Metadata

name: wechat-viral-topic
title: WeChat Viral Topic Alchemy
slug: wechat-viral-topic
display_name: 微信爆款选题炼金炉
display_name_en: WeChat Viral Topic Alchemy
version: 2.0.1
author: SynomosAI
description: 微信爆款选题与内容策划助手——基于「爆款四基因+12 条策划心法+8 维打分卡+6 大内容形态+12 条标题公式+风险自检」的纯规则策划流水线，回答「这周写什么」「这个素材能不能爆」。
description_zh: 微信爆款选题与内容策划：纯规则流水线（爆款四基因+12 心法+8 维打分卡+6 形态+12 标题公式+风险自检），面向公众号/短视频/小红书/漫画/H5，回答写什么、素材能否爆、热点怎么改成选题。
description_en: WeChat viral topic and content planning assistant: a pure-rule pipeline (four viral genes + 12 planning principles + 8-dimension scoring card + 6 content formats + 12 title formulas + risk self-check) for editors and creators, answering what to write, whether a素材 will pop, and how to reframe a hot topic.
license: MIT
tags: ["wechat-viral-topic","爆款选题","微信运营","内容策划","公众号","短视频","小红书"]
brand: SynomosAI
copyright: SynomosAI
platforms: ["windows","macos","linux"]
platform_compatibility: ["windows","macos","linux"]
category: content
classification: {"skillhub":["content","social-media","ideation"],"clawhub":["content","marketing"],"iso_25010":["Functional suitability"]}
read_when: ["用户问「这周写什么/给我选题/10 个方向」需选题发散","手里有素材/热点需改成自己的爆款角度","需要标题公式、长文提纲、短视频脚本、漫画分镜、H5 方案","需要对选题做爆款打分与风险自检"]
ai_protocol: true
spec: usk/1.0
interface: {"type":"prompt","entry_point":"SKILL.md","runtime":"none","call_pattern":"prompt"}
output_schema: {"result":"structured output per SKILL.md (input receipt / Top3 scoring / format plan)"}
exit: no process exit code; main agent executes per doc
capabilities: ["content","wechat","ideation"]
permissions: {"network":false,"filesystem":false,"subprocess":false}
requirements: {"python_packages":[],"min_python":"3.8"}
changelog: 2.0.1: Bilingual rewrite (EN-first) + title/description_en-zh/read_when/classification; 2.0.0 prior

凡自治之物 · LGD 治理学 · 创造术家族｜评测协议：skill-quality-gate

English

WeChat Viral Topic Alchemy

A pure-rule content-planning pipeline (no code execution, no external runtime) that turns a user's material or direction into actionable viral topics and multi-format content plans. Core assets live in references/*.md and templates/output-template.md; this file is the AI entry point and standard contract.

When to use

User asks "what to write this week / will this topic pop / give me 10 topics".
User has material and needs it reframed into their own viral angle.
User needs titles, long-form outline, short-video script, comic storyboard, or H5 plan.
User needs a viral score and risk self-check on a topic.

Not for: layout/illustration/video editing (production); account operation, posting, view-buying, traffic diversion; political/medical/financial high-risk material without compliant sources (refuse outright).

How to use

See "Trigger scenarios" and "Standard SOP" below. Input receipt + 12-principle pre-screen + risk self-check must accompany every output.

Workflow

Collect material + structured input guidance (2.1).
Four-gene match (emotion / information gap / identity / action).
12 planning-principle pre-screen.
8-dimension scoring card (total 40).
Smart Top3 recommendation + matched format (v2).

Boundary

In-scope: topic generation, angle design, title factory (12 formulas × platform word limits), format distribution (long article / short video / comic / H5 / Moments / Xiaohongshu), quality scoring, risk self-check, smart Top3. Out-of-scope: no real-time hot-list scraping (ask user to provide heat), no image/video/code generation, no traffic guarantee, no replacement for human compliance review.

Input & output contract

Structured input fields: material/direction (required), target platform, audience, content format, tone. Missing fields use defaults (2.2) rather than a bare reject. Platform word caps: Moments title ≤22 / Official-account ≤28 / Xiaohongshu ≤20 (emoji counts 2) / Video-account ≤22 / H5 main title ≤14. Output promises: always include input receipt + 12-principle pre-screen + risk self-check; Top3 must carry the 8-dimension score (≥30 pass, ≥35 nuclear); every title passes risk self-check first.

Trigger intent table

Intent	Keywords	Action
Topic divergence	"give me topics"	four-gene divergence + 8-dim score
Material reframe	"how to use this hotspot"	angle design + format distribution
Title optimization	"make a title"	12 formulas × platform limits
Format landing	"write a script"	corresponding format template
Quality assessment	"will this pop"	pre-screen + score + self-check
Rescue	"rejected / no traffic"	risk/weakness diagnosis + alternatives

SOP (no skipping; gates ✅ pass / ⚠️ rework / ❌ reject)

Collect + structured input (⚠️ missing fields → defaults). 2. Four-gene match (✅ ≥2 genes). 3. 12-principle pre-screen (❌ red line → reject material). 4. 8-dim score (✅ ≥30, ≥35 nuclear, <24 reject). 5. Smart Top3 + format. 6. Title factory (≥5 titles per plan). 7. Format distribution. 8. Multi-format scheduling. 9. Risk self-check (10 categories). 10. Assemble per output-template. 11. Error recovery (never hang). Detailed logic: references/four-genes.md, core-principles.md, scoring-card.md, content-formats.md, title-formulas.md, risk-checklist.md; output skeleton: templates/output-template.md.

Key facts

Four viral genes (≥2 per topic, ≥3 nuclear): emotional hook / information gap / identity label / action trigger.
Score thresholds: ≥35 nuclear 💥 / 30–34 pass ✅ / 24–29 edge ⚠️ / <24 reject ❌.
Platform caps: Moments ≤22 / Official-account ≤28 / Xiaohongshu ≤20 / Video-account ≤22 / H5 ≤14.
12 title formulas: identity reversal / number list / counter-intuitive / identity contrast / urgency / insider reveal / Q&A / story hook / contrast / pitfall warning / golden-quote nail / persona self-mockery.
10 risk categories: absolute ad-law words / medical / political-social / anxiety-bait / sexual hints / false promises / group discrimination / fatigue words / brand compliance / platform red lines.

Prohibited (iron rules)

No political sensitivity (leaders, party-state, HK/Macau/Taiwan/Xinjiang/Tibet negatives, living-person scandal, undetermined blame). 2. No fabrication — information gap must be real. 3. No absolute words (最/第一/核心/国家级/100%/永久). 4. No share/follow bait (WeChat blocks links). 5. No despair-peddling. 6. No group discrimination. 7. No view-buying/traffic-diversion/ghost-posting. 8. No image/video/code generation. 9. No output skipping pre-screen + risk check. 10. No unsourced claims — core data traceable or marked "to verify".

Error handling

Error codes: E-INPUT-001/002/003 (missing/vague/format), E-GENE-001 (<2 genes), E-PRINCIPLE-001 (red line), E-RISK-001 (high-risk no source), E-PLATFORM-001 (word/tone), E-SCORE-001 (<24), E-RECOVER-001 (step failure → degrade, never hang). Always emit an E-code + one-line cause + next action.

Evidence & safety

Evidence: TBD; Version: 2.0.1; Security: no independent scan (TBD).
No-keyword dependency: pure-rule pipeline. If future real-time/hot-list features are added, credentials must come from a vault (env injection), never written into SKILL.md or echoed to the user.

中文

微信爆款选题炼金炉

纯规则内容策划流水线（不执行代码、不调外部运行时），把素材/方向炼成可落地的爆款选题与多形态方案。核心资产在 references/*.md 与 templates/output-template.md；本文件是 AI 调用入口与标准契约。

何时用

用户问「这周写什么/这个热点能不能爆/给我 10 个选题」。
手里有素材需改成自己的爆款角度。
需要标题、长文提纲、短视频脚本、漫画分镜、H5 方案。
需要对选题做爆款打分与风险自检。

不用于：排版/配图/剪辑等制作环节；代运营/代发/刷量/导流；政治/医疗/金融高危无信源素材（直接拒绝）。

用法

见「触发场景」与「标准 SOP」。每次输出必带输入回执+12 心法预筛+风险自检。

流程

收料+结构化引导 → 四基因匹配（情绪/信息差/身份/行动）→ 12 心法预筛 → 8 维打分（总分 40）→ 智能 Top3+适配形态。

能力边界

能做：选题生成、角度设计、标题工厂、形态分发、质量打分、风险自检、智能推荐。不能做：抓实时热榜（请用户提供热度）、生成图/视频/代码、保证流量、替代人工合规审核。

输入输出契约

结构化字段：素材/方向（必填）、平台、受众、形态、调性；缺字段按默认值补全而非退回。平台字数上限：朋友圈≤22/公众号≤28/小红书≤20/视频号≤22/H5 主标≤14。每次输出必带三段；Top3 必带 8 维分（≥30 合格，≥35 核弹）；标题先过风险自检。

触发意图表

选题发散/素材改造/标题优化/形态落地/质量评估/救场，按对应动作执行（详见正文表）。

标准 SOP（不许跳步）

收料→四基因匹配→12 心法预筛→8 维打分→Top3+形态→标题工厂→形态分发→排期→风险自检→按模板组装→错误恢复。详细逻辑见 references/ 各文件。

关键事实速查

四基因（≥2 命中，≥3 核弹）：情绪钩子/信息差/身份标签/行动触发。打分阈值：≥35 核弹/30-34 合格/24-29 边缘/<24 淘汰。12 标题公式、10 类风险清单见正文。

铁律（PROHIBITED）

不站队、不造假、不绝对化用词、不诱导分享、不贩卖绝望、不群体歧视、不违规操作、不越界制作、不跳过预筛自检、不无信源断言。

错误处理

错误码 E-INPUT/E-GENE/E-PRINCIPLE/E-RISK/E-PLATFORM/E-SCORE/E-RECOVER；始终输出错误码+一句话原因+下一步动作，不卡死。

证据与安全声明

证据：TBD；版本：2.0.1；安全：未做独立扫描（TBD）。无密钥依赖；未来若加热榜/发布能力，凭据须从保险库读取，严禁写入本文件或回显。

wechat-viral-topic

安装技能