Chrome Lighthouse now checks for llms.txt

Google’s latest experimental Lighthouse update introduces a new “Agentic Browsing” audit category, and one of its key checks is whether a site has an llms.txt file. According to the new documentation, this file is framed as a discoverability and efficiency signal for AI agents, not as a traditional crawling directive for search engines.

The audits fall under Chrome’s emerging Agentic Browsing category, designed to evaluate how well a website is structured for machine interaction. This development comes less than a week after Google published updated guidance on optimizing for generative AI features like AI Overviews and AI Mode, where it explicitly stated that llms.txt files are not necessary for search performance.

What Lighthouse now checks. Google’s documentation explains that the Agentic Browsing category uses deterministic audits to assess “how well your site is constructed for machine interaction.” The specific checks include:

• WebMCP integration.Lighthouse specifically looks for “the presence of a machine-readable summary at the domain root.” Google explains the rationale: “Without llms.txt, agents may spend more time crawling the site to understand its high-level structure and primary content.” Notably, this audit category does not produce a traditional 0-100 score. Instead, Google surfaces a fractional pass ratio alongside pass/fail checks tied to agentic readiness signals.The tension. While the new Lighthouse documentation does not directly contradict Google’s earlier advice about optimizing for generative AI search, it does create a nuanced distinction. Those earlier recommendations focused on Google Search rankings, whereas these audits target AI agents and browser tools. Still, seeing llms.txt included in Chrome’s own readiness checks may prompt some SEOs to reconsider earlier skepticism about the file’s value.Agentic engine optimization. The Lighthouse audits align with concepts that Google Cloud AI engineering director Addy Osmani outlined in April regarding Agentic Engine Optimization. Osmani noted that AI agents with limited context windows may truncate long pages or miss critical information buried deep in content. His recommendations included:
• Cleaner semantic structure.SEO vs. llms.txt. Google’s official stance, as stated in its mythbusting section on generative AI search, remains clear: “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in generative AI search. Note that Google may discover, crawl, and index many kinds of files in addition to HTML on a website: this doesn’t mean that the file is treated in a special way.”Google’s John Mueller further clarified on Bluesky when asked about the apparent irony of Google using llms.txt files while advising others not to. He explained that it’s not done for search: “There’s more to websites than just SEO :-).” He distinguished between discovery (finding a site via a global search engine) and functionality (helping users complete tasks once they arrive). For Google’s developer documentation site, AI coding tools benefit from easily parsed reference material. “In those cases, it can help to give them a way to understand the context of the documentation they’re looking at, as well as a simplified version of the reference page.” However, he added that for non-developer sites, this makes little sense: “Making a markdown version of a shoe’s specs is not going to get you more sales.” He cautioned against prioritizing hypothetical future scenarios over current SEO needs.What Google says agents rely on. Beyond llms.txt, the new Lighthouse category places heavy emphasis on accessibility and interface stability. The documentation states that agents rely on the accessibility tree as their “primary data model.” Specific evaluations include:
• Programmatic labels for interactive elements.Google also warns that dynamically registered WebMCP tools and large DOM changes can affect audit results.Why we care. Google says you don’t need llms.txt for Search, but Chrome is now actively checking whether it exists. At the same time, Google’s agentic tools appear to favor sites that are easier for machines to read and use, particularly those with strong accessibility, stable layouts, and clear agent access. This creates a strategic consideration for SEOs: balancing current search priorities with emerging agentic readiness signals.