Industry News Insight

Overview

Use this skill to turn one industry name into a short, evidence-backed Chinese Markdown article for WeChat-style mobile reading. The output should read like a human industry observer's first-person note, not a formal report, pitch, consultant bio, or AI-generated summary.

Workflow

1. Confirm the target industry and date window.

   • Default to the most recent 30 days from the current date.
   • If the industry name is broad, keep it broad unless the user specifies a sub-sector; avoid silently narrowing the scope.
   • State the date window in the working notes or final response, but do not over-explain it inside the article.

2. Search current evidence.

   • Search web/news sources for the industry name plus terms such as 新闻, 政策, 投融资, 价格, 产能, 需求, 出口, 技术, 并购, supply, demand, market, regulation, funding, earnings, capacity, and price.
   • Prioritize sources published within the last 30 days. Use older sources only for background or to explain a structural fact, and label them as background in notes.
   • Prefer primary or high-signal sources: regulators, exchanges, statistics agencies, company announcements, earnings materials, industry associations, reputable business media, and credible research institutions.
   • Collect enough evidence to cover at least three angles among production端, 消费端, 生产资料/上游投入, 渠道/生态, 政策监管, 技术变化, 资本市场/投融资, and international trade.

3. Validate the story before writing.

   • Separate facts from interpretation.
   • Look for contradictions across sources; resolve them by source quality, date, and measurement definition.
   • Do not write if the last-month evidence is too thin, stale, promotional, or one-sided. Tell the user what is missing instead of producing a weak article.
   • Add supplemental searches when a key causal link depends on missing evidence.

4. Build the article thesis.

   • Use a "news signal -> industry mechanism -> what may change next" structure.
   • Avoid a simple news roundup. Extract one core judgment from the evidence.
   • Make the judgment neutral and falsifiable; avoid investment advice, buy/sell calls, and unsupported predictions.

5. Write for mobile WeChat reading.

   • Write in Chinese unless the user asks otherwise.
   • Keep the article under four mobile screens: normally 900-1500 Chinese characters, excluding references and diagrams.
   • Use short paragraphs, compact headings, and direct sentences.
   • Use first person naturally, such as "我更关注的是..." or "我把这件事放到产业链里看...".
   • Do not reveal an author role. Avoid phrases like "作为第三方顾问", "作为AI", "本文由AI生成", "我们团队", or "研究报告认为".
   • Do not use obvious AI template language such as "综上所述", "值得注意的是" repeated mechanically, "未来可期", or inflated slogans.
   • Do not fabricate quotes, statistics, rankings, or company positions.

6. Add visuals only when they improve comprehension.

   • Prefer one compact Mermaid, SVG, or ASCII diagram when it clarifies a value chain, feedback loop, or supply-demand relation.
   • Keep visuals mobile-readable and text-light.
   • Skip visuals when they would consume space without adding insight.

7. Save the Markdown.

   • Create or use the current project's markdown/ directory.
   • Save as markdown/<YYYYMMDD>-<industry-slug>-industry-news-insight.md.
   • Include the article title, body, optional diagram, and a compact 参考来源 section with source names, dates, and links.
   • In the final response, report the saved absolute path and a brief evidence summary.

Article Shape

Use this structure unless the user requests another format:

# <短标题：行业名 + 核心判断>

<开场：最近一个月最关键的新闻信号，用一两段进入，不写套话。>

## <机制标题>

<解释生产端、消费端、上游或生态中的关键变化。>

## <真正的变化>

<给出第一人称观察和行业机制判断。>

<可选 Mermaid/SVG/ASCII 图>

## <接下来盯什么>

<列出2-4个后续观察指标，不写投资建议。>

---

参考来源：
- <来源名>，<发布日期>，<链接>

Quality Bar

• The article must be readable as a standalone public-account post.
• Every specific data point or event must be traceable to a source.
• The core claim must follow from the cited evidence, not from generic industry common sense.
• The article should feel concise, observational, and human; remove bureaucratic framing, consultant disclaimers, and AI-style transitions before saving.