interview-analysis

使用动态专家路由进行深度访谈分析。根据角色类型自动选择顶级领域思想家，以区分真正的能力和绩效，通过方法论背诵识别战斗伤痕。适用于任何专业职位，包括产品管理、工程、设计、运营、销售和数据科学。

# Interview Analysis Skill

> **Core Mission**: Transform interview transcripts into deep insights.
> **Core Logic**: Don't listen to what candidates "say" (Methodology Recitation), observe what they've "done" (Battle Scars) and "how they think" (First Principles).

## 1. Dynamic Expert Activation (Expert Routing)

### Core Principle
Based on **role type** and **evaluation dimensions**, automatically select the **best minds** combination for that domain:

**Three-Step Expert Selection**:
1. **Identify core competency domain**: Product/Engineering/Operations/Design/Sales/Data Science/...
2. **Match top domain thinkers**: Recognized methodology masters or practitioners in the field
3. **Combine hiring experts**: Geoff Smart (fact-checking) + Lou Adler (competency validation)

### Common Role-Expert Mapping (Non-Exhaustive)

| Role Type | Domain Expert (Methodology) | Hiring Expert (Validation) | Rationale |
|-----------|---------------------------|---------------------------|-----------|
| **Product Manager** | Marty Cagan / Julie Zhuo | Geoff Smart | Product Sense + Fact Check |
| **Software Engineer** | Linus Torvalds / John Carmack | Lou Adler | Engineering Judgment + Results Validation |
| **Growth Hacker** | Sean Ellis / Brian Balfour | Geoff Smart | Growth Methodology + Metrics Verification |
| **UX Designer** | Don Norman / Jony Ive | Lou Adler | UX Principles + Portfolio Validation |
| **Data Scientist** | Andrew Ng / DJ Patil | Geoff Smart | Technical Depth + Project Verification |
| **Operations** | Sheryl Sandberg / Reid Hoffman | Lou Adler | Scale Operations + Results Focus |
| **Sales/BD** | Aaron Ross / Jill Konrath | Geoff Smart | Sales Methodology + Performance Verification |

> [!IMPORTANT]
> **Flexibility Principle**: The table above is for reference only. Flexibly select the most appropriate expert combination based on specific role and candidate background.
> 
> **Encourage Innovation**: If you believe a non-mainstream expert is better suited to evaluate this candidate, make that choice and explain your rationale.
> 
> **Core Question**: "Who can best identify imposters in this role? Whose framework best validates core competencies?"

## 2. Execution Workflow

### Step 1: Fact Reconstruction & Red Flag Scan
*   **Timeline Reconstruction**: Connect experiences scattered across multiple interview rounds, checking for logical gaps.
*   **Consistency Verification**: Compare different versions of the same story told to different interviewers (e.g., reasons for leaving, project failures).
*   **Red Flag Annotation**: Mark all vague titles (e.g., SPM), exaggerated data, and attribution fallacies ("it was all market/technology's fault").

### Step 2: Deep Decoding - STAR Episodes
*   **Tactic**: Select 1-2 core cases (e.g., startup project, most challenging project) for microscopic analysis.
*   **Truth Extraction**:
    *   **Methodology Check**: Is the candidate reciting SOPs (MECE, SWOT) or applying first principles?
    *   **Solution Bias Check**: Did they jump straight to "add features," or first conduct "value validation"?
    *   **Technical Boundary Check**: For technical challenges, did they "deflect blame" or "anticipate"?

### Step 3: Interviewer Meta-Analysis
*   **Subject**: Evaluate interviewer (you/colleagues) performance.
*   **Dimensions**:
    *   **Depth**: Did they probe at critical moments? Or let it pass?
    *   **Bias**: Did they draw conclusions too early or ask leading questions?
    *   **Bar**: Did they maintain A Player standards?

### Step 4: Card-based Output (Zettelkasten Output)
Generate Markdown cards using the following standard templates, saved to `people/{candidate_name}/analysis/`. Be sure to read template content before filling in analysis results.

*   **Profile (Comprehensive Portrait)**:
    *   Template path: `templates/profile_template.md`
    *   Purpose: Fact checking, red flag scanning, core competency assessment.
*   **Insight (Deep Analysis)**:
    *   Template path: `templates/insight_template.md`
    *   Purpose: Deep dive into specific domains (e.g., AI Capability, Product Strategy).
*   **Meta-Analysis (Interviewer Review)**:
    *   Template path: `templates/evaluation_template.md`
    *   Purpose: Evaluate interviewer performance and organizational recommendations.
*   **Structure Note (Hub Document)**:
    *   Template path: `templates/structure_note_template.md`
    *   Purpose: Serves as hub connecting all analysis cards above, forming decision closure.

## 3. Usage Examples

*   "Analyze Li Yashuang's three interview rounds, focusing on AI capabilities."
*   "Review this interview to see where we interviewers did well and where we missed opportunities."
*   "Use Marty Cagan's perspective to analyze this candidate's product thinking."

---

## 中文说明

# Interview Analysis Skill

> **核心使命**：将访谈记录转化为深度洞察。
> **核心逻辑**：不要听候选人"说"了什么（方法论背诵），而要观察他们"做"了什么（战斗伤痕）以及他们"如何思考"（第一性原理）。

## 1. 动态专家激活（专家路由）

### 核心原则
基于**角色类型**和**评估维度**，自动为该领域选择**最佳思想者**组合：

**三步专家选择**：
1. **识别核心能力领域**：产品/工程/运营/设计/销售/数据科学/……
2. **匹配顶级领域思想者**：该领域公认的方法论大师或实践者
3. **组合招聘专家**：Geoff Smart（事实核查）+ Lou Adler（能力验证）

### 常见角色-专家映射（非穷尽）

| 角色类型 | 领域专家（方法论） | 招聘专家（验证） | 理由 |
|-----------|---------------------------|---------------------------|-----------|
| **产品经理** | Marty Cagan / Julie Zhuo | Geoff Smart | 产品感觉 + 事实核查 |
| **软件工程师** | Linus Torvalds / John Carmack | Lou Adler | 工程判断力 + 结果验证 |
| **增长黑客** | Sean Ellis / Brian Balfour | Geoff Smart | 增长方法论 + 指标验证 |
| **UX 设计师** | Don Norman / Jony Ive | Lou Adler | UX 原则 + 作品集验证 |
| **数据科学家** | Andrew Ng / DJ Patil | Geoff Smart | 技术深度 + 项目验证 |
| **运营** | Sheryl Sandberg / Reid Hoffman | Lou Adler | 规模化运营 + 结果导向 |
| **销售/BD** | Aaron Ross / Jill Konrath | Geoff Smart | 销售方法论 + 绩效验证 |

> [!IMPORTANT]
> **灵活性原则**：上表仅供参考。请根据具体角色和候选人背景灵活选择最合适的专家组合。
> 
> **鼓励创新**：如果你认为某位非主流专家更适合评估这位候选人，请做出该选择并说明你的理由。
> 
> **核心问题**："谁最能识别这个角色中的冒牌者？谁的框架最能验证核心能力？"

## 2. 执行工作流

### 第 1 步：事实重构与危险信号扫描
*   **时间线重构**：串联分散在多轮面试中的经历，检查逻辑空白。
*   **一致性验证**：比对同一故事对不同面试官讲述的不同版本（例如，离职原因、项目失败）。
*   **危险信号标注**：标记所有模糊的头衔（例如 SPM）、夸大的数据，以及归因谬误（"全是市场/技术的错"）。

### 第 2 步：深度解码 - STAR 片段
*   **战术**：选择 1-2 个核心案例（例如，创业项目、最具挑战性的项目）进行微观分析。
*   **真相提取**：
    *   **方法论检查**：候选人是在背诵 SOP（MECE、SWOT），还是在运用第一性原理？
    *   **解决方案偏差检查**：他们是直接跳到"加功能"，还是先进行"价值验证"？
    *   **技术边界检查**：面对技术挑战时，他们是"推卸责任"还是"提前预见"？

### 第 3 步：面试官元分析
*   **对象**：评估面试官（你/同事）的表现。
*   **维度**：
    *   **深度**：他们是否在关键时刻深入追问？还是放任而过？
    *   **偏见**：他们是否过早下结论或提出引导性问题？
    *   **标准**：他们是否坚持了 A Player 的标准？

### 第 4 步：卡片式输出（Zettelkasten Output）
使用以下标准模板生成 Markdown 卡片，保存至 `people/{candidate_name}/analysis/`。在填写分析结果前，务必先阅读模板内容。

*   **Profile（综合画像）**：
    *   模板路径：`templates/profile_template.md`
    *   用途：事实核查、危险信号扫描、核心能力评估。
*   **Insight（深度分析）**：
    *   模板路径：`templates/insight_template.md`
    *   用途：深入特定领域（例如 AI 能力、产品战略）。
*   **Meta-Analysis（面试官复盘）**：
    *   模板路径：`templates/evaluation_template.md`
    *   用途：评估面试官表现及组织层面建议。
*   **Structure Note（中枢文档）**：
    *   模板路径：`templates/structure_note_template.md`
    *   用途：作为连接上述所有分析卡片的中枢，形成决策闭环。

## 3. 使用示例

*   "分析李雅双的三轮面试，重点关注 AI 能力。"
*   "复盘这次面试，看看我们面试官哪里做得好，哪里错失了机会。"
*   "用 Marty Cagan 的视角分析这位候选人的产品思维。"