agentleFS
Fei2-LabsFei2-Labs/skill-genie/skills/linuxdo-application/SKILL.md

linuxdo-application

Skill12 starsChanged 48 days agoMITView on GitHub

Craft high-pass-rate, de-AI'd Chinese applications (小作文) for Linux.do registration. Conducts an adaptive survey to learn the applicant's real background, then generates natural, rule-compliant plain text ready to paste. All user interactions in Chinese.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 14 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈14

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask one question at a time.

    line 31
  2. Distinctive:

    If an answer is vague (e.g., "想学习交流"), probe deeper: "具体想学什么?在哪看到过相关讨论?"

    line 42
  3. Distinctive:

    If an answer already covers multiple blocks, skip redundant questions

    line 43
  4. Distinctive:

    If the applicant mentions a specific post/topic, ask them to elaborate — this is gold

    line 44
  5. Very specific:

    Stop when all 4 info blocks are covered (see references/linuxdo-rules.md)

    line 45
  6. Distinctive:

    Verification: Before moving to Phase 2, confirm internally:

    line 47
  7. Distinctive:

    If any block is missing, ask one more targeted question.

    line 53
  8. Distinctive:

    Write in first person, casual Chinese — like telling a friend why you signed up

    line 60
  9. Distinctive:

    Use the applicant's own words and phrasing where possible

    line 63
  10. Distinctive:

    No greetings, no sign-offs, no "你好" or "谢谢" — just the substance

    line 64
  11. Distinctive:

    If any HIGH risk is triggered, rewrite before proceeding.

    line 79
  12. Very specific:

    Run the draft through the full de-AI checklist (see references/de-ai-checklist.md).

    line 83
  13. Distinctive:

    Check all 12 AI markers — flag and rewrite any matches

    line 86
  14. Distinctive:

    Read the text as if you're a human reviewer — does it smell like AI?

    line 87
  15. Distinctive:

    If the user wants changes, revise and re-run Phase 3 checks.

    line 96
  16. Distinctive:

    After final output, confirm:

    line 111
  17. Very specific:

    Rule updates: When Linux.do changes registration rules, update references/linuxdo-rules.md

    line 122

≈ “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

Linux.do Application Writer

Generate a natural, high-pass-rate registration application for Linux.do through an adaptive interview followed by rule-aware, de-AI'd text generation.

Triggers

  • "write linux.do application"
  • "linux.do 小作文"
  • "linux.do 注册申请"
  • "help me apply to linux.do"

Quick Reference

InputOutputDuration
Answers to 5-8 survey questionsPlain text Chinese application (~80-150 chars)5-10 min

Process

Phase 1: Applicant Survey

Conduct an adaptive interview in Chinese to collect the 4 required info blocks. Ask one question at a time. Adapt follow-ups based on answers.

Core questions (ask in Chinese, adapt order based on flow):

  1. 你平时主要做什么?(工作、学习、兴趣方向)
  2. 你是怎么知道 Linux.do 的?(搜索、朋友推荐、看到某个帖子?)
  3. 你在上面浏览过哪些内容?有没有印象深的帖子或话题?
  4. 为什么现在想注册?有什么具体的需求或场景吗?
  5. 注册之后你打算怎么用?(潜水、回帖、关注某类话题、分享经验?)

Adaptive rules:

  • If an answer is vague (e.g., "想学习交流"), probe deeper: "具体想学什么?在哪看到过相关讨论?"
  • If an answer already covers multiple blocks, skip redundant questions
  • If the applicant mentions a specific post/topic, ask them to elaborate — this is gold
  • Stop when all 4 info blocks are covered (see references/linuxdo-rules.md)

Verification: Before moving to Phase 2, confirm internally:

  • Background covered (what they do/follow)
  • Discovery path covered (how they found the site)
  • Join reason covered (why register now, specific scenario)
  • Usage plan covered (what they'll do with the account)

If any block is missing, ask one more targeted question.

Phase 2: Draft Generation + Risk Check

Generate the application draft using collected info.

Generation rules:

  1. Write in first person, casual Chinese — like telling a friend why you signed up
  2. Target 80-150 characters. Not too short (looks lazy), not too long (looks try-hard)
  3. Weave all 4 info blocks naturally — do NOT use a 4-paragraph structure
  4. Use the applicant's own words and phrasing where possible
  5. No greetings, no sign-offs, no "你好" or "谢谢" — just the substance
  6. Allow sentence fragments, colloquialisms, and imperfect grammar
  7. Vary sentence length: mix short punchy lines with longer ones

Risk check — scan the draft against these (see references/linuxdo-rules.md):

Risk LevelCheck
HIGHDoes it mention background? Discovery path? Join reason?
HIGHIs there any concrete fact, or is it all fluff?
HIGHDoes it look like a template that could apply to anyone?
MEDIUMIs it only admiration/flattery without substance?
MEDIUMIs it only "I can contribute X" without explaining why here?
MEDIUMIs info density too low for the character count?

If any HIGH risk is triggered, rewrite before proceeding.

Phase 3: De-AI Polish + Final Output

Run the draft through the full de-AI checklist (see references/de-ai-checklist.md).

Audit steps:

  1. Check all 12 AI markers — flag and rewrite any matches
  2. Read the text as if you're a human reviewer — does it smell like AI?
  3. Apply the 3-question smell test:
    • Could you swap details and reuse this for someone else? → too generic
    • Does every sentence add new information? → cut filler
    • Would you text this to a friend? → loosen if too formal

Output: Present the final application as plain text in a code block. Tell the user in Chinese: "这是你的申请文,可以直接复制粘贴到注册页面。"

If the user wants changes, revise and re-run Phase 3 checks.

Anti-Patterns

AvoidWhyInstead
Parallel structures (我喜欢X,我热爱Y)AI marker #1Vary grammar patterns
三段式 (首先/其次/最后)AI marker #2Drop scaffolding
AI vocab (赋能/深耕/沉淀/赛道)AI marker #5Use plain spoken Chinese
Flattery (久仰大名/慕名而来)High-risk per rulesReplace with real discovery story
Generic motivation (想学习交流)Medium-risk per rulesState specific use case
Press-release toneAI marker #9Add 吧/嘛/其实/反正

Verification

After final output, confirm:

  • All 4 info blocks present
  • 80-150 characters
  • Zero high-risk patterns
  • Passes all 12 de-AI markers
  • Reads like a real person wrote it
  • Plain text, no formatting

Extension Points

  1. Rule updates: When Linux.do changes registration rules, update references/linuxdo-rules.md
  2. New AI markers: As AI detection evolves, add markers to references/de-ai-checklist.md
  3. Multi-community: Adapt the survey + rule framework for other invite-only communities

References

  • Linux.do Rules & Signals — official rules + community patterns
  • De-AI Checklist — 12 AI markers to detect and fix

Related Skills

SkillUse When
humanizer-zhAdditional de-AI processing for Chinese text
psychology-masterDeeper user profiling during survey
wechat-compliance-checkNeed to check Chinese content for platform compliance

Always run openskills list before starting to discover and load any relevant skills that can enhance the application writing process.

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.