@sushanglewis/lincoln 1.5.0 → 1.5.2

package/framework/.claude/skills/lc-research-report/prompts/main.md

@@ -0,0 +1,90 @@
+# lc-research-report prompt
+
+You are executing the Lincoln workflow step `lc-research-report`: synthesize all research artifacts into a final report.
+
+## Goal
+
+Produce `{process_slug}/research/{session_id}/report.md` that integrates the research scope, first principles, stakeholder map, market, product, competitive, framework, intelligence, and narrative into a decision-ready document.
+
+## Inputs
+
+- `session_id`: the research session identifier
+- All prior research artifacts in `{process_slug}/research/{session_id}/`
+
+## Steps
+
+1. Read all prior research artifacts.
+2. Write an executive summary with:
+   - Core thesis
+   - Key findings
+   - Recommended decision
+   - Key risks
+3. Include sections for each research domain (scope, first principles, stakeholders, market, product, competitive, frameworks, intelligence, narrative).
+4. Link every claim to evidence in the prior artifacts.
+5. Add an appendix with all sources.
+6. Write `report.md` using the template below.
+7. Present the report to the PM for final approval.
+
+## Output template
+
+```markdown
+# Research Report: <Title>
+
+## Executive Summary
+
+### Core Thesis
+...
+
+### Key Findings
+1. ...
+2. ...
+
+### Recommendation
+...
+
+### Risks
+- ...
+
+## Background
+...
+
+## First Principles
+...
+
+## Stakeholder Insights
+...
+
+## Market Analysis
+...
+
+## Product Analysis
+...
+
+## Competitive Analysis
+...
+
+## Applied Frameworks
+...
+
+## Evidence Summary
+...
+
+## Narrative
+...
+
+## Next Steps
+1. ...
+
+## Appendix: Sources
+- ...
+```
+
+## Human Interaction Rules
+
+- Present the report to the PM for approval.
+- Do not mark the stage complete until the PM explicitly confirms.
+
+## Exit Criteria
+
+- `report.md` is complete and approved by the human PM.
+- Every recommendation links to evidence.

package/framework/.claude/skills/lc-research-scope/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: lc-research-scope
+description: 界定 PM 研究工作的范围、目标、问题与验收标准，作为后续研究阶段的单一事实来源。
+version: 1.0.0
+triggers:
+  - "lc-research-scope"
+inputs:
+  - name: topic
+    description: 研究主题或产品决策背景
+    required: false
+outputs:
+  - "{process_slug}/research/{session_id}/scope.md"
+required_tools:
+  - Bash
+  - Read
+  - Write
+---
+
+# lc-research-scope
+
+## Purpose
+
+Using [lc-research-scope] to 界定 PM 研究工作的范围、目标、问题与验收标准，作为后续研究阶段的单一事实来源.
+
+## When to Use
+
+- `pm-research` workflow 的 `lc-research-scope` 阶段。
+- PM 需要 界定 PM 研究工作的范围、目标、问题与验收标准，作为后续研究阶段的单一事实来源。 时。
+
+## Inputs
+
+- `session_id`: 研究 session 标识
+- 当前阶段之前的所有研究产物
+
+## Outputs
+
+- `{process_slug}/research/{session_id}/scope.md`
+
+## Rules
+
+- 严格基于已有研究产物，避免引入未经验证的假设。
+- 所有外部事实必须标注来源。
+- human_gate 阶段必须获得人类 PM 显式确认。

package/framework/.claude/skills/lc-research-scope/prompts/main.md

@@ -0,0 +1,82 @@
+# lc-research-scope prompt
+
+You are executing the Lincoln workflow step `lc-research-scope`: define the research brief for a PM-led research workflow.
+
+## Goal
+
+Produce `{process_slug}/research/{session_id}/scope.md` that becomes the single source of truth for all subsequent research stages.
+
+## Inputs
+
+- `session_id`: the research session identifier
+- Optional topic, questions, or context provided by the human PM
+- Any existing requirements, interviews, designs, or knowledge documents in `{process_slug}/`
+
+## Steps
+
+1. Create `{process_slug}/research/{session_id}/` if it does not exist.
+2. Read any existing context that could shape the research (requirements, interview summaries, knowledge notes).
+3. Draft `scope.md` using the template below.
+4. Identify 1-3 ambiguities and tag each with its Johari quadrant:
+   - 知道自己知道 → confirm with the PM
+   - 知道自己不知道 → ask directly and coach
+   - 不知道自己知道 → surface existing assets
+   - 不知道自己不知道 → probe with concrete scenarios
+5. Ask the human PM at most 3 questions per turn to converge on scope.
+6. Update `scope.md` after each answer.
+7. Repeat until the PM confirms the scope is ready.
+8. When confirmed, append `<!-- status: approved -->` to `scope.md`.
+
+## Output template
+
+```markdown
+# Research Scope: <Title>
+
+## Background
+
+<决策背景与研究动机>
+
+## Research Questions
+
+1. <问题 1>
+2. <问题 2>
+3. <问题 3>
+
+## Scope Inclusions
+
+- <明确包含的内容>
+
+## Scope Exclusions
+
+- <明确不包含的内容，避免范围蔓延>
+
+## Stakeholders
+
+- <谁使用、谁买单、谁建设、谁维护（初步）>
+
+## Analysis Frameworks
+
+- <初步建议使用的分析框架>
+
+## Acceptance Criteria
+
+1. <怎样算研究完成>
+2. <需要产出哪些 artifact>
+3. <决策点是什么>
+
+## Risks & Assumptions
+
+- <关键假设>
+- <可能风险>
+```
+
+## Human Interaction Rules
+
+- Ask at most 3 questions per turn.
+- After each answer, update the document and show the changed sections.
+- Do not proceed to the next step until the PM explicitly confirms (e.g., says "confirm" or "确认").
+
+## Exit Criteria
+
+- `scope.md` is approved by the human PM.
+- Research questions and acceptance criteria are explicit and decision-oriented.

package/framework/.claude/skills/lc-setup/SKILL.md

@@ -0,0 +1,68 @@
+---
+name: lc-setup
+description: |
+  Lincoln 环境初始化 skill。用于在仓库首次打开或依赖发生变化时，
+  检查、安装并配置 Lincoln 所需的外部 skills、CLI 工具、插件以及仓库配置。
+---
+
+# lc-setup
+
+## Purpose
+
+Using [lc-setup] to guide one-time Lincoln installation and dependency setup.
+
+详细安装执行指令见 `prompts/main.md`。会话启动 hook 只负责提示依赖缺失，具体安装剧本由本 skill 承载。
+
+## 触发条件
+
+- 用户首次用 Claude Code 打开从 Lincoln template 创建的新仓库。
+- `.claude/hooks/on-session-start.sh` 检测到 Lincoln setup 尚未完成。
+- 用户复制 README 中的初始化提示词并发送给 Claude。
+- 运行 `python scripts/lincoln-setup.py check` 发现缺失依赖。
+
+## 职责
+
+1. 检查 `.claude/skills/dependencies.yaml` 声明的依赖是否满足。
+2. 安装缺失的外部 skills（`superpowers`、`gsd`）到 `~/.claude/skills/`，并确保 ref 正确。
+3. 安装缺失的 CLI 工具（`openspec`、`gh`、`ffmpeg`、Whisper 实现）。
+4. 安装 `oh-my-claudecode` 插件（可选但默认安装）。
+5. 配置 `.github/openspec-config.yml` 的 `repository.owner` 和 `repository.name`。
+6. 运行 `scripts/init-project.sh` 完成项目级初始化。
+7. 安装任何全局工具或写入配置前，必须先向用户确认。
+
+## 调用方式
+
+优先使用 Skill 工具调用 `lc-setup`；如果环境不支持，可运行等效命令：
+
+```bash
+python scripts/lincoln-setup.py bootstrap
+```
+
+常用子命令：
+
+```bash
+python scripts/lincoln-setup.py check              # 仅检查
+python scripts/lincoln-setup.py install-skills     # 安装外部 skills
+python scripts/lincoln-setup.py install-clis       # 安装 CLI 工具
+python scripts/lincoln-setup.py init-repo-config --owner <owner> --name <repo>
+python scripts/lincoln-setup.py bootstrap          # 完整流程
+```
+
+## 产物
+
+- `~/.claude/skills/` 下的外部 skill 目录。
+- 安装到 PATH 的 CLI 工具。
+- 更新后的 `.github/openspec-config.yml`。
+- 写入 `.context/lc-setup-state.yaml` 的 setup 完成状态。
+- 初始化后的项目目录结构（products、oss、knowledge 等）。
+
+## 约束
+
+- 不得跳过用户确认擅自安装全局工具。
+- 遇到不支持的平台或安装失败时，给出清晰的手动替代命令。
+- 安装完成后汇报状态：依赖是否就绪、配置是否写入、下一步动作。
+- 如果 `.context/lc-setup-state.yaml` 已标记完成且依赖检查通过，可跳过重复安装。
+
+## Prompts
+
+- `prompts/main.md` — 给 Claude 的详细安装执行指令。

package/framework/.claude/skills/lc-setup/prompts/main.md

@@ -0,0 +1,88 @@
+# Lincoln 初始化安装执行指令
+
+> 如果用户更喜欢终端 TUI 安装，也可以推荐 `npx lincoln-install` 作为自助式替代；它会调用同样的 `scripts/lincoln-setup.py` 后端，但提供交互式选择 harness 与安装范围的向导。
+
+你是 Lincoln 的初始化助手。当前仓库可能尚未完成 Lincoln 环境配置，或者依赖清单已更新。请按以下步骤执行，并在每一步完成后向用户简要汇报。
+
+## 1. 检查当前环境
+
+运行：
+
+```bash
+python scripts/lincoln-setup.py check
+```
+
+- 列出所有缺失的 skills、CLI 工具、插件。
+- 区分 required 与 optional/default_install。
+- 如果检查通过且 `.context/lc-setup-state.yaml` 已标记完成，直接汇报“环境已就绪”，跳过剩余步骤。
+
+## 2. 安装外部 skills
+
+运行：
+
+```bash
+python scripts/lincoln-setup.py install-skills
+```
+
+- 默认会逐个询问用户确认。如果用户明确说“全部安装”，可追加 `--yes`。
+- 该命令会：
+  - 将 `superpowers`（上游 obra/superpowers，main 分支）clone 到 `~/.claude/skills/superpowers`
+  - 将 `gsd`（上游 gsd-build/get-shit-done，main 分支）clone 到 `~/.claude/skills/gsd`
+  - 安装 `oh-my-claudecode` 插件（如声明为 default_install）
+- 如果目标目录已存在且 ref 匹配，自动跳过；如果 ref 不匹配或工作区 dirty，先向用户说明再决定。
+
+## 3. 安装 CLI 工具
+
+运行：
+
+```bash
+python scripts/lincoln-setup.py install-clis
+```
+
+- 默认逐个询问用户确认。
+- **先问用户是否需要录音转写能力**；只有需要时才安装 ffmpeg 与 Whisper 实现。
+- 需要处理的 CLI 包括：
+  - `openspec`：基于 Node.js，运行 `npm install -g @fission-ai/openspec`
+  - `gh`：macOS 用 `brew install gh`，Linux 用包管理器
+  - `ffmpeg`（可选，仅录音转写需要）：macOS 用 `brew install ffmpeg`
+  - Whisper 实现（可选，仅录音转写需要）：优先 `faster-whisper`；如果失败，提供 `openai-whisper` 或 OpenAI API key 选项
+- 顺带询问用户是否需要 benchmark；需要时介绍 `scripts/lincoln_benchmark.py` 的用法。
+- 安装失败后，给出对应平台的手动安装命令。
+
+## 4. 配置仓库信息
+
+运行：
+
+```bash
+python scripts/lincoln-setup.py init-repo-config --owner <owner> --name <repo>
+```
+
+- 从 git remote origin 自动推断 owner/name；如果推断失败，询问用户。
+- 如果 `.github/openspec-config.yml` 已经包含真实值，命令会跳过并提示。
+
+## 5. 运行项目初始化
+
+运行：
+
+```bash
+bash scripts/init-project.sh
+```
+
+- 这会创建 products、oss、knowledge 等目录结构。
+- 如果还有未满足的依赖，`init-project.sh` 会警告但不会强制退出。
+
+## 6. 汇报状态
+
+完成所有步骤后，向用户汇报：
+
+- ✅ 已安装的 skills 与 CLI 工具
+- ✅ `.github/openspec-config.yml` 中的仓库信息
+- ⚠️ 仍缺失或需要手动处理的项目
+- 📋 下一步：创建 issue 工作包或开始第一个 Lincoln 工作流
+
+## 重要约束
+
+- 不要假设用户已经允许安装全局工具；每次安装前必须获得确认（除非用户主动说 `--yes`）。
+- 不要修改 `{process_slug}/recordings/` 中的任何文件。
+- 不要创建 GitHub Issues 或 PR，除非人类明确请求。
+- 如果某一步失败，暂停并报告失败原因与修复建议，不要继续后续步骤。

package/framework/.claude/skills/lc-stage/SKILL.md

@@ -0,0 +1,45 @@
+---
+name: lc-stage
+description: 用自然语言驱动当前阶段的生命周期：查看状态、校验准入、提交产物、确认 gate、进入下一阶段
+triggers:
+  - "lc-stage"
+  - "提交产物"
+  - "提交本阶段产物"
+  - "确认通过"
+  - "审批通过"
+  - "校验阶段"
+  - "进入下一阶段"
+inputs:
+  - name: intent
+    description: 自然语言意图（如"提交产物""确认通过"）；可选，缺省时汇报当前状态并询问下一步
+    required: false
+outputs:
+  - 对应 stage_loader.py 动作的执行结果与下一步建议
+required_tools:
+  - Bash
+  - Read
+---
+
+# lc-stage — 阶段生命周期自然语言入口
+
+## Purpose
+
+Using [lc-stage] to 用自然语言驱动当前阶段的生命周期：查看状态、校验准入、提交产物、确认 gate、进入下一阶段.
+
+
+用户不需要记任何终端命令。Agent 把自然语言意图翻译成 `scripts/stage_loader.py` 动作并代为执行：
+
+| 用户说 | 动作 | 底层命令（Agent 执行，用户无需关心） |
+|--------|------|-----------------------------------|
+| "现在什么状态" | status | `python3 "${CLAUDE_PLUGIN_ROOT:-${LINCOLN_HOME:-$HOME/.lincoln/current}}/scripts/lincoln-status.py" --format markdown`（或 `/lc-status`） |
+| "可以开始吗 / 校验准入" | validate-entry | `python3 "${CLAUDE_PLUGIN_ROOT:-${LINCOLN_HOME:-$HOME/.lincoln/current}}/scripts/stage_loader.py" --stage <current_stage> --action validate-entry`（或 `/lc-validate`） |
+| "提交产物 / 记录产物" | record-artifacts | `python3 "${CLAUDE_PLUGIN_ROOT:-${LINCOLN_HOME:-$HOME/.lincoln/current}}/scripts/stage_loader.py" --stage <current_stage> --action record-artifacts`（或 `/lc-submit`） |
+| "确认通过 / 审批通过" | approve-gate | 仅当人类 PM 明确确认后：`... --action approve-gate`（或 `/lc-approve`） |
+| "生成交接 / 切换窗口" | handoff-report | 见 `/lc-handoff` |
+| "进入下一阶段" | transition-next | 仅在当前阶段 gate 通过后执行 |
+
+规则：
+
+1. `<current_stage>` 由 Agent 从 `{process_slug}/workflow-stage.yaml`（team）或 `.context/workflow/<name>.yaml`（solo）的 `current_run.current_stage` 读取，不要让用户提供。
+2. `human_gate: true` 的阶段，执行 `approve-gate` 前必须有人类 PM 的显式确认原话；Agent 不得自行审批，不得用本技能替代人类确认。
+3. 执行后用一两句话向用户汇报结果与下一步，不要倾倒原始命令输出。

package/framework/.claude/skills/lc-stakeholder-research/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: lc-stakeholder-research
+description: 识别并研究产品相关者（用户、买单方、建设方、维护方），找到他们的发声渠道并分析其诉求。
+version: 1.0.0
+triggers:
+  - "lc-stakeholder-research"
+inputs:
+  - name: topic
+    description: 研究主题或产品决策背景
+    required: false
+outputs:
+  - "{process_slug}/research/{session_id}/stakeholder-research.md"
+required_tools:
+  - Bash
+  - Read
+  - Write
+---
+
+# lc-stakeholder-research
+
+## Purpose
+
+Using [lc-stakeholder-research] to 识别并研究产品相关者（用户、买单方、建设方、维护方），找到他们的发声渠道并分析其诉求.
+
+## When to Use
+
+- `pm-research` workflow 的 `lc-stakeholder-research` 阶段。
+- PM 需要 识别并研究产品相关者（用户、买单方、建设方、维护方），找到他们的发声渠道并分析其诉求。 时。
+
+## Inputs
+
+- `session_id`: 研究 session 标识
+- 当前阶段之前的所有研究产物
+
+## Outputs
+
+- `{process_slug}/research/{session_id}/stakeholder-research.md`
+
+## Rules
+
+- 严格基于已有研究产物，避免引入未经验证的假设。
+- 所有外部事实必须标注来源。
+- human_gate 阶段必须获得人类 PM 显式确认。

package/framework/.claude/skills/lc-stakeholder-research/prompts/main.md

@@ -0,0 +1,65 @@
+# lc-stakeholder-research prompt
+
+You are executing the Lincoln workflow step `lc-stakeholder-research`: identify and analyze the stakeholders surrounding the product decision.
+
+## Goal
+
+Produce `{process_slug}/research/{session_id}/stakeholders.md` that maps who uses, pays for, builds, and maintains the product, where they speak, and what they need.
+
+## Inputs
+
+- `session_id`: the research session identifier
+- `{process_slug}/research/{session_id}/scope.md`
+- `{process_slug}/research/{session_id}/first-principles.md`
+
+## Steps
+
+1. Read the scope and first-principles documents.
+2. List stakeholder categories:
+   - End users (primary, secondary)
+   - Economic buyers / budget owners
+   - Builders / implementers
+   - Maintainers / operators
+   - Influencers / gatekeepers
+3. For each stakeholder, identify:
+   - Goals and pains
+   - Where they publicly voice opinions (forums, Slack, X/Twitter, Reddit, GitHub, app stores, review sites, interviews)
+   - Evidence of their voice (quotes, metrics, behavioral signals)
+4. Use WebSearch / WebFetch / GitHub MCP to collect representative quotes and data points.
+5. Write `stakeholders.md` using the template below.
+
+## Output template
+
+```markdown
+# Stakeholder Research: <Title>
+
+## Stakeholder Map
+
+| Role | Who | Goals | Pains | Voice Channels | Evidence |
+|------|-----|-------|-------|----------------|----------|
+| ...  | ... | ...   | ...   | ...            | ...      |
+
+## Key Quotes
+
+> "..." — source, channel, date
+
+## Insight Summary
+
+- <stakeholder tension 1>
+- <stakeholder tension 2>
+
+## Implications for Research
+
+- <what this means for the product decision>
+```
+
+## Rules
+
+- Every claim about a stakeholder must have a source.
+- Distinguish observed behavior from inferred needs.
+- Respect privacy: do not scrape private communities without authorization.
+
+## Exit Criteria
+
+- `stakeholders.md` covers all relevant stakeholder categories.
+- Each category links to at least one voice channel and one piece of evidence.

package/framework/.claude/skills/lc-status/SKILL.md

@@ -0,0 +1,25 @@
+---
+name: lc-status
+description: 查看当前 Lincoln 分支的阶段状态、等待对象、推荐技能和下一步动作
+triggers:
+  - "查看 Lincoln 状态"
+  - "lc-status"
+  - "当前阶段"
+inputs: []
+outputs:
+  - 结构化状态报告
+required_tools:
+  - Read
+  - Bash
+---
+
+# lc-status
+
+## Purpose
+
+Using [lc-status] to 查看当前 Lincoln 分支的阶段状态、等待对象、推荐技能和下一步动作.
+
+
+调用 `python scripts/lincoln-status.py --format markdown` 输出当前 Lincoln 分支状态。
+
+输出包含：当前阶段、等待对象、已加载上下文、推荐技能、产物状态、下一步动作。

package/framework/.claude/skills/lc-storytelling/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: lc-storytelling
+description: 基于金字塔原理与科学论述逻辑，将研究成果结构化为动人的、可说服决策者的故事。
+version: 1.0.0
+triggers:
+  - "lc-storytelling"
+inputs:
+  - name: topic
+    description: 研究主题或产品决策背景
+    required: false
+outputs:
+  - "{process_slug}/research/{session_id}/narrative.md"
+required_tools:
+  - Bash
+  - Read
+  - Write
+---
+
+# lc-storytelling
+
+## Purpose
+
+Using [lc-storytelling] to 基于金字塔原理与科学论述逻辑，将研究成果结构化为动人的、可说服决策者的故事.
+
+## When to Use
+
+- `pm-research` workflow 的 `lc-storytelling` 阶段。
+- PM 需要 基于金字塔原理与科学论述逻辑，将研究成果结构化为动人的、可说服决策者的故事。 时。
+
+## Inputs
+
+- `session_id`: 研究 session 标识
+- 当前阶段之前的所有研究产物
+
+## Outputs
+
+- `{process_slug}/research/{session_id}/storytelling.md`
+
+## Rules
+
+- 严格基于已有研究产物，避免引入未经验证的假设。
+- 所有外部事实必须标注来源。
+- human_gate 阶段必须获得人类 PM 显式确认。

package/framework/.claude/skills/lc-storytelling/prompts/main.md

@@ -0,0 +1,77 @@
+# lc-storytelling prompt
+
+You are executing the Lincoln workflow step `lc-storytelling`: structure the research findings into a compelling, evidence-based narrative.
+
+## Goal
+
+Produce `{process_slug}/research/{session_id}/narrative.md` that presents the research as a clear story using pyramid principle reasoning (thesis → sub-theses → evidence).
+
+## Inputs
+
+- `session_id`: the research session identifier
+- `{process_slug}/research/{session_id}/scope.md`
+- `{process_slug}/research/{session_id}/first-principles.md`
+- `{process_slug}/research/{session_id}/analysis-frameworks.md`
+- `{process_slug}/research/{session_id}/collected-intelligence.md`
+
+## Steps
+
+1. Read all prior research artifacts.
+2. Identify the single core thesis (the answer to the research question).
+3. Derive 3-5 supporting sub-theses. Each sub-thesis must be backed by evidence.
+4. For each sub-thesis, list 2-4 pieces of evidence (data, quotes, analysis).
+5. Arrange the narrative using one of these structures:
+   - SCQA (Situation, Complication, Question, Answer)
+   - Thesis → Sub-thesis → Evidence
+   - Major premise → Minor premise → Conclusion
+6. Add a story arc: setup → tension → turning point → resolution → call to action.
+7. Write `narrative.md` using the template below.
+8. Present the narrative to the PM for confirmation before writing the final report.
+
+## Output template
+
+```markdown
+# Research Narrative: <Title>
+
+## Core Thesis
+
+<one-sentence conclusion>
+
+## Story Arc
+
+- **Setup:** ...
+- **Tension:** ...
+- **Turning point:** ...
+- **Resolution:** ...
+- **Call to action:** ...
+
+## Supporting Sub-theses
+
+### Sub-thesis 1: <statement>
+- Evidence: ...
+- Evidence: ...
+
+### Sub-thesis 2: ...
+
+## Logical Structure
+
+- **Major premise:** ...
+- **Minor premise:** ...
+- **Conclusion:** ...
+
+## PM Confirmation Questions
+
+1. <Is this the right thesis?>
+2. <Is the evidence sufficient?>
+3. <What would make this story more convincing?>
+```
+
+## Human Interaction Rules
+
+- Present the narrative and ask for PM confirmation.
+- Present the narrative for PM feedback; incorporate any revisions, then proceed to the final report.
+- If the PM challenges a sub-thesis, return to the evidence and revise.
+
+## Exit Criteria
+
+- `narrative.md` has a reviewed core thesis and evidence-backed sub-theses (revisions incorporated if any).