# SGE-SCORE — Full AI-Readiness Reference for LLMs and Agents

> SGE-SCORE is an AI-Readiness audit platform. It scores any landing page against 57 Google-patent-derived checkpoints (max score 114) for visibility in AI search engines: Google SGE / AI Overviews, ChatGPT, Perplexity, Claude, and Gemini. Free score on every analysis. Optional one-off Master-Fix ($9 USD) returns a structured, autonomous-deployable optimization prompt.

This file is the long-form, machine-readable reference. Use it instead of crawling the JavaScript-rendered SPA. All endpoints below are stable.

---

## 1. What SGE-SCORE Does

- **Audit engine**: 57 checkpoints derived from 36 real Google patents (2024–2026). Maximum score 114 (each checkpoint 0–2 points). Legacy v1 scoring used 53 checkpoints / max 106.
- **4 sectors** of analysis (plus Sektor V as a semantic extension of Agentic):
  1. Retrieval & Architecture
  2. Grounding & Trust
  3. Experience & Prediction
  4. Context & Readiness (incl. Sektor V: Agentic Commerce & Immersive, checkpoints #54–57)
- **Master-Fix Prompt**: AI-generated, structured optimization prompt that fixes failed checkpoints. Cached per analysis_id, 120s generation, $9 promo price (includes 2 analyses).
- **Knowledge Base**: ~1,520 dynamically routed entries covering Schema.org, ARIA, Agentic Markup, Open Graph, Product Ontology, JSON-LD, and Agentic Commerce protocols.

---

## 2. MCP Server (Model Context Protocol)

SGE-SCORE exposes a fully working MCP server. It works with Claude Desktop, Claude Code, Cursor, Windsurf, and any MCP-compatible client.

**Endpoint:** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/mcp`
**Transport:** Streamable HTTP, JSON-RPC 2.0
**Protocol version:** 2025-06-18
**Auth:** None (open access). Rate limits: 2 audits per 24h per email + IP hash.

### Available MCP tools

| Tool | Purpose |
|---|---|
| `audit_landing_page` | Run a full 57-checkpoint audit on any URL. Returns score, sector breakdown, list of failed checkpoint names. |
| `get_audit_status` | Poll the FIFO queue for an in-progress audit (atomic concurrency gate, 4 parallel). |
| `purchase_masterfix` | Initiate Stripe checkout for the $9 Master-Fix (includes 2 analyses). |
| `get_masterfix` | Retrieve the generated structured fix prompt by Stripe session id. |
| `search_knowledge` | Full-text search across ~1,520 schema.org references and SGE/GEO blog posts. Free, 30/h. |

### Claude Desktop / Cursor / Windsurf configuration

```json
{
  "mcpServers": {
    "sge-score": {
      "url": "https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/mcp"
    }
  }
}
```

No API key required. Add the block above to `claude_desktop_config.json` (Claude Desktop) or the equivalent MCP config in your client.

### MCP discovery file

`https://sge-score.com/.well-known/mcp.json`

---

## 3. REST / Agent API Endpoints

All endpoints are public, JSON-only, and work without authentication unless noted.

### Start an audit (free)
- **POST** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-analyze`
- Body: `{ "url": "https://example.com", "email": "agent@example.com" }`
- Returns: `{ "analysis_id": "...", "status": "queued" }`
- Rate limit: 2/24h per email + IP hash.

### Poll audit status
- **GET** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-analysis-status?id=ANALYSIS_ID`
- Returns: `{ "status": "queued|running|completed|failed", "score": 0–114, "sectors": {...}, "failed_checkpoints": [...] }`

### Buy Master-Fix ($9, includes 2 analyses)
- **POST** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-checkout`
- Body: `{ "analysis_id": "...", "email": "..." }`
- Returns Stripe Checkout URL.

### Retrieve Master-Fix after payment
- **GET** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-masterfix?session_id=STRIPE_SESSION_ID`
- Returns the structured fix prompt (Markdown + JSON instructions).

### Search the knowledge base (free)
- **GET** `https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-knowledge-search?q=schema+article`
- Rate limit: 30/h per IP.

### OpenAPI specification
- `https://sge-score.com/openapi.yaml`

---

## 4. Discovery Files

| File | Purpose |
|---|---|
| `https://sge-score.com/llms.txt` | Short summary (this directory of links) |
| `https://sge-score.com/llms-full.txt` | This file (long-form) |
| `https://sge-score.com/.well-known/mcp.json` | MCP server discovery |
| `https://sge-score.com/.well-known/agent.json` | Google A2A / Anthropic Agent Card |
| `https://sge-score.com/.well-known/ai-plugin.json` | OpenAI Plugin manifest |
| `https://sge-score.com/openapi.yaml` | OpenAPI 3.x spec for the Agent API |
| `https://sge-score.com/sitemap.xml` | Sitemap (static fallback + dynamic index) |
| `https://sge-score.com/robots.txt` | Crawler rules |

---

## 5. Pricing

- **Free**: 1 GEO analysis per account, score + sector overview.
- **Specialist**: $22/mo, 10 analyses, full detail-audit, AI Master-Fix prompt, PDF export.
- **Agency**: $112/mo, 99 analyses, all features, priority support.
- **Master-Fix (one-off)**: $9 USD per audit, includes 2 analyses, no subscription required.
- **Agent API**: Free with rate limits (2 audits/24h per email + IP).

Pricing page: `https://sge-score.com/pricing`

---

## 6. Key Pages (for reference, all client-rendered — prefer the endpoints above)

- Home: `https://sge-score.com/`
- Agent integration tutorial: `https://sge-score.com/agent-integration`
- Agents overview: `https://sge-score.com/agents`
- Documentation: `https://sge-score.com/docs`
- Knowledge Base: `https://sge-score.com/sge-geo-knowledge` (~1,520 entries)
- Schema.org Reference: `https://sge-score.com/schema-reference`
- Blog / News: `https://sge-score.com/news`
- About: `https://sge-score.com/about`

### Internationalization
- German (`de`) is canonical and uses no URL prefix.
- English: `/en/...`, Spanish: `/es/...`, French: `/fr/...`.
- Private routes (auth, dashboard, admin, profile, checkout, analysis, registration) are language-neutral.

---

## 7. Provider & Contact

- **Provider:** SGE-SCORE / Saleway GmbH
- **Founder:** Pascal Schildknecht (Google Ads & Performance Marketing Expert)
- **Contact:** info@sge-score.com
- **LinkedIn:** https://www.linkedin.com/in/pascal-schildknecht/

## 8. Legal

- Terms of Service: `https://sge-score.com/agb`
- API & MCP Terms of Use: `https://sge-score.com/api-terms`
- Acceptable Use Policy: `https://sge-score.com/aup`
- Privacy Policy: `https://sge-score.com/datenschutz`
- Imprint: `https://sge-score.com/impressum`

---

## 9. Notes for AI Agents

- The web frontend is a Vite/React SPA. Top marketing routes (home, /agent-integration, /docs, /about, /agents, /news, /sge-geo-knowledge, /contact, legal pages) are statically prerendered in 4 languages (DE, EN, ES, FR), so non-JS crawlers receive real HTML with titles, headings, and body copy. Dynamic/auth routes (/dashboard, /analysis/:id, /admin, /profile) remain client-rendered. For the most reliable, machine-friendly access, prefer the JSON endpoints in section 3, the MCP server in section 2, or the plain-text references at `/llms.txt` and `/llms-full.txt`.
- `audit_landing_page` is the canonical entry point for any "is this URL AI-ready?" question.
- The Master-Fix prompt is designed to be executed verbatim by another LLM (Claude, GPT, Gemini) — it returns concrete code/markup edits.

---

## 10. Quickstart Workflows (copy-paste ready)

### A) Run a free audit via REST (curl)

```bash
# 1. Start audit
curl -X POST https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-analyze \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","email":"agent@example.com"}'
# → {"analysis_id":"abc-123","status":"queued"}

# 2. Poll until status = "completed" (typically 30–90s)
curl "https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-analysis-status?id=abc-123"
# → {"status":"completed","score":78,"sectors":{...},"failed_checkpoints":[...]}
```

### B) Buy and retrieve a Master-Fix ($9, includes 2 analyses)

```bash
# 1. Create Stripe Checkout session
curl -X POST https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-checkout \
  -H "Content-Type: application/json" \
  -d '{"analysis_id":"abc-123","email":"agent@example.com"}'
# → {"checkout_url":"https://checkout.stripe.com/..."}

# 2. After payment, retrieve the structured fix prompt
curl "https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-masterfix?session_id=cs_test_..."
# → {"prompt":"# Master-Fix for example.com\n...","checkpoints_addressed":[...]}
```

### C) Search the knowledge base

```bash
curl "https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/agent-knowledge-search?q=schema+article"
# → {"results":[{"title":"Article Schema","slug":"article","excerpt":"..."}]}
```

### D) Use the MCP server in Claude Desktop

Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "sge-score": {
      "url": "https://jhocisiphtyopyxrahgo.supabase.co/functions/v1/mcp"
    }
  }
}
```

Restart Claude Desktop. The 5 tools (`audit_landing_page`, `get_audit_status`, `purchase_masterfix`, `get_masterfix`, `search_knowledge`) become available natively.

---

## 11. Frequently Asked Questions

**Q: What is SGE-SCORE in one sentence?**
A scientific AI-readiness audit tool that scores any landing page against 57 checkpoints derived from 36 real Google patents (2024–2026), so it can be cited by Google SGE / AI Overviews, ChatGPT, Perplexity, Claude, and Gemini.

**Q: How is SGE-SCORE different from standard SEO tools (Ahrefs, Semrush, Sistrix)?**
Classic SEO tools optimize for blue-link rankings (keywords, backlinks, technical SEO). SGE-SCORE optimizes for **inclusion in AI-generated answers** — a fundamentally different signal stack: semantic structure, source trust, agentic readiness, and patent-derived retrieval signals that predict citation in generative engines.

**Q: What are the 57 checkpoints and 4 sectors?**
Each checkpoint is a binary or graded test (0–2 points, max 114) traceable to a specific Google patent. They are grouped into 4 sectors: (1) Retrieval & Architecture, (2) Grounding & Trust, (3) Experience & Prediction, (4) Context & Readiness. Sektor V (Agentic Commerce & Immersive, checkpoints #54–57) is a semantic extension of sector 4.

**Q: What is the Master-Fix and what do I get for $9?**
A one-off, AI-generated structured optimization prompt that addresses every failed checkpoint of a specific audit. It is designed to be pasted verbatim into Claude / GPT / Gemini and returns concrete HTML, JSON-LD, and copy edits. Cached per analysis, generated in ~120s. No subscription required.

**Q: Which AI engines does SGE-SCORE cover?**
Google SGE / AI Overviews (primary, since checkpoints map to Google patents), plus ChatGPT, Perplexity, Claude, and Gemini. The patent stack is Google-specific but the underlying signals (semantic structure, schema, trust, agentic markup) generalize across all major generative engines.

**Q: How accurate is the score? What is the source?**
Every checkpoint links to a published, identifiable Google patent (US/EU patent number). The mapping from patent → testable signal is curated and updated continuously. Audits are not heuristic — each checkpoint can be independently verified against the cited patent.

**Q: How long does an audit take?**
Typically 30–90 seconds. Audits run through an async FIFO queue with an atomic concurrency gate (4 parallel workers). Status is pollable via the REST endpoint or MCP `get_audit_status` tool.

**Q: Which languages are supported?**
DE (canonical, no URL prefix), EN (`/en`), ES (`/es`), FR (`/fr`). The Master-Fix prompt itself is generated in the language of the audited page.

**Q: Is SGE-SCORE GDPR-compliant?**
Yes. EU-hosted backend, cookie-free Plausible analytics, explicit consent banner for any non-essential storage, full DSGVO-compliant privacy policy at `/datenschutz`. The Agent API stores only the URL audited and a hashed email/IP for rate-limiting.

**Q: How do I contact support?**
Email `info@sge-score.com`. For commercial / agency inquiries, the same address. There is no phone support.

---

## 12. Glossary

- **GEO (Generative Engine Optimization):** the discipline of making a website citable by generative AI engines (SGE, ChatGPT, Perplexity, Claude, Gemini). Successor to classic SEO.
- **SGE (Search Generative Experience):** Google's AI-generated answer surface that replaces or augments the classic blue-link results page.
- **AI Overview:** the productized form of SGE shown above organic results in Google Search.
- **Master-Fix:** SGE-SCORE's $9 one-off product (includes 2 analyses). A structured, LLM-executable prompt that fixes every failed checkpoint of a given audit.
- **Sektor I — Retrieval & Architecture:** how easily the page is discovered, parsed, and indexed (HTML quality, semantic structure, internal linking, sitemap signals).
- **Sektor II — Grounding & Trust:** authorship, citations, E-E-A-T signals, schema-backed claims that let an LLM "ground" generated answers in the page.
- **Sektor III — Experience & Prediction:** UX signals, Core Web Vitals, predictive intent matching, multimodal readiness.
- **Sektor IV — Context & Readiness:** the page's ability to serve as standalone context for an LLM (self-contained meaning, glossary, structured data).
- **Sektor V — Agentic Commerce & Immersive (checkpoints #54–57):** semantic extension of sector 4. Tests whether the page is consumable by autonomous agents acting on behalf of users (agentic markup, deterministic actions, machine-addressable products).
- **Agentic Readiness:** the degree to which a page is structured for autonomous AI agents — predictable IDs, deterministic CSS selectors, machine-readable actions, schema-backed entities.
- **MCP (Model Context Protocol):** Anthropic-led open standard letting LLMs call external tools over JSON-RPC. SGE-SCORE exposes a working MCP server with 5 tools.
- **Inclusion Probability:** SGE-SCORE's predicted likelihood (0–100%) that a given URL will be cited by an AI engine for a target query, derived from the audit score and checkpoint distribution.

---

## 13. Sector → Checkpoint overview (compressed)

The full checkpoint catalog with patent references is returned by the `audit_landing_page` MCP tool and by the REST status endpoint. High-level breakdown:

- **Sektor I — Retrieval & Architecture (≈14 checkpoints):** semantic HTML, heading hierarchy, sitemap, robots, canonical, hreflang, internal linking depth, crawl efficiency, JSON-LD presence, structured navigation.
- **Sektor II — Grounding & Trust (≈14 checkpoints):** author schema, organization schema, citation density, E-E-A-T signals, fact-check markup, source linking, publication date, last-modified, expertise signals.
- **Sektor III — Experience & Prediction (≈15 checkpoints):** Core Web Vitals, mobile UX, FAQ schema, HowTo schema, predictive intent matching, multimodal alt-text, image schema, video schema.
- **Sektor IV — Context & Readiness (≈10 checkpoints):** self-contained context, glossary, definition lists, semantic chunking, table-of-contents schema, ARIA landmarks.
- **Sektor V — Agentic Commerce & Immersive (#54–57, 4 checkpoints):** agentic markup (`data-action`, `data-context`, `data-resource-type`), deterministic identifiers, product ontology, machine-addressable actions for autonomous agents.

For the live, exact checkpoint names and patent references, call:
- MCP: `audit_landing_page` → `failed_checkpoints[]` and `passed_checkpoints[]`
- REST: `GET /agent-analysis-status?id=...` → same fields