March 11, 2026 · 8 min read

MSP-1: The New Semantic Metadata Standard That Tells AI Who You Are

Today we are shipping full MSP-1 support across AgentSEO.guru. Every analysis now checks for MSP-1 metadata at /.well-known/msp.json, scores it as part of your AI Discovery Files, generates both site-level and page-level MSP-1 files for you, includes them in competitor comparisons, and bundles them in your download package.

This is a big deal. MSP-1 fills a gap that none of the other AI discovery files cover: semantic identity and trust. While llms.txt tells AI what you do, and agent.json tells it what you can do, MSP-1 tells it who you are and why you should be trusted.

What Is MSP-1?

The Mark Semantic Protocol (MSP-1) is an open standard released in December 2025 by the MSP-1 organization. It defines a JSON-LD based metadata format that websites can use to declare structured information about their identity, intent, content provenance, and trustworthiness.

The standard operates at two levels:

Site-Level: msp.json

Deployed at /.well-known/msp.json, this file declares your site's identity, purpose, primary intent, verification status, and trust signals. Think of it as a passport for your entire domain that AI systems can read before crawling a single page.

Page-Level: JSON-LD

Embedded in the <head> of individual pages, page-level MSP-1 metadata declares the intent, author, and content provenance of each specific page -- including whether AI was involved in creating the content.

Why MSP-1 Matters for AI Visibility

The AI discovery landscape is moving fast. Files like llms.txt and agent.json help AI agents understand what you do, but they don't address a fundamental question that AI systems increasingly need answered: should I trust this source?

MSP-1 addresses this directly with three core schemas:

1. Identity and Intent

MSP-1's site schema declares who you are (name, URL, description) and what your site's primary intent is. The intent field uses a structured category system -- informational, commercial, transactional, educational, entertainment, or navigational -- giving AI systems immediate context about what kind of content to expect.

2. Content Provenance

The provenance schema at the page level declares how content was created. It includes an explicit aiInvolvement field with five levels: none, research-assist, editing-assist, co-created, and ai-generated. As AI-generated content detection becomes a bigger factor in trust assessment, being transparent about provenance is a competitive advantage.

3. Trust Signals

The trust schema lets you declare structured trust indicators: SSL status, contact availability, physical address, social media presence, and more. AI systems that are evaluating whether to cite you can consume these signals instantly instead of having to infer them from scattered HTML.

What We Generate

When you run a scan on AgentSEO.guru, we now generate two MSP-1 files alongside your other AI discovery files:

Site-Level: /.well-known/msp.json

This is the primary file. It includes your business identity, a description derived from your actual content, your site's intent category (automatically detected), verification fields, and trust signals computed from what we find during the crawl -- SSL, contact info, physical address, social profiles, and your AI visibility score.

Here is a simplified example of what we generate:

{
  "$schema": "https://msp-1.org/schema/site/site.json",
  "id": "acme-plumbing",
  "name": "Acme Plumbing",
  "url": "https://acmeplumbing.com",
  "protocol": "MSP-1",
  "version": "1.0.0",
  "description": "Licensed plumbing service...",
  "intent": {
    "statement": "Provide authoritative information...",
    "category": "commercial",
    "audience": "Homeowners and businesses..."
  },
  "verification": {
    "core": true,
    "verified": false,
    "verificationLevel": "self-declared"
  },
  "trust": {
    "level": "established",
    "signals": ["ssl-secured", "contact-available",
                "physical-address", "social-presence"]
  }
}

Page-Level: MSP-1 JSON-LD

The second file is a page-level template with installation instructions. It shows you exactly how to embed MSP-1 metadata in your homepage's <head>, including the intent, author information, and provenance declaration. You can customize it for each page type on your site.

How Scoring Works

MSP-1 is now part of the AI Discovery Files scoring category. It is worth up to 7 points out of the new 55-point maximum for discovery files:

Score Criteria
7/7 (Pass) Valid JSON with protocol: "MSP-1", name field, plus intent or verification fields
4/7 (Partial) Valid JSON with protocol declaration or name, but missing rich fields
2/7 (Partial) File exists but is missing required protocol and name fields
0/7 (Fail) No msp.json found at /.well-known/msp.json

We also now check competitors for MSP-1 during competitor analysis. The comparison table on your results page shows a new MSP-1 column so you can see at a glance whether your competitors have adopted the standard.

How to Deploy

Deploying MSP-1 is straightforward -- it follows the same .well-known convention as agent.json:

  1. Site-level: Upload msp.json to your website's .well-known folder. It should be accessible at yourdomain.com/.well-known/msp.json.
  2. Page-level: Copy the <script type="application/ld+json"> block from the page template into the <head> section of your pages.

If you already deployed agent.json to .well-known/, then you already have the folder -- just add msp.json next to it. If you're using the AI Visibility Report in Cursor, Replit, or Windsurf, the report already includes the MSP-1 files and deployment paths.

Why We Think This Matters

MSP-1 is version 1.0.0 and adoption is still early. As of today, virtually no websites have deployed it. That's exactly the same position llms.txt was in six months ago -- and look how fast that changed.

The AI ecosystem is converging on a common need: structured metadata about who a source is, not just what it contains. Google's E-E-A-T framework already rewards expertise and trustworthiness. AI models are developing their own trust heuristics. MSP-1 provides a machine-readable way to present that information directly.

By adding MSP-1 now, you are:

What Changed in the Product

Here's a summary of everything that shipped today:

See Your MSP-1 Score

Run a free scan to check whether your site has MSP-1 metadata deployed -- and get your complete AI Discovery Files score across all 10 layers.

Scan Your Website Free →

Further Reading