WSS
Web Specification Studio Home
Free AI Agent Audit Tool

How Ready Is Your Website for AI Agents?

Scan your website to see how ready it is for AI agents. We check multiple web standards and emerging protocols: from robots.txt, XML sitemaps, and Schema.org to Markdown content negotiation, llms.txt, and MCP.

Free live scan for any public domain

Evolution

Why agent readiness matters for your website

A great website delivers a wonderful experience to people while also communicating smoothly with the automated tools that connect visitors to your work.

Every day, search crawlers, AI research assistants, API integrations, and developer agents visit your site to understand your content, catalog your services, and answer questions for users.

By providing clean HTTP responses, well-organized sitemaps, rich Schema.org structured data, and reliable endpoints, you make it easy for these systems to find your pages, understand your entities, and cite your content accurately.

Agent readiness gives your website a clear competitive advantage: faster discovery, higher citation accuracy in search, and dependable interfaces for automated tools and workflows.

The 3 Levels of Agent Readiness

How websites evolve from traditional human browsing to full AI agent automation:

Maturity Model
01

Discovery: Can AI Crawlers Find You?

Foundation

If crawlers like GPTBot or ClaudeBot hit bot blocks, slow redirects, or missing sitemaps, your website remains invisible to AI answers.

Key signals:robots.txtsitemap.xmlHTTP 200 OK
02

Understanding: Can AI Models Read Your Facts?

Accuracy

Scrapers parsing cluttered HTML layouts and heavy JavaScript often hallucinate prices or misquote specifications when answering user queries.

Key signals:/llms.txtJSON-LD SchemaSemantic HTML
03

Action: Can Autonomous Agents Use Your Tools?

Automation

Beyond reading articles, coding agents and assistants can query databases, run workflows, and execute tasks directly on your platform.

Key signals:MCP ServersOpenAPI SpecsAgent Skills
Architecture

The 6 Layers of an Agent-Friendly Website

AI agents, search engines, and automated crawlers interact with your website across six practical areas. Each layer helps automated visitors find, read, and use your pages effectively:

01

Discovery

Can AI tools and search engines find your pages easily?

02

Content Retrieval

Can automated visitors download clean, readable content quickly?

03

Content Understanding

Can AI assistants understand what your content and data actually mean?

Established methods:
04

Crawler Rules

Do you give search bots clear rules on what to visit and what to skip?

Established and emerging signals:
05

Tool Capabilities

Can AI agents interact with your tools, endpoints, or services?

Established and emerging tools:
06

Safe Actions & Transactions

Can automated clients complete actions and checkouts securely?

Emerging protocols:
  • Universal Commerce Protocol (UCP)
  • Agentic Commerce Protocol (ACP)
  • Machine Payment Protocol (MPP) and HTTP 402
  • authenticated state transitions
  • structured machine checkout interfaces

These layers are complementary. Implementing one layer does not automatically satisfy another, and not every website requires capabilities in all six layers.

Recommendations

Best Practices for an Agent-Friendly Site

Three simple guidelines help keep your website reliable for automated tools and completely secure for your users:

Clear Guidance & Strong Security

Publishing robots.txt guidelines helps search engines and automated crawlers discover your content responsibly. Pair these directives with standard authentication to keep private areas completely secure.

End-to-End Response Quality

Beyond confirming files exist, testing HTTP status codes, content-type headers, and schema syntax ensures automated tools parse your resources quickly and reliably.

Reliable Web Standards

Following established W3C and IETF standards ensures long-term compatibility across all search engines, AI assistants, and future web clients.

Methodology

How the Audit Works

The scanner inspects your domain step by step across six clear phases:

01

Discover

Find Public Endpoints

Locate robots.txt, XML sitemaps, canonical tags, discovery headers, and well-known configuration paths.

02

Retrieve

Fetch Resources

Request resources with standard HTTP calls, checking response codes, content types, and caching headers.

03

Parse

Read Content Signals

Parse robots rules, XML sitemap URLs, Schema.org JSON-LD graphs, and document text structures.

04

Validate

Check Web Standards

Compare actual results against RFC web standards, Schema.org definitions, and official protocol specs.

05

Classify

Assign Clear Status

Label each finding as PASS, FAIL, WARNING, INFO, or N/A based on clear, reproducible evidence.

06

Explain

Provide Fixes

Give clear explanations, show the exact impact on automated tools, and provide ready-to-use fixes.

Scope of Audit

What We Check on Your Website

The scanner runs automated checks across seven practical areas on your production domain:

01

Web Discovery

We inspect the mechanisms that help automated clients locate your site's resources without brute-force crawling.

Checks include:robots.txt accessibility • XML sitemap declaration • sitemap HTTP response • canonical URL link tags • HTTP Link headers • emerging discovery signal (DNS-AID TXT: Experimental) • discovery consistency

Objective: Verify whether an automated client can determine your site structure deterministically. DNS-AID TXT records are classified as an experimental discovery convention rather than an established web standard.

02

Content Delivery & Page Rendering

We examine how your website delivers content to automated clients, verifying real content delivery and distinguishing fast server responses from empty client-rendered shells.

Checks include:HTTP status codes • Content-Type declaration • server-rendered HTML payload • Markdown content negotiation (Accept: text/markdown) • Vary: Accept header presence • canonical URI preservation • representation efficiency (Informational)

We do not merely detect Markdown. We test whether the negotiated representation behaves correctly: preserving resource identity, matching canonical URLs, returning accurate media types, providing caching directives (Vary: Accept), and serving equivalent semantic content.

03

Crawler Rules & Bot Policies

We analyze the policies your site publishes for automated crawlers and automated systems.

Checks include:RFC 9309 robots.txt syntax • User-agent rule blocks • Allow and Disallow consistency • automated crawler rules (ClaudeBot, PerplexityBot, GPTBot) • Content Signals (search, ai-input, ai-train)

Content Signals are an emerging machine-readable policy mechanism for communicating permitted uses of content. They complement crawler access directives but do not replace authentication or network-level enforcement.

04

Structured Data & Schema Markup

We inspect structured information exposed by your pages to assess explicit entity grounding.

Checks include:JSON-LD script blocks • Schema.org types • entity identifiers • Organization • Article • Product • BreadcrumbList • entity relationship graph • schema syntax validity

Structured data gives machines explicit information about entities and relationships that would otherwise have to be inferred from page text.

05

AI Documentation & llms.txt

We look for curated indexes intended to help automated systems navigate a site with minimal token overhead.

Checks include:llms.txt existence • llms.txt Content-Type • valid Markdown parsing • curated link validity • llms-full.txt presence • API catalogs

llms.txt is an emerging convention for LLM-oriented site summaries. It is not an established web standard and is not required for search indexing or Google AI features.

06

Interactive Tools & MCP Protocols

We evaluate whether a site exposes machine-actionable tools and capability declarations.

Checks include:MCP Server Card convention (/.well-known/mcp.json) • WebMCP browser tool registration API • Agent Skills directory • OAuth 2.0 discovery • authentication documentation (auth.md)

WebMCP is an emerging W3C Community Group browser API allowing websites to expose in-page tools. MCP Server Cards are an ecosystem convention for advertising endpoints. Files like auth.md and skill directories are reported as informational machine documentation signals, not formal standards.

07

Automated Checkout & Payments

For transactional web applications, we inspect emerging machine payment and commerce signals.

Checks include:Universal Commerce Protocol (UCP) • Agentic Commerce Protocol (ACP) • Machine Payment Protocol (MPP) • HTTP 402 payment headers • structured product offers

Transactional capabilities are evaluated against specification metadata when applicable. Sites that do not conduct commerce are marked Not Applicable rather than penalized.

Taxonomy

How Standards and Protocols Are Grouped

The web contains official standards, community protocols, and emerging AI conventions. We classify each check clearly so you know what is an established web standard and what is an emerging practice:

ClassificationDefinitionExamples in Scanner
Web StandardFormal specification maintained through a standards body or standards processHTML living standard, RFC 9110 HTTP Semantics, RFC 9309 Robots Exclusion Protocol
ProtocolA defined machine-to-machine communication mechanismHTTP, DNS, Model Context Protocol (MCP)
VocabularyA defined semantic vocabulary or data modelSchema.org, JSON-LD serialization
PracticeWidely adopted implementation practice across production websitesXML Sitemap protocol, Open Graph metadata, canonical link tags
EmergingDeveloping specification, Community Group draft, or ecosystem conventionllms.txt, WebMCP, Content Signals, UCP, MPP
Vendor-SpecificMechanism defined primarily by a particular provider or platformCrawler-specific token rules, cloud bot headers
InformationalObservable signal without a normative compliance requirementRepresentation efficiency, raw byte size, response timing

The absence of an emerging convention or informational signal is never treated as a standards failure.

Capability Framework

Agent Readiness Profile and Levels

Agent readiness is not a single competition where every website must reach the same destination. Different websites serve different operational roles. The Level is a summary headline, while the profile evaluates distinct capability dimensions independently:

Capability DimensionAssessment FocusTarget Outcome
Discoveryrobots.txt, XML sitemaps, canonical links, Link headersDeterministic discovery without brute-force crawling
RetrievalRepresentation integrity, Content-Type, Vary headers, server HTMLPredictable payloads matching resource identity
SemanticsSchema.org vocabularies, JSON-LD graphs, entity groundingExplicit entity disambiguation without text inference
PolicyRFC 9309 crawler directives, Content Signals permissionsExplicit machine-readable crawler and usage preferences
CapabilityREST/OpenAPI catalog, MCP server convention, WebMCP toolsDiscoverable and invocable machine interfaces
TransactionMachine payments, HTTP 402, UCP/ACP commerce manifestsStructured state-changing commerce (or marked N/A)
Level 0level-0

Unavailable

The machine-facing layer is substantially inaccessible or non-functional.

Typical conditions: HTTP connection errors • broken response headers • severe crawler restrictions • empty client-side DOM with zero server-rendered content

Level 1level-1

Crawlable

The website exposes basic, retrievable web documents.

Typical capabilities: valid 200 OK responses • retrievable HTML • valid robots.txt • fundamental web crawler access

Level 2level-2

Discoverable

The site provides explicit mechanisms for automated resource discovery.

Typical capabilities: XML sitemaps • canonical URL declarations • consistent internal linking • well-known discovery resources

Level 3level-3

Structured

The site exposes explicit machine semantics and entity grounding.

Typical capabilities: Schema.org types • valid JSON-LD graph • entity relationships • structured metadata

Level 4level-4

Interoperable

The site provides machine-oriented representations and clear policy signals beyond standard HTML.

Typical capabilities: HTTP content negotiation (Accept: text/markdown) • Vary: Accept headers • Content Signals • curated machine indexes

Level 5level-5

Agent-Interactive

The service exposes machine-invocable tools, interfaces, or transaction protocols.

Capabilities: Model Context Protocol (MCP) server endpoints • WebMCP browser tool declarations • API catalogs • authenticated tool execution

Describes an observable architectural capability, not certification and not a guarantee of third-party client behavior.

Transparency

How We Report Audit Findings

Instead of giving you a mysterious single score, every check gives you transparent findings and clear explanations:

Status & Confidence

PASS • FAIL • WARN • INFO

Clear verification status accompanied by High, Medium, or Low certainty.

Scope & Method

Tested Endpoints

The exact URL, HTTP method, and negotiation headers checked during the audit.

Observed Facts

Direct Observations

Directly measured response status codes, header values, and parsed structures.

Derived & Action

Why It Matters & Fixes

Clear analysis of tool impact plus exact, copy-paste code to fix issues.

Audience

Who This Tool Is Built For

Agent Ready Scanner provides technical teams with clear, reproducible signals to test machine-facing endpoints:

Web Developers

Validate HTTP behavior, content negotiation, Schema.org grounding, and machine interfaces before deploying to production.

Platform Developers

Inspect headers, discovery endpoints, crawler policy directives, and edge infrastructure behavior across environments.

Security Teams

Separate crawler policy from actual access control and identify unsafe assumptions regarding bot directives and endpoints.

API and Integration Teams

Evaluate machine-discoverable capabilities, Model Context Protocol (MCP) server endpoints, and automated client interfaces.

Technical SEO Teams

Understand the machine-readable infrastructure underlying automated retrieval, crawling efficiency, and entity grounding.

Q&A

Frequently Asked Questions

Common questions regarding agent readiness, standards compliance, and how the scanner evaluates modern websites:

Is agent readiness an official web standard?

No single specification named 'Agent Readiness' exists. It is an operational profile evaluating how well your site implements official web standards (RFC 9110 HTTP Semantics, RFC 9309 Robots Exclusion Protocol, W3C Schema.org/JSON-LD, HTML living standard) alongside emerging community protocols (Model Context Protocol, WebMCP, Content Signals, llms.txt).

Does a website need llms.txt?

No. llms.txt is an emerging proposal for curated site summaries. Established search crawlers and AI assistants primarily discover and read standard HTML pages, XML sitemaps, robots.txt, and Schema.org markup. llms.txt is helpful context, not a prerequisite.

Does robots.txt control automated clients?

robots.txt is an advisory protocol (RFC 9309). Reputable crawlers respect it, but malicious scrapers and arbitrary clients can ignore it. It is a discovery and indexing policy, never a security boundary. Private resources must be protected by authentication.

Does this guarantee visibility in search engines or AI assistants?

No. The scanner evaluates technical infrastructure: whether your endpoints respond correctly, provide valid machine metadata, and support automated retrieval. Editorial relevance, domain authority, and content quality are assessed by the downstream platforms themselves.

What is WebMCP and how does WSS evaluate it?

WebMCP is an emerging W3C Community Group proposal for exposing machine-actionable tools inside web pages. The scanner checks whether your site declares tool capabilities via standard descriptors or in-page APIs. Because it is an emerging proposal, absence of WebMCP is reported as informational, not a failure.

How does the scanner audit MCP Server Cards?

The scanner inspects the ecosystem convention resource at /.well-known/mcp.json, testing if it returns 200 OK with valid JSON, contains expected metadata fields, and points to valid server interfaces.

What is the difference between a machine-readable website and an agent-ready website?

A machine-readable website focuses on retrieval and semantics: providing clean content, structured JSON-LD, and understandable text. An agent-ready website also provides machine-discoverable interfaces for invocation and state transitions: tool declarations, API endpoints, authorization flows, and transaction capabilities.

Documentation

Related Specifications & Guides

Deepen your understanding of machine-readable protocols, technical SEO, and agent discovery with our documentation and guides:

Agent Readiness Specifications

Complete technical specifications covering llms.txt, Content Signals, Model Context Protocol, and machine content negotiation. Learn how automated clients discover and evaluate your site.

21 Specificationsllms.txtContent SignalsWebMCPMCP Tools
Browse all specifications

The Standards Checklist

Every spec item, flat and tickable. Use as a complete self-audit across web performance, technical SEO, accessibility, and machine-readable signals for your domain.

Self-AuditInteractiveWeb StandardsPrint-Friendly
Open the checklist

Sample Technical Audit

Review a real-world example of an in-depth audit report examining crawler access, rendering speed, and structured data.

Agent Web Architecture

How search crawlers, AI assistants, and autonomous agents discover and process content on modern sites.

Content Negotiation for AI

Implement standard HTTP Accept headers and Markdown responses to reduce token costs and improve retrieval.

Audit your site for the agent-powered web.

Agent Ready Scanner provides a live, point-in-time check of your site's machine-readable infrastructure. Find broken discovery links, invalid robots rules, missing structured data, and unexposed APIs so your site is ready for AI tools and automated agents.

This scanner checks publicly accessible HTTP signals at the time of the scan. Results reflect a point-in-time snapshot. Specifications and ecosystem conventions continue to develop. Review all configurations before deploying to production.