pm-interview

Preamble (run first)

if [ -f "docs/01-需求调研/需求调研报告.md" ]; then
  echo "✅ 需求调研报告 - 已找到（可复用假设与目标用户）"
else
  echo "⏳ 需求调研报告 - 未找到（可选，缺失时由本技能快速采集）"
fi

---

Write-time boundaries

When only discussing, do not write files; startup checks do not create directories. When the user asks to save artifacts, first confirm the current product, the exact output path, and the overwrite/append choice for existing files; after confirmationonly immediately before the actual Writecreate the parent directory of that output file. For full installation, see skills/_shared/artifact-policy.md, and installing this skill alone also follows this paragraph.

---

Cross-Agent interaction rules

When the process requires interaction with the user:

1. If the current environment supports AskUserQuestion, use AskUserQuestion (best experience).
2. If the current environment does not support AskUserQuestion, you must ask the same question via a normal chat message.
3. Ask only one question at a time.
4. After asking, you must stop the current turn and wait for the user's answer (STOP and WAIT).
5. Do not generate documents or write to docs before the user answers.
6. Existing docs files cannot substitute for the user's answer in this turn.

---

Applicable scenarios

• The user says "user interview", "interview outline", "user research plan", "want to talk with users", "qualitative research"
• Distinguish from pm-demand : pm-demand leans toward desk/secondary research; this skill leans towardprimary qualitative research(interview design and execution).

---

Execution flow

Step 1: Clarify interview goals and hypotheses (main agent - user interaction)

Use AskUserQuestion to ask:

🎯 Interview goal

What do you mainly want to figure out with this interview?

A) Validate requirement hypotheses (whether the feature is truly needed) B) Explore user pain points and scenarios (open-ended discovery) C) Evaluate product experience (usability/satisfaction) D) Understand decision-making and payment (selection/payment motivation)

What is the core hypothesis you want to validate/explore? (one sentence)

Record to variable INTERVIEW_GOAL and HYPOTHESIS。

---

Step 2: Determine interview subjects and sample (main agent)

Combine with the requirements research report (if any) to extract target user personas, and use AskUserQuestion to confirm:

👥 Interview subjects

A) Follow the target user segmentation from the requirements research report B) I will manually specify the group C) Focus on a certain extreme/core user group

Sample size: 5-8 people per segment recommended (stop when saturated); duration: {weeks}

Record SEGMENTS and SAMPLE_PLAN。

---

Step 3: Design the interview outline (main agent + optional subagent)

Generate the outline by stage, default structure:

1. Warm-up(build trust, understand background)
2. Behavior and scenarios(actual usage/alternatives)
3. Pain points and motivations(how the problem is currently solved, pain point intensity)
4. Needs and expectations(expectations for the solution, willingness to pay)
5. Wrap-up(open-ended additions, willingness for follow-up interviews)

Use the Agent tool (optional, for complex topics) to dispatch a subagent to generate segment-specific outlines:

Tool: Task
Parameters:
  subagent_type: "general-purpose"
  description: "访谈提纲生成"
  prompt: |
    你是用户研究专家。请基于以下信息设计一份用户访谈提纲。
    目标：{INTERVIEW_GOAL}
    核心假设：{HYPOTHESIS}
    对象分层：{SEGMENTS}
    要求：开放式问题为主、避免引导性提问、每阶段 3-5 题、标注追问点。
    输出 Markdown 提纲。

The main agent integrates and finalizes.

---

Step 4: Recruitment and Execution Guidelines (Main Agent)

Generate:

• Recruitment Script / Screening Questionnaire(including inclusion/exclusion criteria)
• Execution Guidelines: Questioning techniques (probing with "why" and "can you give an example"), avoiding leading questions, recording and informed consent, time control (45-60 minutes)
• Recording Template: Key points per session + verbatim quotes

---

Step 5: Analysis and Output (Main Agent)

Use the Write tool to generate docs/01-需求调研/用户访谈方案.md：

# {产品名称} 用户访谈方案

## 一、访谈目标与假设
- 目标: {INTERVIEW_GOAL}
- 核心假设: {HYPOTHESIS}
- 待回答的关键问题: {列表}

## 二、对象与样本
| 分层 | 特征 | 样本量 | 招募渠道 |
|------|------|-------|---------|
| {层1} | {特征} | {n} | {渠道} |

## 三、访谈提纲（按阶段）
### 阶段1 暖场
1. {问题}
### 阶段2 行为与场景
1. {问题}（追问：...）
...

## 四、招募与执行
- 招募话术: {话术}
- 执行指引: {要点}
- 记录模板: {模板}

## 五、分析方法
- 归纳编码（开放→主题）
- 假设验证矩阵（支持/反驳/需补充）
- 输出物: 洞察报告 + 用户原话引用

## 六、下一步建议
1. /pm-demand - 整合访谈洞察进入需求调研
2. /pm-clarify - 细化高价值需求
3. /pm-priority - 将洞察转优先级

---

Step 6: Recommend Next Steps (Main Agent)

✅ User interview plan generated:docs/01-需求调研/用户访谈方案.md

Recommended actions:

1. /pm-demand - Integrate interview insights
2. /pm-clarify - Refine requirements
3. /pm-priority - Convert to priorities

---

Output Quality Comparison

✅ Good Example：

• Open-ended question: "How did you handle XX last time? Can you give a specific example?"
• Has a sampling plan and inclusion criteria
• Has analysis methods (coding/hypothesis matrix)

❌ Bad Example：

• Leading question: "Don't you think XX is hard to use?"
• No sampling plan
• No analysis methods

---

Common Pitfalls / Red Flags — STOP

Pitfall	Correct Approach
Using leading questions	Open-ended, neutral, probing with "why"
No sample stratification	Recruit by user segments to avoid homogeneity
Collecting without analyzing	Must define coding and hypothesis validation methods
Treating a single case as a conclusion	Cross-validate across multiple samples, cite verbatim quotes

---

Output Quality Check / Verification Checklist

• Interview goals and hypotheses clearly defined
• Participant stratification and sampling plan determined
• Phased open-ended outline generated
• Recruitment and execution guidelines provided
• Output document generated to docs/01-需求调研/
• 2-3 follow-up skills recommended

⚠️ If any item fails → complete it before marking as done.

---