SEO & GEO content flywheel — research, plan, draft, GEO/SEO + CORE-EEAT quality gate, publish, feedback. Zero-dependency Markdown skill for AI agents.
概览
Research a topic, plan it with the strongest reasoning model, draft it, de-AI it, run a GEO/SEO quality gate, fan it out to social, publish, and feed analytics back into the next cycle — as a reusable skill for Claude Code, Cursor, Codex CLI, Gemini CLI, and any agent that loads Markdown skills. Built for — content that AI search engines (ChatGPT, Perplexity, Google AI Overviews) actually cite — on top of classic SEO. A decides how autonomous each run is — from "pause for human review at every step" up to "fully automated" — and rises or falls based on how much you edit the output. The quality of the output comes from frameworks, not vibes. See frameworks/: - — definition optimization (term-first, 25–50 words, citable as a standalone answer), quotable statements (stats/facts/comparisons), authority signals, Q&A structure, and factual density so generative engines quote you.
README
以下为原文 README 节选，完整内容请查看原文。
Content Flywheel — SEO & GEO Content Skills
An open content-production methodology for AI agents. Research a topic, plan it with the strongest reasoning model, draft it, de-AI it, run a GEO/SEO quality gate, fan it out to social, publish, and feed analytics back into the next cycle — as a reusable skill for Claude Code, Cursor, Codex CLI, Gemini CLI, and any agent that loads Markdown skills.
Built for Generative Engine Optimization (GEO) — content that AI search engines (ChatGPT, Perplexity, Google AI Overviews) actually cite — on top of classic SEO.
🌐 Languages: English · 繁體中文 · 简体中文 · 日本語 · 한국어 · Español · Português · Français · Deutsch · Italiano · Русский · العربية · हिन्दी · Bahasa Indonesia · Tiếng Việt · ไทย · Türkçe · Nederlands
The Flywheel
┌─────────────────────────────────────────────────────────┐
│ ↓
Research → Plan → Draft → De-AI → GEO/SEO Quality Gate → Publish → Feedback
↑ │
└─────────────────── analytics tune the next cycle ───────────────┘
| Phase | What happens | Best-fit agent role |
|---|---|---|
| 1. Research | Trends, SERP intent, competitor gaps, audience needs | a research/grounding model |
| 2a. Plan | Outline (H2/H3), thesis, internal-link plan, CTA, structured-data plan | the strongest reasoning model |
| 2b. Draft | Full prose written to the plan | a fast drafting model |
| 2c. De-AI | Strip filler, generic phrasing, “AI tells” | a fast editing model |
| 2d. Quality gate | Score the draft against CORE-EEAT + GEO rules; list gaps | a review model |
| 3–4. Fan-out | Split to platform copy (FB/IG/Threads/X/LinkedIn) + carousels | a layout/copy model |
| 5. Technical SEO | Schema/JSON-LD, Core Web Vitals (INP/LCP/CLS) | a technical model |
| 6. Publish | Publish article + schedule social | automation |
| 7. Feedback | Analytics → tune content-pillar weights + cadence | the reasoning model |
A trust score decides how autonomous each run is — from “pause for human review at every step” up to “fully automated” — and rises or falls based on how much you edit the output.
GEO/SEO Frameworks
The quality of the output comes from frameworks, not vibes. See frameworks/:
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add zhenheco/content-flywheel-skills