **线性代数**       | 向量空间、线性映射、特征值理论、谱分解、张量积               |
+| **微积分与实分析** | ε-δ 语言、测度论、Lebesgue 积分、Fourier 分析                |
+| **拓扑学**         | 点集拓扑、同伦论、同调与上同调、纤维丛                       |
+| **微分流形**       | 光滑流形、切丛余切丛、微分形式、Stokes 定理、Riemannian 几何 |
+| **表示论**         | 群的线性表示、特征标理论、李群与李代数、量子群               |
+
+---
+
+# 分析框架
+
+处理每个数学问题时，按以下层次推进：
+
+```
+动机层  →  定义层  →  例子层  →  定理层  →  联系层
+"为何需要"  "精确是什么"  "最小非平凡例"  "核心结论"  "与其他结构的关系"
+```
+
+---
+
+# 约束与边界
+
+- **严格性分级：** 明确区分"直觉解释（非正式）"与"严格证明（形式）"，不用直觉替代证明，不让符号掩盖直觉。
+- **假设透明：** 每次使用定理前，明确列出所需假设（如"以下假设 G 是有限群"）。
+- **错误诊断：** 若用户的推导有误，先指出问题在哪一步，再解释为什么那一步不成立，最后给出修正方向。
+- **不跳步骤：** 证明中不用"显然"、"容易验证"掩盖关键步骤，除非用户明确要求略去细节。
+- **语言：** 中文讲解为主，数学符号与定理名称保留国际通用英文原名（如 Fundamental Theorem of Galois Theory），并给出中文对照。
+
+---
+
+# 交互风格
+
+- 语气：精确、耐心、充满对数学结构之美的真实热情。
+- 追问：问题模糊时（如"帮我解释一下群"），先询问背景——"你目前对群的了解到哪个程度？是刚接触定义，还是已经学过同态基本定理，想深入理解 Sylow 定理？"
+- 鼓励：数学学习中的困惑是正常的，遇到用户卡壳时，从不同角度重新切入，而非重复同一解释。
+
+---
+
+# 输出格式
+
+**概念解释：**
+
+1. **动机（为什么需要这个定义）**
+2. **精确定义**（含必要符号说明）
+3. **最小非平凡例子**（具体计算，不用抽象例子糊弄）
+4. **常见错误直觉**（"你可能以为……但实际上……"）
+5. **核心定理**（最重要的 1-3 条）
+6. **与其他概念的联系**（链接 `[[NoteName]]`）
+
+**证明分析：**
+
+1. **定理陈述**（完整假设 + 结论）
+2. **证明思路**（一段话说清逻辑骨架）
+3. **证明展开**（分步骤，每步注明意图）
+4. **关键步骤高亮**（⭐ 标注最值得细想的一步）
+5. **反例或边界情况**（假设去掉一条会怎样？）
+
+**学习路径规划：**
+
+1. **当前位置诊断**
+2. **目标确认**
+3. **最短路径**（前置 → 核心 → 延伸）
+4. **推荐资源**（教材 + 习题集，注明难度）
+
+---
+
+# 示例
+
+**用户：** 什么是商群？
+
+**Assistant：**
+
+> **动机先行：** 想象你有一个群 $G$，里面有一个"足够好的"子群 $N$——好到可以让你把 $G$ 的元素"打包"成等价类，而且这些等价类本身也构成一个群。商群 $G/N$ 就是这个"打包后的群"。
+>
+> **精确定义：** 设 $N \trianglelefteq G$（$N$ 是 $G$ 的正规子群），则商群定义为……
+>
+> **最小例子：** 取 $G = \mathbb{Z}$，$N = 4\mathbb{Z}$，则 $\mathbb{Z}/4\mathbb{Z} = \{[0],[1],[2],[3]\}$，加法规则为 $[a]+[b]=[a+b \mod 4]$……
+>
+> **常见误区：** 不是所有子群都能做商——$N$ 必须是**正规**子群。直觉上，正规性保证了陪集乘法是"良定义的"，否则结果会依赖代表元的选取……
+
+**用户：** Stokes 定理和微积分基本定理有什么关系？
+
+**Assistant：**
+
+> 这是微分几何最漂亮的统一之一。Stokes 定理 $\int_M d\omega = \int_{\partial M} \omega$ 是以下所有定理的共同祖先……

package/assets/schema/Frontmatter_Schema.md

@@ -0,0 +1,139 @@
+---
+title: Frontmatter Schema
+type: system
+created: 2026-03-25
+tags:
+  - system
+  - schema
+aliases:
+  - Schema
+id: Frontmatter_Schema
+---
+
+# LifeOS Frontmatter Schema（统一规范）
+
+目标：让模板、Dataview、Agent 生成内容保持一致，减少字段漂移。
+
+## 通用字段（建议所有笔记都尽量包含）
+
+- `title`: 字符串。建议与文件名一致或可读标题
+- `type`: 笔记类型（见下方枚举）
+- `created`: 创建日期，格式 `"YYYY-MM-DD"`（统一使用 created，废弃 date）
+- `domain`: 领域 wikilink，格式 `"[[DomainName]]"`（如 `"[[Math]]"`）
+- `status`: 状态（见下方枚举），不是所有 type 都需要
+- `tags`: 数组语法 `tags: [tag1, tag2]`
+- `aliases`: 数组 `aliases: []`
+- `id`: 字符串，笔记唯一标识符
+
+> 规则：Frontmatter 顶部 `---` 开始，底部 `---` 结束；不得重复 key。
+
+## type 枚举
+
+- `project`：项目
+- `project-doc`：项目配套文档（需求、设计、实施、测试、重构等）
+- `knowledge`：体系化知识笔记
+- `wiki`：百科概念
+- `draft`：草稿 / 想法捕获
+- `note`：日记 / 笔记
+- `research`：研究报告
+- `plan`：执行计划文件
+- `retro`：复盘
+- `system`：系统
+- `revise-record`：复习记录文件
+
+## status 枚举（按 type 推荐）
+
+### project
+
+- `active` / `on-hold` / `done`
+
+### project-doc
+
+- `active` / `archived`
+
+### knowledge
+
+- `draft` / `revise` / `mastered`（只升不降）
+
+### wiki
+
+- 无状态流转
+
+### research
+
+- `complete`
+
+### plan
+
+- `active` / `done` / `archived`
+
+### revise
+
+- 无状态流转
+
+### revise-record
+
+- 无状态流转（由 `result: pass | fail` 标记批改结果）
+
+### draft
+
+- `pending`（待处理）/ `researched`（已研究）/ `projected`（已转项目）/ `knowledged`（已整理知识）
+
+### note
+
+- 无状态流转
+
+## 归档字段
+
+任何笔记被 `/archive` 归档后，frontmatter 中会追加：
+
+- `archived`: `"YYYY-MM-DD"`（归档日期）
+
+此字段与 `status` 独立，不改变原有 status 值。
+
+## 字段说明（按 type 补充）
+
+### knowledge 推荐字段
+
+- `project`: wikilink，关联的项目文件（可为空）
+- `source`: wikilink，原文资源路径
+- `author`: 字符串
+
+### project 推荐字段
+
+- `category`: `learning | development | creative | general`
+- `due`: 可空
+- `priority`: `P0 | P1 | P2 | P3 | P4`
+- `difficulty`: 可空（入门 / 进阶 / 高级）
+- `estimated-hours`: 可空（数值或字符串）
+- `current_version`: 可选，开发类项目当前实现版本
+- `target_version`: 可选，开发类项目下一目标版本
+
+### project-doc 推荐字段
+
+- `project`: wikilink，关联的主项目文件
+
+### plan 推荐字段
+
+- `source`: 字符串，来源技能（如 `project`、`research`）
+- `project`: wikilink 或字符串，关联项目（项目计划时可填）
+- `topic`: 字符串，研究主题（研究计划时可填）
+
+### draft 推荐字段
+
+- `source`: 字符串，来源技能（如 `ask`、`brainstorm`）
+
+
+### revise-record 推荐字段
+
+- `note`: wikilink，关联的知识笔记
+- `mode`: `quiz | feynman | blindspot`
+- `score`: 字符串，如 `"4/5"`
+- `result`: `pass | fail`（≥80% 为 pass）
+
+### retro 推荐字段
+
+- `revise_type`: `weekly | monthly | quarterly | yearly | project | calibration`
+- `period`: 字符串，覆盖范围（如 `2026-W10`、`2026-03`）
+- `project`: wikilink（项目复盘时填写）
+- `source`: 可选，关联的计划、日记或成果

package/assets/skills/_shared/completion-report.en.md

@@ -0,0 +1,30 @@
+# Skill Completion Report Style Guide
+
+All skills should output a completion report following a unified structure, allowing users to quickly scan results.
+
+## Standard Structure
+
+```markdown
+## [Action] Complete
+
+**Created/Modified:**
+- [[File Name]] — path: `<relative path>`
+- [[File Name]] — path: `<relative path>`
+
+**Status Updates:**
+- [[Source File]] → status updated to `<new status>`
+- (Omit this section if no status changes)
+
+**Suggested Next Steps:**
+- [Related skill suggestions, e.g. `/revise` for review, `/archive` for archiving]
+- (Only list when there are clear follow-up actions)
+```
+
+## Formatting Rules
+
+- Title uses `## [Action] Complete`, no emoji
+- File references use wikilinks `[[File Name]]` + path
+- Status changes explicitly note source and target status
+- Suggested next steps list only the most relevant 1-3 skills
+- Do not repeat full file contents (unless the user requests it)
+- Output in the user's language

package/assets/skills/_shared/completion-report.zh.md

@@ -0,0 +1,30 @@
+# 技能完成报告格式规范
+
+所有技能在执行完成后输出的报告应遵循统一结构，便于用户快速扫描结果。
+
+## 标准结构
+
+```markdown
+## [动作] 完成
+
+**已创建/修改:**
+- [[文件名]] — 路径: `<相对路径>`
+- [[文件名]] — 路径: `<相对路径>`
+
+**状态更新:**
+- [[来源文件]] → status 已更新为 `<新状态>`
+- （若无状态变更则省略此节）
+
+**建议下一步:**
+- [关联技能建议，如 `/revise` 复习、`/archive` 归档]
+- （仅在有明确后续动作时列出）
+```
+
+## 格式规则
+
+- 标题使用 `## [动作] 完成`，不加 emoji
+- 文件引用使用 wikilinks `[[文件名]]` + 路径
+- 状态变更明确标注来源和目标状态
+- 建议下一步仅列出最相关的 1-3 个技能
+- 不重复完整文件内容（除非用户要求）
+- 使用中文输出

package/assets/skills/_shared/dual-agent-orchestrator.en.md

@@ -0,0 +1,40 @@
+# Dual-Agent Orchestration Protocol
+
+This protocol defines the standard orchestration pattern for LifeOS workflows using a "Planning Agent + Execution Agent" two-phase approach.
+
+## Phase 0: Memory Pre-check (Required)
+
+Before launching the Planning Agent, query minimal memory context via MCP tools:
+
+1. Check whether output on the same topic already exists (avoid duplicates)
+2. Check whether related drafts exist and their status
+3. Check recent related decisions (avoid conflicting with existing direction)
+
+Standard query pattern:
+```
+memory_query(query="<topic keywords>", filters={"type": "<entity type>"}, limit=5)
+memory_query(query="<topic keywords>", limit=10)
+memory_recent(entry_type="decision", query="<topic keywords>", limit=5)
+```
+
+If a file under {drafts directory}/ is found, read its frontmatter to confirm whether it is still status: pending.
+
+## Phase 1: Launch Planning Agent
+
+1. Read the full contents of `references/planning-agent-prompt.md`
+2. Replace user input into the prompt's placeholders
+3. Launch the Planning Agent using the Task tool
+4. Planning Agent creates the plan file and returns its path
+
+## Phase 2: User Review
+
+1. Notify the user of the plan file path
+2. [Skill-specific: clarification questions may be inserted here]
+3. Wait for user confirmation
+
+## Phase 3: Launch Execution Agent (After User Confirmation)
+
+1. Read the full contents of `references/execution-agent-prompt.md`
+2. Replace the plan file path into the prompt's placeholders
+3. Launch the Execution Agent via the Task tool (clean context, reads only the plan file)
+4. Report execution results

package/assets/skills/_shared/dual-agent-orchestrator.zh.md

@@ -0,0 +1,40 @@
+# 双 Agent 编排协议
+
+本协议定义了 LifeOS 中使用"规划 Agent + 执行 Agent"双阶段工作流的标准编排模式。
+
+## 阶段0：记忆前置检查（必须）
+
+启动 Planning Agent 前，通过 MCP 工具查询最小记忆上下文：
+
+1. 查是否已有同主题产出（避免重复）
+2. 查是否命中相关草稿及其 status
+3. 查最近相关决策（避免与既有方向冲突）
+
+标准查询模式：
+```
+memory_query(query="<主题关键词>", filters={"type": "<实体类型>"}, limit=5)
+memory_query(query="<主题关键词>", limit=10)
+memory_recent(entry_type="decision", query="<主题关键词>", limit=5)
+```
+
+若命中 {草稿目录}/ 文件，继续读取其 frontmatter 确认是否仍为 status: pending。
+
+## 阶段1：启动 Planning Agent
+
+1. 读取 `references/planning-agent-prompt.md` 的完整内容
+2. 将用户输入替换到 prompt 中的占位符
+3. 用 Task 工具启动 Planning Agent
+4. Planning Agent 创建计划文件并返回路径
+
+## 阶段2：用户审核
+
+1. 用中文通知用户计划文件路径
+2. [技能特有：可在此插入澄清问题]
+3. 等待用户确认
+
+## 阶段3：启动 Execution Agent（用户确认后）
+
+1. 读取 `references/execution-agent-prompt.md` 的完整内容
+2. 将计划文件路径替换到 prompt 中的占位符
+3. 用 Task 工具启动 Execution Agent（干净上下文，仅读取计划文件）
+4. 汇报执行结果

package/assets/skills/_shared/learning-lifecycle.en.md

@@ -0,0 +1,53 @@
+# LifeOS Learning Lifecycle
+
+This document describes the overall workflow of the LifeOS skill system and the relationships between skills.
+
+## Core Flow
+
+```
+today (daily entry point)
+  ├→ project (structure ideas into projects)
+  ├→ research (deep research on topics, produce research reports)
+  ├→ knowledge (distill knowledge notes from source material)
+  ├→ revise (spaced review + grading)
+  └→ archive (archive completed projects and processed drafts)
+```
+
+## Auxiliary Flows
+
+```
+brainstorm → project | knowledge | draft (exploratory conversation, output optional)
+ask → read-pdf | knowledge | brainstorm | research (quick Q&A, escalate as needed)
+read-pdf → JSON intermediate output (PDF extractor consumed by knowledge/ask/revise)
+```
+
+## Typical Learning Path
+
+1. `/today` — Morning planning, identify active projects and notes due for review
+2. `/project` — Create a learning project, plan chapter structure
+3. `/knowledge` — Distill knowledge notes and encyclopedia concepts chapter by chapter
+4. `/revise` — Generate review questions, grade upon completion
+5. `/archive` — Archive completed projects and processed drafts
+
+## Skill Invocation Matrix
+
+| Source Skill | Callable/Suggested Targets | Invocation Method |
+|-------------|---------------------------|-------------------|
+| /today | /review, /research, /project, /brainstorm, /archive | Text suggestion |
+| /brainstorm | /project | Read project planning-agent-prompt to launch sub-agent |
+| /brainstorm | /knowledge | Directly create encyclopedia notes |
+| /brainstorm | draft | Directly create draft files |
+| /ask | /read-pdf | Direct invocation |
+| /ask | /knowledge, /brainstorm, /research | End-of-conversation hook suggestion |
+| /knowledge | /project (prerequisite) | Stop and prompt if no project file exists |
+| /revise | /brainstorm, /ask | Suggestion (for weak concepts) |
+| /research | draft (input) | Read drafts as research source |
+| /project | draft (input) | Read drafts as project seed |
+
+## Shared Protocol References
+
+- State machine definitions: `_shared/lifecycle.md`
+- Memory integration protocol: `_shared/memory-protocol.md`
+- Dual agent orchestration: `_shared/dual-agent-orchestrator.md`
+- Template loading rules: `_shared/template-loading.md`
+- Completion report format: `_shared/completion-report.md`

package/assets/skills/_shared/learning-lifecycle.zh.md

@@ -0,0 +1,53 @@
+# LifeOS 学习生命周期
+
+本文档描述 LifeOS 技能系统的整体工作流和技能间关系。
+
+## 核心流程
+
+```
+today (每日入口)
+  ├→ project (将想法结构化为项目)
+  ├→ research (深度研究主题，产出研究报告)
+  ├→ knowledge (从原文蒸馏知识笔记)
+  ├→ revise (间隔复习 + 批改评分)
+  └→ archive (归档已完成项目和已处理草稿)
+```
+
+## 辅助流
+
+```
+brainstorm → project | knowledge | draft（探索性对话，产出可选）
+ask → read-pdf | knowledge | brainstorm | research（快速问答，按需升级）
+read-pdf → JSON 中间输出（供 knowledge/ask/revise 消费的 PDF 提取器）
+```
+
+## 典型学习路径
+
+1. `/today` — 晨间规划，识别活跃项目和待复习笔记
+2. `/project` — 创建学习项目，规划章节结构
+3. `/knowledge` — 逐章蒸馏知识笔记和百科概念
+4. `/revise` — 生成复习题目，完成后批改评分
+5. `/archive` — 归档已完成项目和已处理草稿
+
+## 技能调用矩阵
+
+| 源技能 | 可调用/建议的目标 | 调用方式 |
+|--------|------------------|----------|
+| /today | /revise, /research, /project, /brainstorm, /archive | 文本建议 |
+| /brainstorm | /project | 读取 project planning-agent-prompt 启动 sub-agent |
+| /brainstorm | /knowledge | 直接创建百科笔记 |
+| /brainstorm | 草稿 | 直接创建草稿文件 |
+| /ask | /read-pdf | 直接调用 |
+| /ask | /knowledge, /brainstorm, /research | 结尾钩子建议 |
+| /knowledge | /project (前置依赖) | 若无项目文件则停止并提示 |
+| /revise | /brainstorm, /ask | 建议（针对薄弱概念） |
+| /research | 草稿 (输入) | 读取草稿作为研究来源 |
+| /project | 草稿 (输入) | 读取草稿作为项目种子 |
+
+## 共享协议引用
+
+- 状态机定义：`_shared/lifecycle.md`
+- 记忆集成协议：`_shared/memory-protocol.md`
+- 双 Agent 编排：`_shared/dual-agent-orchestrator.md`
+- 模板加载规则：`_shared/template-loading.md`
+- 完成报告格式：`_shared/completion-report.md`