メニュー
Use before publishing any article. Triggers: "check article", "preflight", "validate post", "is this ready to publish", "QA my post", "pre-publish check". Run this before distribute or any publish command.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when the user wants to set up or update their content profile. Also use when the user says "my blog is about", "content profile", "set up content context", or "who am I writing for". Stores context as .agents/content-context.md — read by all other content-ops skills first.
Use when the user wants content ideas or a backlog. Triggers: "content ideas", "what to write next", "build backlog", "ideas from changelog", "what should I post about", "content calendar".
Use when the user wants feedback on marketing copy, headlines, or landing page text. Triggers: "edit landing page", "improve headline", "copy feedback", "review this copy", "is this headline good", "landing page copy review".
Use when the user wants to publish an article to multiple platforms at once. Triggers: "publish article", "distribute to platforms", "send to devto and hashnode", "cross-post", "publish everywhere", "distribute my post".
Use when the user wants to write a welcome sequence or email drip campaign. Triggers: "welcome sequence", "newsletter drip", "onboarding emails", "email series", "welcome email". Note: generates drafts only — no sending, no API.
Use when the user wants to announce a product launch or new feature. Triggers: "HN Show post", "announce feature", "product launch post", "announce on HN", "write launch announcement", "post on HN", "Product Hunt post", "IndieHackers post", "GitHub launch".
| name | content-qa |
| description | Use before publishing any article. Triggers: "check article", "preflight", "validate post", "is this ready to publish", "QA my post", "pre-publish check". Run this before distribute or any publish command. |
Run pre-publish validation on the article. Check all platforms' requirements before sending.
Run content_qa.py for automated, deterministic checks before any LLM-driven analysis:
python3 ${CLAUDE_PLUGIN_ROOT}/tools/content_qa.py "$ARTICLE_PATH" --format text
Show the output. If exit code is 2 (block), stop immediately — hard failures must be fixed first.
If exit code is 1 (review), show the issues and continue to LLM checks below for deeper analysis.
If exit code is 0 (pass), still run the LLM checks below for editorial quality.
Read the article at the path the user provides. Run through references/platform-limits.md.
title — present and non-emptydescription — present, under 160 charsdate — present, valid YYYY-MM-DD formattags — present, list formatlang — present (en, ru, etc.)canonical_url — present (or can be derived from slug)## Content QA Report: <title>
### Frontmatter
OK title: "..."
OK description: "..." (143 chars)
WARNING tags: 6 tags — dev.to will truncate to 4
### Body
OK Has headings
OK Has CTA in last paragraph
WARNING Line 47: relative image path ./images/foo.png — convert to absolute URL
### Verdict
2 warnings — fix before publishing to dev.to
Fix any failures before publish. Warnings are platform-specific.