News Story Fact-Checking and Editing
Paste the following prompt into your AI chat to install this skill:
Please install @user_a5c5f999/newsgatekeeper according to https://skillhub.cn/install/skillhub.md.
About this skill
What problem it solves
Manual pre-publication review can miss typos, misplaced facts, unsupported data, absolute claims, and public-opinion risk. The problem gets worse when the draft comes from screenshots, posters, or long images, because text extraction can introduce uncertain characters. This skill turns news-draft review into a repeatable workflow: obtain a reviewable text version, check it against editorial rules, and output a risk-graded issue list instead of generic “polishing.”
How it works
- Text or image input: images can be transcribed visually or extracted with
scripts/ocr.swift; uncertain OCR characters should be marked as “needs verification.” - Eight-dimension checks: language basics, news facts,
5W1Hcompleteness, structure and logic, objectivity, public-opinion risk, title review, and final grading. - Evidence-based findings: each conclusion cites a specific location and excerpt. Clear errors are marked 🔴, expression risks 🟡, and optimization suggestions 🟢.
- Structured report: the output follows a standard template with overall assessment, issue list,
5W1Hcheck, logic issues, risk scan, and editing advice; a revised draft is provided only on request.
Boundaries
This skill is for editorial quality checks on news stories, features, comments, social posts, and press releases. It does not replace fact-checking or investigation, and it should not add unconfirmed facts. scripts/ocr.swift is macOS-only; on other platforms, use visual transcription and keep “needs verification” flags for blurry or uncommon characters.
Use Cases
- Before publishing a government news draft, check typos, data sources, and public-opinion risk.
- Extract text from screenshots or long images, then review the story for factual accuracy and wording.
- Ask the skill to flag clear errors, expression risks, and optimization suggestions in a ready-to-publish draft.
- Check whether the title is misleading and whether the body has complete 5W1H details and clear logic.
Best For
- Editors handling final pre-publication review who need to locate factual errors and risky wording quickly.
- Social media operators checking posts or posters for typos, data claims, and sensitive language before release.
- Journalists writing press releases who want the draft reviewed for structure, objectivity, and title accuracy.
- Reviewers working from screenshots who need image text extraction before clause-by-clause fact checking.
Related Skills
Rewrite self-introductions by scenario, audience, and duration, with memory hooks, structure feedback, and speakable practice versions.
Helps AI Agents apply local CSS themes to Markdown articles with wenyan-cli, including preview, WeChat draft publishing, and theme registration.
Turn Markdown or plain text into styles for Xiaohongshu, WeChat Official Accounts, Zhihu, Weibo, Douyin, Bilibili, and Toutiao.
A multi-platform writing workflow for WeChat that emphasizes information gain, scannability, SEO/GEO, and localization to produce verifiable, high-signal blog posts.