agentleFS
cellinlabcellinlab/cell-skills/skills/xhs-title-selector/SKILL.md

xhs-title-selector

Skill29 starsChanged 7 months agoMITView on GitHub

Xiaohongshu title strategy skill for Chinese creators. Use when Codex needs to pick the right title formula for a topic, draft, or existing Xiaohongshu title; match the content to the right trigger family instead of free-generating titles; produce traceable title options with formula IDs and rationale; and choose the best testing candidates.

The file

Show its rules and text

We split this skill into the rules it gives an agent

23 rules. About 0 are common, 22 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈22

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to treat Xiaohongshu titles as a matching problem, not a random generation problem.

    line 11
  2. Distinctive:

    Check whether the topic itself is strong enough to deserve title work.

    line 25
  3. Distinctive:

    If the topic itself is weak or unclear, say so before generating titles.

    line 30
  4. Distinctive:

    use the formula catalog instead of pure freeform generation

    line 38
  5. Distinctive:

    keep titles concise and testable

    line 40
  6. Distinctive:

    prefer titles that can actually be paid off by the body

    line 41
  7. Distinctive:

    If the topic is still too fuzzy, read references/workflow.md.

    line 60
  8. Distinctive:

    Before matching formulas, ask:

    line 64
  9. Very specific:

    If the topic is weak, recommend $content-strategy-diagnosis instead of pretending title work can rescue it.

    line 70
  10. Distinctive:

    Use the trigger-family map in references/trigger-families.md.

    line 74
  11. Distinctive:

    Do not force families that do not fit.

    line 91
  12. Distinctive:

    keep the formula logic intact

    line 99
  13. Distinctive:

    Read references/quality-bar.md before finalizing.

    line 113
  14. Distinctive:

    If the user already has a title:

    line 117
  15. Distinctive:

    Default to assets/title-pack-template.md.

    line 127
  16. Distinctive:

    use niche jargon if broader language would widen relevance

    line 144
  17. Distinctive:

    keep multiple strategic directions alive

    line 151
  18. Distinctive:

    keep the final recommendation opinionated

    line 153
  19. Distinctive:

    Read for input-state handling, title-before-content checks, and existing-title optimization.

    line 158
  20. Distinctive:

    Read for the 12 trigger families and when each should be used.

    line 160
  21. Distinctive:

    Read for the traceable formula list used in title generation.

    line 162
  22. Distinctive:

    Read for title quality rules, compression guidance, and failure checks.

    line 164
  23. Distinctive:

    Use for the standard output shape.

    line 166

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

XHS Title Selector

Overview

Use this skill to treat Xiaohongshu titles as a matching problem, not a random generation problem.

The core job is to:

  • identify what the content is really selling in the first glance
  • match it to the right trigger family
  • generate titles that can be traced back to a formula
  • explain why each title fits

This is a title-strategy skill, not a generic copywriting skill.

Quick Start

  1. First understand the topic, audience, and content goal.
  2. Check whether the topic itself is strong enough to deserve title work.
  3. Match the content to 3-6 trigger families.
  4. Select 5-8 formulas from the catalog instead of free-generating blindly.
  5. Return title options with formula IDs, rationale, and Top 3 recommendations.

If the topic itself is weak or unclear, say so before generating titles.

Default Contract

Assume the following unless the user says otherwise:

  • write in Chinese
  • generate Xiaohongshu-oriented titles
  • use the formula catalog instead of pure freeform generation
  • cover at least 3 distinct trigger families unless the topic is unusually narrow
  • keep titles concise and testable
  • prefer titles that can actually be paid off by the body

Workflow

Step 1: Understand the Input

The user may provide:

  • a topic
  • an industry plus topic
  • a paragraph or full draft
  • a current title that needs diagnosis

Lock these first:

  • what the content is about
  • who it is for
  • what reaction the user wants: click, save, trust, comment, or conversion

If the topic is still too fuzzy, read references/workflow.md.

Step 2: Decide Whether Title Work Should Start Yet

Before matching formulas, ask:

  • does the content have a clear promise
  • is there enough material to support a stronger title
  • is the issue really the title, or is the topic itself weak

If the topic is weak, recommend $content-strategy-diagnosis instead of pretending title work can rescue it.

Step 3: Match Trigger Families

Use the trigger-family map in references/trigger-families.md.

Typical matching logic:

  • counterintuitive claim -> cognitive conflict
  • hidden knowledge or inside information -> curiosity gap
  • warnings, pitfalls, correction -> loss avoidance
  • precise audience callout -> identity mirror
  • steps, methods, lists -> numeric anchor
  • measurable outcome -> outcome promise
  • story or personal turnaround -> case reversal
  • strong stance or potential debate -> controversy
  • specific user state -> scenario or condition
  • push to action -> action call
  • borrowed credibility -> authority leverage
  • comments or self-check desire -> interaction or test

Do not force families that do not fit.

Step 4: Select Formulas

Choose 5-8 formulas from references/formula-catalog.md.

For each formula:

  • keep the formula logic intact
  • adapt the wording to the user's topic
  • label the formula ID
  • explain why it fits this content

Step 5: Generate and Rank Titles

Return:

  • grouped title options by trigger family
  • formula IDs
  • one-line rationale
  • Top 3 recommendations for immediate testing

Read references/quality-bar.md before finalizing.

Step 6: Handle Existing Titles

If the user already has a title:

  1. infer which family and formula it is closest to
  2. diagnose whether the problem is formula choice or execution quality
  3. provide:
    • same-formula improvements
    • alternative-family replacements

Output Format

Default to assets/title-pack-template.md.

At minimum, include:

  • topic and audience read
  • matched trigger families
  • title options with formula IDs
  • Top 3 recommendation
  • brief next-step advice

Hard Rules

Do not:

  • free-generate titles without formula attribution
  • spoil the whole answer inside the title when suspense is needed
  • stuff every option into one trigger family
  • use niche jargon if broader language would widen relevance
  • promise outcomes the content cannot support

Always:

  • identify topic and audience first
  • make formula choice explicit
  • keep multiple strategic directions alive
  • distinguish between wrong formula and weak execution
  • keep the final recommendation opinionated

Resource Map

  • references/workflow.md
    • Read for input-state handling, title-before-content checks, and existing-title optimization.
  • references/trigger-families.md
    • Read for the 12 trigger families and when each should be used.
  • references/formula-catalog.md
    • Read for the traceable formula list used in title generation.
  • references/quality-bar.md
    • Read for title quality rules, compression guidance, and failure checks.
  • assets/title-pack-template.md
    • Use for the standard output shape.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.