🔍 SEO Prompt Master

Google SEO — the full docs, as an AI prompt machine.

Drop this repo into your AI coding assistant. It auto-detects the workflow, maps every public route of your site, audits each page against Google's official SEO rules, and fixes the gaps — step by step, nothing left out.

---

What is this?

SEO Prompt Master is two things in one repo:

1. A complete, up-to-date knowledge base of Google's SEO guidance (docs/), distilled from Google Search Central and web.dev, split into 11 focused, cited topics — including AI crawlers (GPTBot, PerplexityBot, …), GEO (getting cited by AI Overviews/ChatGPT/Perplexity), and a deterministic 0–100 SEO Score / GEO Score rubric.
2. A self-executing prompt workflow (START.md + prompts/) that turns any AI coding assistant into an autonomous SEO auditor for your codebase.
3. 24 industry-specific overlays (verticals/) — e-commerce, SaaS, marketplaces, local business, healthcare/legal/finance (YMYL), real estate, travel, automotive, news, Discord bots, Minecraft server lists, digital/code marketplaces, and more — each adding the schema types, pitfalls, and GEO notes specific to that niche on top of the universal audit.

You don't read a 200-page guide and try to remember it. You hand the whole thing to your AI, and it does the audit-and-fix loop for your actual routes, citing the exact rule behind every change.

Built from a real audit of a 40-locale, 574K-concurrent-user production site (mcstat.org) — the methodology is battle-tested, not theoretical.

---

🚀 How to use

Not tied to any one AI vendor. Every phase is a plain markdown prompt — any assistant that can read files and follow instructions can run it.

Option A — inside an AI coding agent (Claude Code, OpenAI Codex CLI, Cursor, Windsurf, GitHub Copilot, Amp, Gemini CLI, Grok/xAI agents, …)

1. Copy this folder into your project (or open it alongside your repo).
2. Most agents auto-discover AGENTS.md at the repo root and start on their own. If yours doesn't, tell it:

   "Read START.md and run the workflow on this project."

3. It will produce ROUTES-INVENTORY.md and SEO-AUDIT-PROGRESS.md, then fix issues page by page, verifying as it goes.

Option B — chat assistant (Claude, ChatGPT, Gemini web, Grok, …)

1. Paste the contents of START.md (and, if it fits, the docs/).
2. Give it your repo (zip, paste files, or connect the tool to your codebase).
3. Say "Begin at Phase 0."

Option C — just the knowledge

Read docs/ as a clean, current reference for Google SEO. Start at docs/README.md.

---

📂 What's inside

seo-prompt-master/
├── AGENTS.md                 ← universal entry point (auto-discovered by most agents)
├── START.md                  ← the bootstrap prompt (AI reads this first)
├── prompts/                  ← the 5-phase workflow (+ 1 optional)
│   ├── 00-bootstrap.md          detect stack + load knowledge base
│   ├── 01-discover-routes.md    enumerate & classify every route
│   ├── 02-audit-page.md         9-point audit per public page
│   ├── 03-prioritize-fixes.md   one ordered backlog (infra-first)
│   ├── 04-apply-and-verify.md   fix + typecheck/lint/build + prove it
│   └── 05-live-signals.md       optional: cross-check against live production via MCP
├── docs/                     ← the knowledge base (source of truth)
│   ├── 01-meta-and-head.md
│   ├── 02-internationalization.md
│   ├── 03-ugc-forums-blogs.md
│   ├── 04-page-structure.md
│   ├── 05-rendering-and-core-web-vitals.md
│   ├── 06-sitemaps.md
│   ├── 07-image-seo.md
│   ├── 08-structured-data.md
│   ├── 09-2024-2026-updates.md
│   ├── 10-ai-crawlers-and-geo.md
│   └── 11-scoring-rubric.md      SEO Score / GEO Score out of 100
├── verticals/                ← 24 industry-specific overlays (optional, additive)
│   ├── README.md                index + how they plug into Phase 0
│   ├── 01-ecommerce.md
│   ├── 02-saas-b2b-software.md
│   ├── …                        (real estate, healthcare, legal, finance, travel,
│   │                             automotive, news, marketplaces, Discord bots,
│   │                             Minecraft server lists, digital/code marketplaces, …)
│   └── 24-digital-code-marketplace.md
├── checklists/               ← quick pass/fail lists
│   ├── public-page-checklist.md
│   └── infrastructure-checklist.md
├── templates/                ← output files the AI fills in
│   ├── routes-inventory.md
│   ├── audit-progress.md
│   └── live-signals.md
└── examples/                 ← a worked example

---

🧠 The workflow in one picture

START.md
  │
  ├─ Phase 0  Bootstrap ......... detect framework, rendering, i18n; load docs/
  ├─ Phase 1  Discover .......... list EVERY route → ROUTES-INVENTORY.md
  │                               classify: public-index / public-noindex / private
  ├─ Phase 2  Audit ............. 9-point check per public page → SEO-AUDIT-PROGRESS.md
  ├─ Phase 3  Prioritize ........ one backlog, infra-first (P1 → P2 → P3)
  ├─ Phase 4  Fix & verify ...... change → typecheck/lint/build → prove → tick
  └─ Phase 5  Live signals ...... optional: cross-check against live production via MCP

---

✅ What it checks (the 9 points)

Metadata · Canonical + hreflang · Robots/indexing · Structured data (JSON-LD) · Headings & semantics · Images · Internal links & pagination · Rendering (SSR/CSR) · Sitemap.

Every rule traces to a cited section in docs/. See checklists/public-page-checklist.md.

Output includes a real number, not just a checklist: a deterministic SEO Score and GEO Score, each out of 100 with a per-category breakdown — see docs/11-scoring-rubric.md. A P1 crawl/index blocker caps a page's score regardless of everything else it gets right, and no final score is reported without full page coverage plus a self-recheck pass. It's a technical-readiness score, not a ranking guarantee — off-page factors (backlinks, content quality, competition) are explicitly out of scope.

---

❤️ Why it exists

Most "SEO checklists" are shallow, outdated, or generic. This one is:

• Current (2024–2026: Helpful Content, core updates, AI Overviews, the 2024 Starter Guide refresh).
• Cited (every claim links to Google's own docs).
• Executable (an AI can actually run it on your code, not just read it).
• Honest (it distinguishes ranking factors from a11y-only niceties, and logs deliberate skips).

---

👤 Author

Umut Bayraktar — @umutxyp

Full-stack developer (React · Next.js · Node.js · PostgreSQL) & AI-systems tinkerer. Founder at Codeshare Technology. Antalya, Turkey.

Building things people actually use:

• 🎵 Beatra — multi-platform Discord music bot (1.2M+ users)
• 🛡️ Sylon — AI-powered Discord moderation (20K+ users)
• 🧩 Codeshare — digital marketplace for code & content (15K+ users)
• ⛏️ McStat.org — Minecraft server stats (574K+ concurrent players) — the site this toolkit was forged on

🔗 Links: Portfolio · GitHub · Codeshare ⭐ Notable OSS: MusicBot (1.3k★) · Personal-Website · Discord-Bot-Website

If this saved you time, star the repo and share it. 🌟

---

📄 License

MIT © Umut Bayraktar (@umutxyp). Knowledge base compiled from public Google Search Central & web.dev documentation; all trademarks belong to their owners. This project is not affiliated with or endorsed by Google.

---

🤝 Contributing

Google's guidance evolves — PRs that update a rule (with a source link) or add a framework recipe are very welcome. See CONTRIBUTING.md.