WeChat Viral Radar

Use this skill to find public WeChat Official Account hot-article signals and turn them into an actionable content report.

Quick Start

Choose the path that fits the user:

• Chat-only mode: use when the user wants results directly in the conversation or cannot run Python. Follow references/chat_only_workflow.md.
• Script mode: use when local execution is available and the user wants Markdown/CSV/JSON files.
• Manual fallback mode: use when public search is blocked or results are too thin. Use the generated manual_fallback_queries.csv or the query patterns in references/public_sources.md.

Common script commands:

python scripts/start.py --track all --limit 10 --report-style deep
python scripts/start.py --track ai --freshness 7d --limit 10 --ideas-per-article 3
python scripts/start.py --mode hotlist --track all --freshness today --limit 12
python scripts/start.py --track career --keywords "副业,失业,普通人" --limit 10

When To Use

Use this skill when the user says things like:

• "帮我找公众号最近 10 万+ 爆文"
• "看看公众号总榜热点"
• "找 AI / 财经 / 情感 / 教育赛道的爆款标题"
• "拆一下这些公众号爆文为什么火"
• "给我生成今天公众号选题"
• "监控几个竞品公众号的公开文章方向"

Input Parameters

See references/parameters.md for full parameter descriptions and examples.

Core options:

• track: all, ai, finance, emotion, education, career, or another built-in track.
• freshness: today, 7d, or 30d.
• mode: articles or hotlist.
• report-style: concise or deep.
• keywords: extra niche terms.
• limit: maximum article candidates.

Output Shape

A complete report should include:

• insight summary
• top topics
• article candidates
• signal level and confidence
• title formulas
• audience pain points
• creative angle bank
• original writing ideas
• 3-day publishing plan
• risk notes
• source warnings
• manual fallback queries

If no live articles are collected, still produce a useful fallback report with query links, troubleshooting steps, and the next best manual action.

Important Boundaries

• This free public version uses public web/search signals only. It cannot guarantee full coverage of all WeChat articles.
• Public pages usually show 10万+ instead of exact read counts. Treat results as confirmed_100k_public, suspected_viral, or topic_candidate.
• If a public source shows captcha, throttling, expired links, or no parseable results, the script retries and still produces a fallback query pack.
• Do not scrape private dashboards, bypass login, or ignore a platform's access rules.
• Paid rankings, exact read counts, low-follower viral lists, and long-term dashboard monitoring are enhanced-source features for a later version.

Track Names

Built-in track names are defined in references/tracks.json.

Common tracks:

• ai
• business
• finance
• emotion
• career
• education
• health
• parenting
• local_life
• culture
• entertainment
• auto
• real_estate
• all

If the user gives a custom track, map it to the closest built-in track and add the user's terms as extra keywords.

Signal Labels

• confirmed_100k_public: the title/snippet/page explicitly includes 10万+, 100000+, 10w+, or 十万+.
• suspected_viral: public search result has strong hot-topic/title signals but no visible read-count proof.
• topic_candidate: useful for trend research but not enough evidence to call viral.
• source_warning: source access, parsing, or throttling issue; use the fallback query pack.

Use cautious wording. Prefer "公开信号显示" or "疑似爆款候选" unless the public page visibly confirms 10万+.

Public Source Strategy

Start with these public paths:

• Sogou Weixin public article search.
• General web search with 微信公众号 10万+ 爆文 plus track keywords.
• Public list pages or articles from Newrank, Qingbo, Xigua Data, Yiban, and similar platforms, when visible without login.

If live public search is weak or blocked:

1. Check the Source Warnings section.
2. Open manual_fallback_queries.csv.
3. Try the generated Sogou Weixin and general web queries manually.
4. Paste any found article links back into the conversation for analysis.

FAQ And Examples

• Read references/faq.md when the user asks about reliability, exact read counts, paid tools, installation issues, or risk boundaries.
• Read examples/workflows.md for common workflows.
• Read examples/full_report_sample.md to see a complete final-report shape.

Useful Scripts

• scripts/start.py: main CLI wrapper.
• scripts/wechat_viral_radar.py: public search, retry, fallback query pack, scoring, and report generation.
• scripts/doctor.py: environment check.
• scripts/verify_package.py: package validation, demo run, fallback-mode run, and deep-report check.

Read script help with:

python scripts/start.py --help