微信爆文选题 (WeChat Hot-Topic Researcher)
Overview
Helps a WeChat official-account operator discover popular (10W+/10万+) articles within a given industry keyword, for topic ideation and competitor analysis. Given a keyword, it searches accessible web sources (搜狗微信、新榜、西瓜数据、微信指数、聚合爆文站等), screens for high-heat articles, and outputs a structured list (with data source and confidence tagging) plus actionable topic insights.
Core principle: real data first, never fabricate read counts, always label source & confidence.
When to use
- The user runs a 公众号 and asks for 选题参考 / 竞品分析.
- The user provides an industry keyword and asks to find 微信热门文章 / 爆文 / 10万+ / 阅读量高 的内容.
- Trigger phrases: "帮我找…热门文章", "选题调研", "竞品爆文", "这个领域有什么 10 万+", "分析一下…的爆款".
Workflow
Step 1 — Clarify & scope the keyword
- Accept the keyword. If only a vague domain is given, ask one concise question (target audience / region / time window) before searching; if a concrete keyword is already provided, proceed directly.
- Default time window: latest 6–12 months unless the user specifies. Default region: 中国大陆.
- Record the exact keyword (and any synonyms) for query reuse.
Step 2 — Multi-source discovery (WebSearch)
Run several independent queries in parallel where possible. WeChat article read counts are usually not exposed in search snippets, so cast a wide net and rely on alternative heat signals. Use the query library in references/sources_and_queries.md. Suggested base queries (replace {K} with the keyword):
微信公众号 {K} 10万+ 阅读 爆文搜狗微信 {K} 热门文章新榜 {K} 公众号 文章 阅读量{K} 微信爆文 10万+{K} 公众号 刷屏 热文site:mp.weixin.qq.com {K}(locates actual articles; note read counts are hidden on the page)
Capture for each candidate: 标题、来源公众号、URL、任何明示的阅读量/热度、发布日期、信号来源。
Step 3 — Verify read counts & assign confidence
For each candidate, determine a read-count / heat value and a confidence tag. Rubric (full detail in references):
- 已核实 (verified): a source explicitly states "10万+" / an exact number / a third-party heat value tied to that specific article (e.g. 新榜/西瓜 单篇热度、页面"在看/点赞"数、搜一搜置顶 + 10万+ 标识).
- 估算 (estimated): no per-article number, but a strong proxy exists (account has 100W+ followers and an aggregator flags the article 10万+; or 微信指数 peak + 搜一搜 top rank for the same title).
- 未核实 (unverified): only rumor / screenshot / secondary claim with no retrievable source. Include only if clearly labeled, otherwise drop.
Never invent a specific read number. If unverifiable, write "未核实（无公开数据）" and state the proxy used.
Step 4 — Filter to 10W+ / high-heat
- Keep articles that are 10万+ (verified or estimated).
- If too few verified exist, include estimated ones clearly tagged and note the gap.
- If none are found even by estimate, report honestly: state the searches performed, sources checked, and that no 10W+ content surfaced for this keyword (the niche may be low-volume) — then propose 2–3 adjacent keywords to retry.
Step 5 — Build the structured table (strict format)
Output exactly the following structure:
一、热门文章列表（表格）
| 序号 | 文章标题 | 来源公众号 | 阅读量/热度 | 发布日期 | 核心选题角度 | 链接 |
| --- | --- | --- | --- | --- | --- | --- |
Format every 阅读量/热度 cell as: 值（来源：X｜置信度：已核实/估算/未核实）. Every row must carry a confidence tag. Add a short footnote listing which sources were checked and any data gaps.
Step 6 — Topic insights (3–5 条)
- Synthesize: common 选题角度、标题套路、情绪钩子、结构特征 across the listed articles. Be concrete — cite 1–2 specific examples per point.
- Then provide 2–3 个下周可直接落地的选题建议, each with:
- 借鉴点: what to borrow from the 10W+ pattern.
- 差异化空间: how the user's account (domain / voice / data) can differentiate.
- Tie suggestions back to the user's stated industry/domain.
Step 7 — Honesty & quality gate
Before finalizing, verify:
- No fabricated numbers.
- Every row has a confidence tag + named source.
- Insights are specific and actionable (avoid vague platitudes like "内容要优质").
- Data gaps and limitations are stated openly.
Output template (must follow)
一、热门文章列表（表格）
| 序号 | 文章标题 | 来源公众号 | 阅读量/热度 | 发布日期 | 核心选题角度 | 链接 |
| --- | --- | --- | --- | --- | --- | --- |
（每条 阅读量/热度 标注：值（来源：X｜置信度：已核实/估算/未核实））
注：本次检索覆盖 [来源列表]；数据缺口：[说明]。
二、选题洞察（3-5 条）
• [共同选题角度 / 标题套路 / 情绪钩子 / 结构特征，附 1–2 个实例]
• …
【下周可直接落地的选题建议】
1. 选题：… ｜ 借鉴点：… ｜ 差异化空间：…
2. 选题：… ｜ 借鉴点：… ｜ 差异化空间：…
Resources
references/sources_and_queries.md— query library, searchable source list, confidence rubric, and WeChat read-count caveats.
Notes / limitations
- WeChat read counts are gated behind the platform; treat most figures as estimated unless a third-party analytics source states them. Always label.
- WebSearch result availability varies; if a key source is blocked or returns little, note it and try the next query in the library.
- This skill researches public web signals only — it does not log into any private analytics dashboard.
