pm-market

Preamble (run first)

# 检查是否有需求调研报告
if [ ! -f "docs/01-需求调研/需求调研报告.md" ]; then
  echo "⚠️  未找到需求调研报告"
  echo ""
  echo "建议先执行 /pm-demand 完成需求调研"
fi

Write-time boundaries

When only discussing, do not write files; startup checks do not create directories. When the user requests saving artifacts, first confirm the current product, the exact output path, and the overwrite/append choice for existing files; after confirmation,only 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:

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

Execution flow

Step 1: Read prerequisite data (main agent)

If there is a requirements research report：

Use the Read tool to read docs/01-需求调研/需求调研报告.md

Extract:

Product name
Industry track
Target users
Core pain points

If there is no requirements research report：

Enter quick mode and use AskUserQuestion to collect basic information.

Step 2: Determine search directions (main agent - user interaction)

Use AskUserQuestion to ask:

🎯 Which aspects should I focus on analyzing?

A) Market size and growth trends B) Competitor analysis C) User needs and behavior D) Industry policies and risks E) All analyses (recommended but time-consuming)

After the user selects, record it in the variable MARKET_FOCUS

Step 3: Dispatch subagents to conduct market research (core optimization)

Optimization notes：

The main agent is only responsible for dispatching tasks and receiving results
Subagents perform all search and analysis work
Search details do not occupy the main agent's context

3.1 Build the search task list

Based on the user's selection, build the task list:

If "market size" is selected：

[
  {
    "task_id": "market_size",
    "keywords": ["{行业赛道} 市场规模 2025", "{行业赛道} 增长趋势"],
    "domains": ["iresearch.cn", "analysys.cn", "iimedia.cn"]
  }
]

If "competitor analysis" is selected：

[
  {
    "task_id": "competitor_analysis",
    "keywords": ["{行业赛道} 竞品分析", "{行业赛道} 主要玩家"],
    "domains": ["36kr.com", "huxiu.com", "qimai.cn"]
  }
]

If "all" is selected：

[
  {
    "task_id": "market_size",
    "keywords": ["{行业赛道} 市场规模"],
    "domains": ["iresearch.cn", "analysys.cn"]
  },
  {
    "task_id": "competitor_analysis",
    "keywords": ["{行业赛道} 竞品"],
    "domains": ["36kr.com", "huxiu.com"]
  },
  {
    "task_id": "user_behavior",
    "keywords": ["{目标用户} 消费行为"],
    "domains": ["questmobile.com.cn"]
  }
]

3.2 Use the Agent tool to dispatch subagents

Single task：

Use the Agent tool:

Tool: Agent
Parameters:
  subagent_type: "general-purpose"
  description: "市场调研-{task_id}"
  prompt: |
    你是一个市场调研专家。请执行以下任务：

    **任务 ID**: {task_id}
    **产品名称**: {产品名称}
    **行业赛道**: {行业赛道}
    **搜索关键词**: {keywords}
    **目标数据源**: {domains}

    **要求**：
    1. 搜索工具优先级：首选 AnySearch（Bash 调用 `anysearch_cli.py（先定位：检查 ~/.claude/skills/anysearch/scripts/ ~/.opencode/skills/anysearch/scripts/ ~/.openclaw/skills/anysearch/scripts/ ~/.cursor/skills/anysearch/scripts/ ~/.anysearch/scripts/ 或 which anysearch_cli.py，找到后 python3 <路径>）`search "查询词" --max_results 5 --domain finance --sub_domain finance.cn_stock），失败降级到 Exa MCP（mcp__exa__web_search_exa），最后才用 WebSearch。使用 WebSearch 时标注「⚠️ 降级模式」
    2. 搜索每个关键词
    3. 优先搜索目标数据源
    4. 提取关键数据（市场规模、增长率、主要玩家等）
    5. 返回结构化的 JSON 结果

    **输出格式**：
    ```json
    {
      "task_id": "{task_id}",
      "findings": [
        {
          "title": "数据标题",
          "source": "数据来源",
          "value": "数据值",
          "date": "数据日期"
        }
      ],
      "summary": "关键发现总结"
    }
    ```

Multiple tasks (parallel optimization)：

Dispatch multiple subagents simultaneously:

# 并行派发 3 个 subagent
Agent 1: 市场规模调研
Agent 2: 竞品分析
Agent 3: 用户行为分析

The main agent waits for all subagents to complete, then collects the results.

Step 4: Integrate results and generate the report (main agent)

The main agent receives the return results from all subagents：

Read the JSON data returned by the subagents
Integrate all findings
Generate a structured market research report

Use the Write tool to generate the report：

# 市场调研报告

## 一、市场规模与增长趋势

（从 subagent 结果中提取）

## 二、竞品分析

（从 subagent 结果中提取）

## 三、用户需求与行为

（从 subagent 结果中提取）

## 四、关键发现与建议

（主 agent 综合分析）

---

**生成时间**: {时间}
**调研范围**: {MARKET_FOCUS}

Step 5: Recommend next steps (main agent)

✅ Market research report has been generated:docs/01-需求调研/市场调研报告.md

Recommended to execute:

/pm-priority - Prioritize based on market research

/pm-position - Determine product positioning based on competitive analysis

/pm-mvp - Plan the MVP solution

Comparison: Before vs After Optimization

Before Optimization (pm-market v1)

主 agent 执行流程：
1. 询问用户（10 tokens）
2. 执行搜索1（500 tokens 结果）
3. 执行搜索2（500 tokens 结果）
4. 执行搜索3（500 tokens 结果）
5. 整理结果（200 tokens）
6. 生成报告（100 tokens）

总计：1810 tokens
耗时：串行执行，约 3-5 分钟

After Optimization (pm-market-v2)

主 agent 执行流程：
1. 询问用户（10 tokens）
2. 派发 subagent（20 tokens）
   └─ Subagent 执行搜索（不占用主 agent 上下文）
3. 接收结果（100 tokens 结构化数据）
4. 整合报告（100 tokens）

总计：230 tokens
耗时：并行执行，约 1-2 分钟
效率提升：87% token 节省，2.5x 速度提升

Key Optimization Points

1. Context Savings

✅ Search results do not occupy the main agent's context
✅ The main agent only processes structured results
✅ Token usage reduced by 70-90%

2. Execution Efficiency

✅ Multiple search tasks execute in parallel
✅ Total time reduced by 50-70%
✅ Significantly improved user experience

3. Maintainability

✅ The main agent's logic is clear and focused on decision-making
✅ Subagents have a single responsibility and are easy to test
✅ Clearer code structure

Notes

When to Use a Subagent

✅ Suitable：

Web search (results consume a large number of tokens)
Document generation (can be parallelized)
Data analysis (compute-intensive)
Competitive analysis (multiple searches)

❌ Not Suitable：

User interaction (AskUserQuestion)
Simple decisions (if-else judgment)
Quick tasks (< 10 seconds)

Exception Handling

If the subagent fails to execute:

The main agent detects the error
Provide a fallback (manual input or simplified search)
Log the error information for the user to troubleshoot

Version Notes

v2.0.0 - Subagent architecture refactor

Added Agent tool support
Parallelized search tasks
Optimized the main agent's context
Efficiency improved by 2.5x

v1.0.0 - Initial version

Basic market research functionality

Output Quality Comparison

✅ Good Example：

- 有数据引用：「根据 Q4 数据，留存率从 35% 降至 28%」
- 有验证来源：「数据来源：Google Analytics, 2025-12-01」
- 有明确建议：「建议将新手引导步骤从 5 步减少至 3 步」

❌ Bad Example：

- 模糊结论：「数据表明留存率有所下降」
- 无来源：「根据经验，这个功能很重要」
- 没有行动建议：「留存是个问题」

Common Misconceptions / Red Flags — STOP

Stop immediately and backtrack if any of the following occur:

Misconception	Correct Approach
Drawing conclusions using "should", "probably", "seems"	Must be based on actual data and verification
Claiming completion without running checks	Verify first, then state
Skipping key steps due to time pressure	No exceptions; the tighter the time, the stricter you must be
The thought "this time it should be fine"	Re-verify every time

Output Quality Check / Verification Checklist

Prerequisite dependencies are satisfied (input documents/data have been collected)
All core steps have been executed
The output document has been generated to the docs/ directory
Every judgment is supported by data/evidence
2-3 follow-up skills have been recommended

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