geo-fundamentals

name	geo-fundamentals
description	Generative Engine Optimization for AI search engines (ChatGPT, Claude, Perplexity).
allowed-tools	Read, Write, Edit, Glob, Grep
version	1.0.0
last-updated	"2026-03-12T00:00:00.000Z"
applies-to-model	gemini-2.5-pro, claude-3-7-sonnet

Generative Engine Optimization (GEO)

Traditional SEO optimizes for keyword ranking in blue links. GEO optimizes for citation and inclusion in AI-generated answers.

---

What GEO Is

When users ask ChatGPT, Claude, Perplexity, or Google SGE a question, an AI synthesizes an answer from sources it considers authoritative. GEO is the practice of making your content the kind that AI systems cite, quote, and summarize.

The mechanisms differ from traditional SEO:

• AI doesn't rank URLs — it synthesizes information
• Backlinks matter less than citation-worthiness
• Content density and factual specificity matter more than keyword density

---

How AI Systems Select Content to Cite

Based on observable patterns in AI retrieval:

Signal	What It Means
Authoritative domain	.gov, .edu, established publications get higher base trust
Factual specificity	Numbers, dates, named sources > vague descriptions
Structured content	Lists, tables, step-by-step = easier to extract
Complete answers	Content that fully addresses the question in one place
Original research	Data, studies, surveys the AI can't find elsewhere
Freshness	Recently updated content has advantage for time-sensitive topics

---

Writing for AI Consumption

Answer First (Inverted Pyramid)

AI models extract answers. Put the answer at the top of the content, not after a 300-word preamble.

❌ Old blog pattern:
   [intro paragraph explaining what the article is about]
   [background context]
   [finally... the answer in paragraph 5]

✅ GEO pattern:
   [Direct answer in first 1–2 sentences]
   [Supporting evidence, data, explanation]
   [Deeper context for readers who want more]

Use Citation-Ready Structures

# Format that AI can easily extract:

## Definition
[Term] is [concise definition]. [Supporting context].

## Key Facts
- [Specific, numbered fact with source]
- [Specific statistic — "X% of Y according to Z study"]
- [Named, verifiable claim]

## Step-by-Step Process
1. [Precise step]
2. [Precise step]
3. [Precise step]

Be Specific with Data

❌ "Many developers prefer TypeScript"
✅ "In the 2024 Stack Overflow Developer Survey, 64% of respondents said they used TypeScript"

❌ "Deployment takes some time"  
✅ "Vercel cold-start latency for serverless functions averages 200–400ms for Node.js 20 runtimes"

---

Content Formats AI Favors

Format	GEO Value	Why
FAQ pages	High	Matches question-answer format of AI responses
Comparison tables	High	Easily extracted for comparison queries
How-to guides with numbered steps	High	Directly answerable procedural questions
Definition/explanation articles	High	Definitional queries are common AI use cases
Long-form opinion pieces	Low	Hard to extract a clear answer from
News articles	Medium	Good for recency, lower for evergreen queries

---

Technical Requirements for AI Indexing

<!-- Ensure Perplexity and other AI crawlers can access your content -->
<!-- Check robots.txt — don't accidentally block AI crawlers -->

# robots.txt — allow AI crawlers
User-agent: PerplexityBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: ClaudeBot
Allow: /

<!-- Structured data helps AI understand content context -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [{
    "@type": "Question",
    "name": "What is tribunal-kit?",
    "acceptedAnswer": {
      "@type": "Answer",
      "text": "tribunal-kit is an npm package that installs an anti-hallucination agent kit..."
    }
  }]
}
</script>

---

Metrics for GEO

Traditional SEO metrics (keyword rank, backlinks) don't fully apply. Track:

• AI mention rate — manually query AI systems for your topic and check if your brand/content appears
• Citation count — if identifiable quotes from your content appear in AI outputs
• Perplexity source appearances — Perplexity shows its sources; track mentions
• Direct traffic — users who find AI mentions and navigate directly (not via search)

---

Scripts

Script	Purpose	Run With
scripts/geo_checker.py	Audits content for GEO best practices	python scripts/geo_checker.py <url>

---

Output Format

When this skill produces a recommendation or design decision, structure your output as:

━━━ Geo Fundamentals Recommendation ━━━━━━━━━━━━━━━━
Decision:    [what was chosen / proposed]
Rationale:   [why — one concise line]
Trade-offs:  [what is consciously accepted]
Next action: [concrete next step for the user]
─────────────────────────────────────────────────
Pre-Flight:  ✅ All checks passed
             or ❌ [blocking item that must be resolved first]

---

🤖 LLM-Specific Traps

AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:

1. Over-engineering: Proposing complex abstractions or distributed systems when a simpler approach suffices.
2. Hallucinated Libraries/Methods: Using non-existent methods or packages. Always // VERIFY or check package.json / requirements.txt.
3. Skipping Edge Cases: Writing the "happy path" and ignoring error handling, timeouts, or data validation.
4. Context Amnesia: Forgetting the user's constraints and offering generic advice instead of tailored solutions.
5. Silent Degradation: Catching and suppressing errors without logging or re-raising.

---

🏛️ Tribunal Integration (Anti-Hallucination)

Slash command: /review or /tribunal-full Active reviewers: logic-reviewer · security-auditor

❌ Forbidden AI Tropes

1. Blind Assumptions: Never make an assumption without documenting it clearly with // VERIFY: [reason].
2. Silent Degradation: Catching and suppressing errors without logging or handling.
3. Context Amnesia: Forgetting the user's constraints and offering generic advice instead of tailored solutions.

✅ Pre-Flight Self-Audit

Review these questions before confirming output:

✅ Did I rely ONLY on real, verified tools and methods?
✅ Is this solution appropriately scoped to the user's constraints?
✅ Did I handle potential failure modes and edge cases?
✅ Have I avoided generic boilerplate that doesn't add value?

🛑 Verification-Before-Completion (VBC) Protocol

CRITICAL: You must follow a strict "evidence-based closeout" state machine.

• ❌ Forbidden: Declaring a task complete because the output "looks correct."
• ✅ Required: You are explicitly forbidden from finalizing any task without providing concrete evidence (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.