メニュー
Preflight review of blog posts before publication. Use when reviewing or doing a final check on a blog post (.mdx), or when asked to "review this post" or "preflight check".
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
Write or edit blog posts in Katie Schilling's voice. Use when drafting case studies, product feature posts, or partner-focused content for the Tigris blog. Katie writes technically grounded prose that centers the customer's problem and lets specifics do the persuading. Also use when reviewing prose that should match Katie's style.
Rewrite prose so it reads like one real person actually said it. Use when text feels constructed, overcorrected, evenly paced, or generically "well-written" in the way AI output tends to be. Strips rehearsed phrasing, breaks rhythmic patterns, and restores a clear point of view.
Audit a piece of prose for AI-generated tells. Use when checking whether text reads as AI-written, when reviewing AI-assisted drafts before publication, or when triaging third-party content. Triggers on "is this AI?", "does this sound like AI?", "check for AI patterns", "audit for slop", "scan for tells". Output is a structured forensic report — quotes, locations, severity, confidence. This skill detects; `stop-slop` rewrites.
Audit a blog post for engagement and reader pull. Use when checking whether a draft will actually hold attention — separate from correctness, AI tells, or SEO. Triggers on "is this interesting", "make this more engaging", "will anyone read this", "why does this feel flat", or any review where pulling the reader through is the question.
Create outlines for "build with me" blog posts -- dramatic retellings of engineering decisions, not tutorials. Use when outlining a "deep dive", "how we built", or "walkthrough" style post.
Use when optimizing images for web performance by converting PNG, JPG, or other formats to WebP using ffmpeg. Use when blog posts, documentation sites, or web projects need smaller image files with equivalent quality.
| name | review-writing |
| description | Preflight review of blog posts before publication. Use when reviewing or doing a final check on a blog post (.mdx), or when asked to "review this post" or "preflight check". |
blog/**/index.mdx files via
git status.Apply each of the following skills to the post, noting these overrides:
xe-writing-style — Do NOT use Xe's character dialogues (Mara, Cadey, etc.);
this blog engine doesn't support them.writing-clearly-and-concisely — Focus on substantive edits, not stylistic
nits.stop-slop — Flag but don't rewrite entire paragraphs; suggest specific
phrase replacements.using-inline-cta — Check that an <InlineCta> component exists; suggest one
if missing.seo-aeo-best-practices — Focus on frontmatter and structure, not in-body
keyword density.detect-ai-writing - Check that an article doesn't display obvious signs of
AI writing.hold-the-reader - Audit the blogpost for engagement and reader pull.Favor specific, opinionated writing with concrete examples over generic claims.
Run npm run seo:check against the target post and incorporate its output into
the findings. Refer to README-seo-reviewer.md for interpreting the output
categories.
xe-writing-style (voice and tone trump SEO polish)stop-slop (remove AI patterns)writing-clearly-and-concisely (tighten prose)seo-aeo-best-practices (optimize metadata and structure)using-inline-cta (add CTA if missing)Deduplicate findings across skills. If multiple skills flag the same span of text, combine them into a single finding and note which skills flagged it.
Start the report with:
Advisory only — verify suggestions against author intent. Ask in #gtm if unsure.
If the post has no high-severity issues and fewer than 3 medium issues, lead with: "This post is ready for publication" before listing remaining nits.
path:line.webp filenames