Glossary

Lighthouse Agentic Browsing

The experimental Lighthouse category focused on machine interaction and agentic readiness signals.

  • glossary
  • lighthouse
  • agentic web

Lighthouse Agentic Browsing refers to the experimental Lighthouse category that evaluates selected signals of how a site is constructed for machine interaction.

Chrome documentation frames it as a set of deterministic audits related to agent use, rather than a classic SEO score.

What the category observes

The publicly documented signals include:

  • WebMCP and possible structured tool exposure;
  • accessibility-tree integrity;
  • programmatic names for interactive elements;
  • visual stability, including CLS;
  • the presence of a machine-readable root summary such as llms.txt.

These signals do not cover everything. They do not determine whether the offer is clear, whether the brand is properly disambiguated, whether proof is sufficient or whether a complete business journey remains coherent.

Why this is not an SEO factor

The commercial shortcut should be avoided: “Lighthouse checks llms.txt, therefore llms.txt becomes an SEO factor.”

Google Search documentation for AI features states that site owners do not need to add an llms.txt file, Markdown or a special machine file to appear in generative Search experiences.

Lighthouse is therefore not validating a ranking promise. It is making another layer observable: the site’s preparation for machine interaction.

How Pagup uses it

Pagup treats Lighthouse Agentic Browsing as an input signal, not a verdict.

In the Agentic Readiness Audit, these audits help open the discussion, but the decision is made on real journeys: finding a service, understanding an offer, choosing a CTA, filling a form, interpreting a confirmation or keeping the right context after interaction.

The Agentic Readiness Score therefore goes beyond a Lighthouse result: it connects technical structure, interface, accessibility, content, proof and governance.