技能 yuwen-publish-precheck

📦

yuwen-publish-precheck

Name: yuwen-publish-precheck
Author: yuwen-cool

內容修訂 r1 低風險 📁 檔案系統存取⚙️ 外部命令🌐 網路存取

Review Chinese Social Media Drafts Before Publishing

Chinese social media creators need to identify publishing risks without removing useful detail or personal voice. This skill combines local keyword checks with contextual review, targeted revisions, and reusable personal rules.

支援: Claude Codex Code(CC)

🥈 82 銀級

Maintain this Skill? Claim it 🥈 82 銀級

使用我的智慧代理安裝

將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。

智能體請求

Review the Skillstore skill "yuwen-publish-precheck" from https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck.md and its manifest at https://skillstore.io/api/skills/yuwen-cool-yuwen-publish-precheck/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。

Agent 可讀資源

當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。

Markdown 詳情 GET /skills/yuwen-cool-yuwen-publish-precheck.md 已簽名 manifest GET /api/skills/yuwen-cool-yuwen-publish-precheck/manifest 已簽名 lockfile GET /api/skills/yuwen-cool-yuwen-publish-precheck/lockfile

測試它

正在使用「yuwen-publish-precheck」。 Illustrative request: review a Chinese course advertisement for Douyin that guarantees payback within thirty days.

預期結果:

Conclusion: revise before publishing.
Required change: the guaranteed payback claim conflicts with commercial-expression.R03.
Suggested replacement: This course covers practical methods and exercises; individual financial results are not guaranteed.
Next checks: verify supporting claims, confirm commercial disclosures, and recheck the revised Chinese draft.
Boundary: this review does not guarantee platform approval.

正在使用「yuwen-publish-precheck」。 Illustrative request: investigate low views on a Douyin post discussing an AI model, with no platform penalty notice.

預期結果:

Conclusion: the cause remains unknown; low views alone do not establish a policy violation.
Do not treat the model name as a prohibited term without supporting platform evidence.
Check whether the published images, audio, or video were AI-generated and required a platform disclosure.
Compare titles, covers, originality, posting times, and audience response before drawing a conclusion.
Do not add a permanent personal prohibition based only on this observation.

安全審計

低風險

v1 • 2026/10/8 開啟版本化報告

All 97 static findings are false positives involving documentation, Chinese text, local parsing, installation paths, or isolated tests. No evidence found of malicious prompt injection, credential collection, or skill-controlled data exfiltration. One low-severity semantic finding concerns privacy assurances that do not distinguish local scanning from processing by the hosting AI service.

已掃描檔案

2,692

分析行數

複核項目

已忽略的誤報

已確認的安全問題 (1)

低

README.md:108-115 SKILL.md:64-70

Unqualified Local-Only Privacy Assurance

The README promises that drafts and personal data never reach third parties, but also directs users to submit drafts to an AI assistant. Local scanner execution does not ensure local model inference, so this assurance can encourage disclosure of sensitive drafts under an incorrect privacy assumption.

The unconditional privacy claim and AI review workflow are explicit in the cited text. Actual transmission depends on the AI host; no skill-controlled upload or malicious exfiltration was found.

風險因素

📁 檔案系統存取 (17)

data/cases/README.md:16 data/README.md:14 data/wordlists/README.md:5 evals/test_scan.py:66 evals/test_scan.py:68 evals/test_scan.py:101 evals/test_scan.py:109 evals/test_scan.py:123 evals/test_scan.py:129 evals/test_scan.py:141 evals/test_scan.py:143 MAINTAINING.md:9 MAINTAINING.md:29 MAINTAINING.md:33 README.md:105 README.md:105 scripts/scan.py:138

⚙️ 外部命令 (42)

🌐 網路存取 (2)

README.md:99 README.md:105

審計單位: codex

分享與引用此報告

分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。

開啟版本化報告

複製報告連結

https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/yuwen-cool-yuwen-publish-precheck/security.svg)](https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/yuwen-cool-yuwen-publish-precheck/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/yuwen-cool-yuwen-publish-precheck.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

學術引用 (APA · BibTeX · CFF)

APA 引用

yuwen-cool. (2026). yuwen-publish-precheck security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck/audits/1

BibTeX 引用

@techreport{yuwen-cool-yuwen-cool-yuwen-publish-precheck-2026,
  author = {yuwen-cool},
  title = {yuwen-publish-precheck security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "yuwen-publish-precheck security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "yuwen-cool"
date-released: "2026-10-08"
url: "https://skillstore.io/skills/yuwen-cool-yuwen-publish-precheck/audits/1"
identifiers:
  - type: other
    value: "skillstore:yuwen-cool-yuwen-publish-precheck:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore 評分

評分依據證據置信度: 中

架構

可維護性

內容

社群

規範符合度

你可以打造什麼

Review a Creator's Spoken Script

Check a Chinese video script for contextual risks and receive targeted replacements with a platform-specific publishing checklist.

Check Commercial Campaign Copy

Review sponsored posts for unsupported claims, missing disclosures, and off-platform calls to action before sending them for approval.

Investigate a Publishing Penalty

Compare platform notices with the original content, separate evidence from assumptions, and retain confirmed lessons as personal review rules.

試試這些提示

Check a First Draft

Review this Chinese script for Douyin: [draft]. It is a noncommercial personal post. Identify actual risks, explain uncertain points, and give a publishing checklist.

Revise a Sponsored Post

Review this Chinese sponsored post for Xiaohongshu: [draft]. Preserve my tone, identify required disclosures, suggest replacement sentences, and recheck the revised version.

Build Personal Review Rules

Import these terms and explanations as personal review rules: [terms]. Use contextual review unless I explicitly request zero tolerance. Test sample sentences and summarize what changed.

Diagnose and Recheck Regulated Content

Analyze this WeChat Channels notice, Chinese finance draft, and timeline: [evidence]. Separate confirmed causes from uncertainty. Propose compliant revisions, recheck them for Douyin and WeChat Channels, and ask before archiving.

最佳實務

Provide the target platforms, commercial relationship, relevant industry, and complete draft before review.
Verify important policy references and supporting evidence, then recheck every revised draft before publishing.
Redact sensitive information, review the AI host's privacy settings, and decline local archiving when retention is unnecessary.

避免

Treating a keyword match as a confirmed violation or an empty scan as guaranteed approval.
Requesting misspellings, coded substitutions, or other methods intended to evade moderation.
Turning low views or another account's apparent success into permanent rules without reliable evidence.

常見問題

Which AI tools can use this skill?

The listing supports Claude, Codex, and Claude Code. The complete workflow requires an environment that can read local files and run Python.

What dependencies are required?

The scanner requires Python 3.10 or newer and uses only the standard library. Contextual judgment and revisions require the hosting AI assistant.

Which platforms and content types are covered?

It focuses on Chinese scripts, articles, captions, titles, and post text for Douyin, Xiaohongshu, and WeChat Channels. Image checks are best effort.

Does a clean result guarantee approval?

No. Keyword checks only locate candidates, and contextual review cannot predict hidden platform decisions. Important legal or regulated claims still need qualified review.

Where are drafts and review records processed?

The scanner reads locally, and review records are stored locally unless you decline archiving. AI processing may send content to your chosen provider.

Can I add personal terms or exceptions?

Yes. The skill supports wordlists, personal rules, and contextual exceptions. These can reduce keyword false positives but cannot override legal or platform requirements.