agentsclimarketplace

Chinese blog review

Skill yclgkd/chinese-blog-review-skill/skills/chinese-blog-review

面向中文博客和技术文章的 Agent Skill：像 Code Review 一样审稿，检查标题、结构、事实、证据、风格和中文排版。

Install
npx -y skills add yclgkd/chinese-blog-review-skill --skill chinese-blog-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when reviewing or proofreading Chinese blog posts, Markdown drafts, technical notes, or long-form articles. Reports findings on title fit, topic focus, structure, technical accuracy, evidence, and Chinese copy formatting. NOT for English-only content, academic papers, marketing or brand copy, or auto-rewriting an article.

SKILL.md

4.1 KB, as published. Nobody here has run it

Chinese Blog Review

Overview

Use this skill as a Chinese blog editorial review guide. Treat article review like code review: report concrete issues with evidence and impact, preserve author voice, and only edit text when the user explicitly asks for changes.

Workflow

  1. Identify the article type: blog post, Markdown draft, technical note, essay, tutorial, or long-form documentation.
  2. Read references/review.md and give editorial feedback first.
  3. If the draft makes technical, factual, version-sensitive, or empirical claims (API behavior, protocol or standard, performance numbers, version availability, legal or compliance, pricing, second-hand 据说 / 听说 content), also load references/verifiability.md and apply its primary-source mapping, time-sensitivity tiers, four-state annotation, and AI-hallucination checklist.
  4. Read references/rules.md only for copy formatting and Chinese mixed-script typography.
  5. If the draft has English sentences, quotations, titles, labels, book / journal names, or punctuation-heavy Chinese-English mixing, also load references/mixed-punctuation.md.
  6. Preserve Markdown structure, inline code, code blocks, URLs, frontmatter keys, package names, file paths, identifiers, command flags, and official brand spelling.
  7. Treat formatting-only issues as lower priority than content correctness, topic fit, structure, and factual accuracy.
  8. When fact-checking is needed, fetch primary sources and distinguish verified facts, inferences, and items still pending verification using the four-state annotation in verifiability.md.
  9. If a project-specific style guide conflicts with this skill, follow the project-specific rule.

Output

Follow the structure and finding format defined in references/review.md (Output Format section). The rules below cover policy only, not formatting:

  • For review requests, return findings first. Do not rewrite the article unless asked.
  • For polishing requests, return the corrected text first and keep changes scoped.
  • For file edits, modify only the relevant text and avoid unrelated formatting churn.
  • For exact upstream wording or version-sensitive questions, fetch the current upstream README before answering.
  • Calibrate review length by value, not by forcing every dimension:
    • Report all blocking issues: confirmed factual errors, title/content mismatch, broken core argument, or materially misleading claims.
    • Keep the count of actionable findings (严重 / 中等) within a practical budget:
      • under ~1000 Chinese characters: usually 5 actionable findings, up to 8 when the article has 2+ blocking issues or title/topic problems that require separate fixes.
      • 1000-3000 characters: usually 8-12 actionable findings.
      • 3000-8000 characters: usually 12-18 actionable findings.
      • over 8000 characters: group findings by the article's original H2 sections when available; usually 18-25 actionable findings total unless the user asks for exhaustive review.
    • 轻微 findings do not count toward the actionable budget, but should be grouped or omitted unless they meaningfully affect readability.
    • 事实核查 flags do not count toward the actionable budget, but report only confirmed errors or claims that must be checked before publishing; group related verification concerns.
    • Repeated copy, punctuation, or wording issues should be reported once with examples, not item by item.
    • If over budget, drop or merge low-severity findings first; never hide blocking issues just because the budget is full.

Source

Copy-formatting guidance is a compact local adaptation of sparanoid/chinese-copywriting-guidelines, CY/T 154-2017 中文出版物夹用英文的编辑规范, and the Chinese technical writing guide referenced in references/mixed-punctuation.md. It is not a verbatim mirror.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.