Readiness Levels

The baseline before any deliberate MX work. The site may have auto-generated meta tags from a CMS, boilerplate social sharing cards, or templated HTML. None of this was placed with deliberate machine readability in mind. An AI agent visiting this site must infer meaning from visual layout, guess at business identity, and hallucinate missing context.

Characteristics

• Auto-generated <title> from CMS or template
• No Schema.org JSON-LD
• No llms.txt
• No semantic HTML structure beyond CMS defaults
• Meta description either missing or auto-generated
• No MX governance metadata

What It Proves

Nothing. This is the absence of deliberate MX. Most websites on the internet are at Level 0.

Agent Outcome

Agents must guess, infer, and hallucinate. They cannot reliably identify the publisher, extract structured facts, or verify any claims. Recommendations based on Level 0 content are unreliable.

Typical State

Any website that has not been deliberately structured for machine consumption. CMS-generated pages with default templates, marketing sites built for visual impact without metadata consideration, legacy sites with outdated HTML.

The entry point. MX metadata is present and the publisher is identified. This is the minimum structure that makes content machine-parseable rather than just human-readable. Any file type can qualify — markdown, HTML, JavaScript, CSS, or shell scripts — each using its own carrier format.

Requirements

• MX metadata present in the appropriate carrier format
• title (or equivalent) and description fields present
• author field identifying the publisher
• Structured data parseable by machines without inference

What It Proves

Someone has deliberately structured this content for machine consumption. It is not a random file — it has identity.

Agent Outcome

Agents can parse and discover this content. They can find it, extract basic facts, and confirm the publisher's identity. Discovery-stage agent tasks succeed.

Typical Use

Internal wikis, team documentation, product pages, personal knowledge bases. Content that needs to be parseable by internal AI tools but does not require external trust verification.

---
title: "Internal API Guide"
description: "Authentication flows for the payments service"
author: "Engineering Team"
version: "1.0"
---

# Internal API Guide

Content here...

<!-- Schema.org structured data -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "TechArticle",
  "name": "Internal API Guide",
  "description": "Authentication flows for the payments service",
  "author": { "@type": "Organization", "name": "Engineering Team" }
}
</script>

<!-- MX governance metadata -->
<meta name="mx:status" content="active">
<meta name="mx:contentType" content="guide">

The document follows the full MX metadata specification. It has a named maintainer, contact details, and the operational metadata that tells AI agents not just what the content is, but who is responsible for it.

Requirements

• Everything from Level 1
• Full MX metadata in the appropriate carrier format
• mx:status, mx:contentType, mx:tags fields
• Named maintainer with contact information
• created and modified dates (or equivalent)

What It Proves

This content has governance. A named person or team is responsible for its accuracy. An AI agent can assess not just the content, but who stands behind it.

Agent Outcome

Agents can cite and attribute this content. They can reference it as a source, link to the named maintainer, and assess when it was last updated. Citation-stage agent tasks succeed.

Typical Use

Public-facing documentation, product pages, API references. Content that external parties will read and rely upon, but where cryptographic attestation is not yet required.

---
title: "Product Specification"
description: "Rancilio Silvia Pro X technical spec"
author: "Rancilio Group"
created: 2026-01-15
modified: 2026-03-01
version: "2.1"

mx:
  status: active
  contentType: product
  tags: [espresso, dual-boiler, pid]
  maintainer: "product-team@rancilio.com"
  audience: [humans, machines]
---

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Rancilio Silvia Pro X",
  "description": "Dual-boiler espresso machine with PID",
  "brand": { "@type": "Brand", "name": "Rancilio" },
  "offers": {
    "@type": "Offer",
    "price": "1299.00",
    "priceCurrency": "GBP",
    "availability": "https://schema.org/InStock"
  }
}
</script>
<meta name="mx:status" content="active">
<meta name="mx:contentType" content="product">
<meta name="mx:tags" content="espresso, dual-boiler, pid">
<meta name="mx:audience" content="humans,machines">
<meta name="mx:content-policy" content="extract-with-attribution">

The full contract. Attested content is registered with REGINALD and carries a service-level agreement. The publisher commits to response times, update frequencies, and availability guarantees.

Requirements

• Everything from Level 3
• Registered namespace in REGINALD
• Full Contract of Governance with SLA terms
• Publisher provenance confirmed
• Resolution analytics and cache refresh priority

What It Proves

This publisher is committed to accuracy as a service. Not just "I wrote this" or "I will review this" — but "I guarantee this will be correct and available within defined parameters."

Agent Outcome

Agents can transact with confidence using this content. SLA guarantees, aliveness checks, and registry presence give agents the assurance needed for procurement and commerce workflows. Transaction-stage agent tasks succeed.

Typical Use

Enterprise documentation, regulated industry content, partner integrations. Organisations where downstream systems depend on the accuracy of this data and where SLA breaches have contractual consequences.

The highest level of trust. Everything from Level 4, plus independent third-party verification. An external auditor has confirmed that the content is accurate, the governance processes are followed, and the publisher's claims are substantiated.

Requirements

• Everything from Level 4
• Third-party audit completed and recorded
• Audit trail cryptographically linked to the COG
• Periodic re-audit schedule defined

What It Proves

Someone independent has checked this. Not just the publisher saying "trust me" — a third party confirming "we have provenanced this."

Why It Matters

Level 5 exists for industries where a wrong answer does not just waste tokens — it harms people. A misquoted drug dosage. A wrong financial disclosure. An incorrect legal precedent. These are domains where computational trust must be backed by human accountability.

Agent Outcome

Agents can guarantee accuracy of this content. Independent third-party verification means agents can make safety-critical recommendations backed by human accountability. Guarantee-stage agent tasks succeed.

Typical Use

Healthcare documentation, pharmaceutical data, financial disclosures, legal references. Content where regulatory compliance demands independent verification.