xhs-title-ab-test
Skill1 starsChanged 14 days agoMITView on GitHub
Generates 4-6 Xiaohongshu title variants for the same note and picks one by CTR-hook rules. Use when 起标题, 标题备选, A/B, 同一篇换标题, or writing title.md.
In this repository30 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .cursor/rules/00-agents-standard.mdcCursor rule
- .cursor/rules/dy-publish.mdcCursor rule
- .cursor/rules/jev-claude.mdcCursor rule
- .cursor/rules/xhs-bundesliga.mdcCursor rule
- .cursor/rules/xhs-package-dod.mdcCursor rule
- .cursor/rules/xhs-publish-plain-text.mdcCursor rule
- .cursor/rules/xhs-series-v2.mdcCursor rule
- .cursor/rules/xhs-traffic.mdcCursor rule
- football-content-opsSkill
- atutun-xhs-coverSkill
- dbs-xhs-titleSkill
- douyin-cliSkill
- dy-content-complianceSkill
- dy-culture-rxSkill
- dy-flowus-dbSkill
- dy-note-adaptSkill
- ip-character-systemSkill
- jev-judgeSkill
- tt-note-adaptSkill
- xhs-account-positioningSkill
- xhs-content-complianceSkill
- xhs-flowus-dbSkill
- xhs-note-structureSkill
- xhs-pipelineSkill
- xhs-title-skillSkill
- xhs-topic-bankSkill
- xhs-voice-polishSkill
- xhs-weekly-retroSkill
- xiaohongshu-cliSkill
The file
Files
xhs-title-ab-test
同一篇内容多组标题,不把第一发当唯一解。先读定位卡和选题行。
生成(4–6 条,来源要可追溯)
必须混用两套已有公式,禁止只换形容词:
- 读
skills/dbs-xhs-title/SKILL.md,点名 触发器编号(如 数字锚定 #26、好奇缺口 #7) - 读
skills/xhs-title-skill/SKILL.md,点名 8 模板之一(痛点/悬念/对比…)
每条输出一行:
标题(≤20字含标点) | 公式来源 | 封面大字(6-10字) | 为什么会点至少覆盖三种张力:数字、悬念、冲突/身份。足球快讯要有人名或金额之一;心理要有具体场景,不要「你一定要看」。
五项硬检
- 字数、悬念、不说完答案、用户词够大、≥2 个张力
- 封面大字和标题同钩子,不要标题讲凯恩、封面写「夏窗」
- 过
xhs-content-compliance(无最/第一/唯一、无点赞收藏)
怎么「测」
本号没有标题实验后台。用替代实验,不要假装有统计显著:
- 发前:你挑 1 条进
title.md,其余留在doc/<题>/title-ab.md - 发后 24–48h:看 CTR。
<10%→ 用备选在 App「编辑标题/换封面」再测一组(小红书允许改封面) - 周复盘:把胜出公式写回选题库(例如「金额+位置」比「情怀」香)
输出
- 写入
doc/<题>/title.md(只放采用条) - 写入
doc/<题>/title-ab.md(全部变体) - 聊天里只展示表格,并推荐一条,说明不推荐另外两条的原因(各一句)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

