GEO & AI Search · October 05, 2026 · 6 min read

Inside /llms.txt and ai-catalog.json: Blueprint for AI Agent Discovery

How autonomous procurement agents and LLM web crawlers read your domain root to verify services, transparent pricing, and contact endpoints.

By Xweba AI Architecture Team (Search & Entity Systems)

Machine-readable llms.txt and AI catalog JSON specification layout

Technical deep-dive into structuring /llms.txt, /llms-full.txt, and /.well-known/ai-catalog.json so AI models index and recommend your offerings accurately.

01. Structuring Clean /llms.txt and /.well-known/ai-catalog.json Files

A compliant /llms.txt begins with an H1 organization heading, a concise blockquote value summary, and clean H2 lists pointing to canonical URLs. Keeping the summary crisp prevents token truncation by fast inference bots.

Publishing Verifiable JSON Catalogs

Exposing /.well-known/ai-catalog.json lets procurement agents confirm service definitions, currency support, and official endpoints in a single call. Pair this guide with AI-Ready Website Data and GEO Playbook for ChatGPT & Perplexity.

  • Keep /llms.txt focused on canonical primary URLs and factual summaries.
  • Serve complete pricing tables and FAQs inside /llms-full.txt for deep research.
  • Validate JSON syntax cleanly under /.well-known/ai-catalog.json.

Continue Reading on Digital Pulse

View all Digital Pulse articles →