icp-2

name	icp-2
version	2.1.0
description	ICP-2 — Human-First SEO & Answer Engine Optimization (AEO) standard for 2026
type	guardrail
enforcement	block
triggers	[{"file_pattern":"**/*.html"},{"file_pattern":"*.html"}]

ICP-2: Intelligent Content Protocol v2.1

Successor to: ICP-Lite v1.4 Status: Active standard (blocking enforcement) Purpose: Every page must be findable by humans, citable by AI, and honest about what it knows. Principle: Human first. No SEO theater. No AEO theater.

Soli Deo Gloria

---

What Changed from v2.0

v2.1 incorporates findings from multi-LLM research (GPT + Gemini consensus audit). Both models were asked: "What do you actually use vs. ignore?" Results:

1. Relaxed exact-match parity — JSON-LD description must be consistent with ai-summary, not character-identical. Exact-match enforcement was brittle process overhead, not AI value.
2. Demoted content-protocol tag — no crawler reads it. Kept for internal tooling only, removed from blocking validation.
3. Demoted llms.txt — not an established standard yet. Recommended, not required.
4. Removed ai-breadcrumbs HTML comments — both models confirmed: no crawler reads HTML comments. Remove them from pages.
5. Removed meta keywords — dead since 2009. Added to anti-theater forbidden list.
6. Removed geo.region/geo.placename — irrelevant for these sites.
7. Added static HTML requirement — key content must not be behind JS rendering.
8. Added volatile data discipline — prices, menus, policies must have as-of dates.
9. Added multi-surface framing — pages serve Google, AI assistants, and social previews simultaneously.

---

The ICP-2 Checklist (Blocking)

When any .html file is edited, all of the following must be true or the edit is blocked.

A. Required Head Meta Tags (3 blocking + 1 internal)

<meta name="ai-summary" content="..."/>
<meta name="last-reviewed" content="YYYY-MM-DD"/>
<meta name="description" content="..."/>

<!-- Internal tooling only — not read by any crawler -->
<meta name="content-protocol" content="ICP-2"/>

Rules:

• ai-summary: max 250 chars, first 155 chars must be a complete standalone answer
• ai-summary must be factual, specific, no hype, no CTAs
• last-reviewed: must be a real date when a human last verified the content
• description: traditional meta description, can differ from ai-summary but must be honest
• content-protocol: internal only — tells our validation which standard to check against. Not blocking if missing, but should be present for tooling.

B. Required JSON-LD Structured Data

Every page must have at minimum a JSON-LD block. Prefer specific types (Article, HowTo, TouristDestination) over generic WebPage — specific types give AI models much more to work with.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "name": "PAGE TITLE",
  "description": "Consistent with ai-summary (not necessarily identical)",
  "dateModified": "MUST MATCH last-reviewed EXACTLY",
  "datePublished": "YYYY-MM-DD"
}
</script>

Parity rules:

• JSON-LD description must be consistent with ai-summary — same facts, same meaning, but doesn't need to be character-identical. (Both models confirmed: exact-match adds maintenance burden without AI benefit.)
• JSON-LD dateModified must exactly match last-reviewed — conflicting dates actively confuse both humans and AI about content freshness. This is data integrity.

C. Content-Type Schema (Required for Entity Pages)

Pages about a specific thing must declare mainEntity with the appropriate type:

Page Type	Schema Type	Key Properties
Ship	Product	name, category, manufacturer
Port	TouristDestination	name, description, geo
Restaurant/Venue	FoodEstablishment	name, servesCuisine, priceRange
Article/Guide	Article	headline, author, datePublished
How-To/Planning	HowTo	name, step[], tool[]
Tool/Calculator	SoftwareApplication	name, applicationCategory
Person/Author	Person	name, jobTitle, knowsAbout
Hub/Index	CollectionPage	name, description

D. Answer-First Body Structure

1. First paragraph must answer: What is this? Why does it matter? What can I do here?
   • For complex topics that need context before the answer, lead with a brief orienting sentence, then answer.
2. Semantic headings: H2/H3 mapped to real user questions (naturally, not stuffed)
3. Internal linking: Every page links to 2+ related pages (ships link ports, ports link ships)
4. AI-extractable formatting: Use HTML tables for comparisons and data, ordered lists for steps/rankings, unordered lists for features/options. AI models extract structured HTML (tables, lists, heading hierarchies) more reliably than prose paragraphs. This isn't about formatting for crawlers — it's about making content scannable for humans AND parseable for AI simultaneously.
5. Semantic HTML5 elements: Use <article>, <section>, <nav>, <aside> to segment content. These help AI identify which part of the page contains the primary content vs. navigation vs. supplementary material.

E. Static HTML Content (New in v2.1)

Key content — the text a human reads, the facts an AI cites — must be in the static HTML. Do not hide content behind JavaScript rendering, client-side fetch, or dynamic injection. If a search engine or AI crawler can't see it without executing JS, it doesn't exist.

F. Canonical URL (Required)

<link rel="canonical" href="https://DOMAIN/exact-path.html"/>

G. OpenGraph + Social (Required for public-facing sites)

Only applies to repos with public websites (InTheWake, flickersofmajesty). Recipe repos, sheep repo, and sermon repo are exempt unless they have public pages.

<meta property="og:title" content="..."/>
<meta property="og:description" content="..."/>
<meta property="og:url" content="..."/>
<meta property="og:image" content="..."/>
<meta property="og:type" content="website"/>

---

AEO-Specific Requirements

1. Zero-Click Optimization

The ai-summary is the page's "answer" for AI citation. Write it as if an AI assistant will read it aloud as the complete answer to someone's question.

Test: "If an AI reads only this summary to a user, would they get an accurate, useful answer?"

Multi-surface reality: This summary serves Google snippets, AI Overviews, ChatGPT citations, Perplexity answers, and social share previews simultaneously. Write it for all of them — which means write it for humans.

How AI actually uses your content (March 2026): AI models increasingly synthesize from multiple sources rather than extracting verbatim snippets. This means your page may inform an AI answer without being quoted directly. The best strategy is the same one that works for humans: be the most accurate, specific, and well-structured source on the topic. Original data, unique statistics, and firsthand experience are what make your content indispensable — AI can't synthesize what doesn't exist elsewhere.

2. AI Crawler Access

robots.txt must explicitly allow AI crawlers:

User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

Do NOT block AI crawlers unless there is a specific, documented reason.

3. llms.txt (Recommended, Not Required)

If the site has a public presence, a /llms.txt at site root is recommended containing:

• Site overview (what it is, who it's for)
• Content scope and limitations
• Author credentials (E-E-A-T)
• Citation preference (how to attribute)
• Key content sections with entry points
• What the site does NOT do (no booking, no affiliate sales)

Note: Both GPT and Gemini confirmed llms.txt is not an established standard. It may be ignored. Include it as a low-cost bet, not a requirement.

4. Freshness Honesty

• Sitemap lastmod must reflect actual last content modification per page
• No uniform timestamps across all pages
• last-reviewed in HTML must match reality
• Stale content (>6 months without review) should be flagged for update

5. Volatile Data Discipline (New in v2.1)

When content includes prices, menus, operating hours, policies, or any fact that changes:

• As-of date must be visible (YYYY-MM-DD)
• Verification posture stated: "Verified in Cruise Planner" / "Observed onboard" / "Community-reported"
• Disclaimer: "Subject to change without notice"

This is not SEO theater — both AI models and humans need to know when time-sensitive data was last checked.

6. Multimedia Metadata (New in v2.1)

Images, video, and audio are increasingly processed by AI. Help them:

• Images: Descriptive alt text that states what's in the image factually (not "beautiful sunset photo" — instead "Ship wake at sunrise departing Port Canaveral"). Use WebP format, lazy loading for below-fold.
• Video: If embedded, include transcript text in static HTML. AI can't watch videos.
• Captions: Table/chart captions should state what the data shows, not just label it.

This isn't about stuffing keywords into alt text — it's about making visual content accessible to both screen readers and AI simultaneously.

7. E-E-A-T Author Authority

• Author pages must exist with Person schema
• knowsAbout must list specific, real expertise areas (not aspirational)
• sameAs should link to other presences only if they actually exist and are active — do not create social profiles just for SEO
• Content pages should link back to author where it makes sense for the reader (not forced into every page footer for crawlers)

---

What NOT To Do (Anti-Patterns)

These are theater. ICP-2 explicitly forbids them:

SEO Theater

1. Keyword stuffing — writing for algorithms instead of humans
2. Fake freshness — updating lastmod without actually reviewing content
3. Schema spam — adding structured data that doesn't match visible content
4. Meta keywords — dead since 2009, no engine reads them. Remove if present.
5. Link farming — artificial internal linking that doesn't help navigation
6. Forced FAQs — adding FAQPage schema to pages without genuine Q&A content
7. Geo meta tags — geo.region, geo.placename on non-local sites

AEO Theater

8. ai-breadcrumbs HTML comments — no AI crawler reads HTML comments. Both GPT and Gemini confirmed. Remove them.
9. AI-blocking — blocking AI crawlers out of fear rather than policy
10. Generic summaries — ai-summary that could apply to any page on the internet
11. Speculative standards — ai-plugin.json, .well-known/ai, or other standards that don't exist yet
12. Exact-match meta gymnastics — forcing character-identical text across multiple tags adds maintenance burden, not AI value. Be consistent, not identical.
13. Social profile farming — creating social accounts just to populate sameAs links
14. JS-hidden content — putting key content behind JavaScript rendering where crawlers can't reach it

---

ai-summary Writing Guide

Templates by Page Type

Ship: [Ship Name] is a [Class] ship ([GT] GT, [capacity] guests, [year]) featuring [2-3 signature features] and [itinerary type] itineraries.

Port: [Port Name] features [signature experience]. [Key fact]. [What cruisers do here].

Restaurant: [Name] is [line]'s [type] ([surcharge/included]) featuring [signature items] [availability].

Tool: [Tool name] helps cruisers [primary benefit]. [What it does in one sentence].

Article: [Topic] — [key insight or answer]. [Why it matters for cruise planning].

Hub: [Action verb] all [entity type] by [filters]. [Scope/count].

Rules

• Lead with the most important fact
• Use exact numbers (GT, capacity, year, price)
• No marketing language ("amazing", "best", "incredible")
• No CTAs ("book now", "learn more", "click here")
• No vague qualifiers ("many", "several", "various")
• First 155 chars must be a complete thought
• Total max 250 chars

---

Validation Procedure (Blocking)

When this skill fires on an HTML file edit, check in order:

1. ai-summary exists and is <= 250 chars
2. last-reviewed exists and is valid YYYY-MM-DD
3. description meta tag exists
4. JSON-LD block exists with @type
5. JSON-LD description is consistent with ai-summary (same facts/meaning)
6. JSON-LD dateModified matches last-reviewed exactly
7. Canonical URL exists
8. Key content is in static HTML (not behind JS rendering)
9. OpenGraph tags exist (og:title, og:description, og:url) — public sites only
10. First paragraph is answer-first (not "Welcome to..." or "This page...")
11. No forbidden anti-patterns present (meta keywords, ai-breadcrumbs comments)

If ANY check fails:

• Report the specific failure(s)
• Show what needs to be fixed
• Block the edit until fixed

If all checks pass:

• Confirm ICP-2 compliance
• Note any optional enhancements available (mainEntity, BreadcrumbList, FAQPage, llms.txt)

---

Migration from ICP-Lite v1.4

Pages with content-protocol="ICP-Lite v1.4" are allowed during migration. When editing such a page:

1. Update to ICP-2 compliance
2. Change protocol tag to ICP-2
3. Add any missing content-type schema
4. Verify freshness dates are accurate
5. Remove ai-breadcrumbs HTML comments if present
6. Remove meta name="keywords" if present
7. Remove meta name="geo.region" / geo.placename if present

---

Maintenance Schedule

• On every edit: ICP-2 validation fires (blocking)
• Monthly: Review hub pages for accuracy
• Quarterly: Review entity pages (ships, ports)
• On creation: New pages must be ICP-2 compliant from the start
• Annually: Review llms.txt and reassess emerging standards

---

End of ICP-2 v2.1 Protocol