About the Book
What is MX: The Protocols about?
MX: The Protocols: Designing the Web for AI Agents and Everyone Else examines how modern web design optimized for human users fails for AI agents, and how fixing this benefits everyone. The book provides practical guidance for making websites accessible to both humans and AI agents through semantic HTML, explicit state management, and structured data.
Want a preview? View the full chapter outline on the book homepage.
Who should read this book?
The book targets four primary audiences:
- Web Professionals - developers, designers, accessibility specialists looking to make their websites machine-readable
- Agent System Developers - engineers building AI agents that need to browse websites reliably
- Business Leaders - CTOs and product owners making strategic decisions about agent-mediated commerce
- Partners & Investors - evaluating opportunities in the emerging agent economy
What format is the book available in?
The book is published in Kindle format (6"×9" paperback dimensions) with a planned publication in Q1 2026. All appendices are freely available online at the appendices index.
Technical Concepts
What are AI agents in the context of this book?
AI agents are autonomous software systems that browse websites on behalf of users. The book addresses a diverse ecosystem:
- CLI agents - Command-line tools running locally (e.g., Claude Code, Cline)
- Browser automation agents - Using tools like Playwright or Selenium
- Server-based agents - Cloud-hosted agents accessing websites remotely
- Browser extension assistants - In-browser AI tools
- IDE-integrated browser controls - Development environments with browser integration
How is this different from accessibility?
Whilst there's significant overlap with accessibility best practices (semantic HTML, clear structure, explicit feedback), AI agent compatibility has distinct requirements:
- Agents need explicit state attributes (data-state, data-validation-state)
- Structured data for machine reading (Schema.org JSON-LD)
- Clear feedback that persists in the DOM (not transient animations)
- Form field naming conventions (email, firstName, lastName vs custom names)
The book demonstrates how these patterns benefit both agents and humans, creating a universally accessible web.
Are the patterns in this book standardized?
The book carefully distinguishes between different maturity levels:
- Established Standards - Schema.org, semantic HTML, ARIA (use with confidence)
- Emerging Conventions - llms.txt from llmstxt.org (early adoption phase)
- Proposed Patterns - ai-* meta tags, data-agent-visible (not yet standardized)
All proposed patterns are forward-compatible and won't break if agents don't recognize them.
Implementation and Resources
Where can I find the appendices?
All appendices are available online at the appendices index. They include:
- Appendix A - Implementation Cookbook (quick-reference recipes)
- Appendix B - Proven Lessons (production learnings)
- Appendix C - Web Audit Suite (CogNovaMX measurement service)
- Appendix D - AI-Friendly HTML Guide (comprehensive patterns)
- Appendix E - AI Patterns Quick Reference
- Appendix F - Implementation Roadmap
- Appendix G - Resource Directory (150+ curated resources)
- Appendix H - Example llms.txt File
- Appendix I - Pipeline Failure Case Study
- Appendix J - Industry Developments
How do I get started with implementation?
Start with Appendix F (Implementation Roadmap) which provides priority-based guidance:
- Priority 1: Critical Quick Wins - Semantic HTML, form field naming, Schema.org structured data
- Priority 2: Essential Improvements - Error handling, state attributes, llms.txt file
- Priority 3: Core Infrastructure - Platform-wide patterns and consistency
- Priority 4: Advanced Features - Comprehensive long-term enhancements
Appendix A (Implementation Cookbook) provides code examples for common patterns.
What is the Web Audit Suite?
The Web Audit Suite is a comprehensive Node.js tool that analyzes websites for:
- AI agent compatibility (LLM suitability metrics)
- SEO performance
- Accessibility compliance (WCAG 2.1)
- Performance metrics
- Security headers
The tool implements the patterns described in the book and generates detailed reports. It's available as a separate purchase or professional audit service. See Appendix C for complete documentation.
Where can I find code examples?
Code examples are available in multiple locations:
- Implementation Cookbook (Appendix A) - Quick-reference code snippets
- AI-Friendly HTML Guide (Appendix D) - Comprehensive patterns with examples
- This Website - View the source of any page to see the patterns in action
Getting Started
How much does the Web Audit Suite cost?
The Web Audit Suite is available as a separate purchase or professional audit service. Pricing varies based on your needs and the size of your website. Contact info@cognovamx.com for detailed pricing information tailored to your specific requirements.
What programming languages are the examples in?
The book uses simple JavaScript for code examples with minimal dependencies, ensuring they're accessible to developers across all frameworks. Examples focus on fundamental patterns that work with any technology stack, from vanilla JavaScript to React, Vue, Angular, and server-rendered HTML.
Does this work for my CMS/framework?
Yes. The patterns in this book are framework-agnostic and work with any CMS or framework including WordPress, Drupal, React, Vue, Angular, Next.js, and more. The book focuses on HTML output and browser behavior, not specific implementation technologies. If your system generates HTML, these patterns apply.
How long does implementation take?
Implementation is priority-based rather than time-based. Priority 1 quick wins (semantic HTML, form field naming, Schema.org) can be implemented immediately. Priority 2-4 improvements build systematically on foundations. Appendix F (Implementation Roadmap) provides detailed priority-based guidance for phased rollout.
What's the difference between served and rendered HTML?
Served HTML is the static HTML sent from the server before JavaScript execution, what CLI agents and server-based agents see. Rendered HTML is the dynamic state after JavaScript runs, what browser-based agents see. Different agents operate in different modes, so the book addresses both states for universal compatibility.
Do I need to know JavaScript?
Basic web development knowledge is helpful but not required. Many patterns involve HTML attributes and structured data that don't require JavaScript. The book explains concepts clearly with practical examples, making implementation accessible to web professionals of all backgrounds, from designers to backend developers.
Can I use these patterns with React/Vue/Angular?
Absolutely. Modern frameworks like React, Vue, and Angular all generate HTML as final output. The patterns apply to that HTML regardless of how it's generated. The book shows framework-agnostic patterns that work universally, add data attributes to JSX components, include Schema.org in your head section, use semantic HTML in your templates.
How do I validate my implementation?
Use the Web Audit Suite to analyze your implementation and generate detailed reports. Additionally, use Google's Structured Data Testing Tool for Schema.org validation, browser DevTools to inspect HTML attributes, and Pa11y for accessibility compliance. The book provides validation guidance in Appendix C and Chapter 10.
What's the ROI of implementing these patterns?
The field is too new for validated ROI data, but benefits include improved accessibility for all users, better SEO performance, reduced support costs through clearer error messages, and competitive advantage in agent-mediated commerce. Many patterns overlap with accessibility best practices, providing immediate user experience improvements whilst positioning for agent-mediated traffic.
How do I test for AI agent compatibility?
The Web Audit Suite provides comprehensive AI agent compatibility testing through LLM suitability metrics. Test both served HTML (view page source) and rendered HTML (browser DevTools). Check for semantic structure, explicit state attributes, Schema.org data, and clear feedback. Appendix C provides detailed testing procedures and validation criteria.