Google Flags LLMs.txt As Speculative, John Mueller Favors WebMCP

Search Engine Journal and Search Engine Land report a split in Google product guidance over the proposed llms.txt convention. Google Search Central's public guidance, as summarized by Search Engine Journal, states publishers "do not need special AI files such as llms.txt" to appear in generative AI search experiences. In parallel, Chrome Lighthouse 13.3 added an experimental Agentic Browsing audit that checks for the presence of an llms.txt file and for WebMCP integration, per Search Engine Land and Lucid Media. The Lighthouse documentation describes llms.txt as an "emerging convention" and warns that "Without llms.txt, agents may spend more time crawling the site to understand its high-level structure and primary content," a phrasing cited in Search Engine Land. Several outlets, including Search Engine Journal and Lucid Media, report that Google Search Relations lead John Mueller has publicly compared llms.txt to the deprecated keywords meta tag.

Industry reporting frames llms.txt as a proposed, machine-readable summary placed at a site's root, meant to speed agent discovery of high-level site structure. Chrome's Agentic Browsing audit bundles four signals for "agentic readiness": WebMCP registration, llms.txt presence, accessibility-tree integrity, and layout stability (CLS), according to Search Engine Land and Lucid Media. Coverage describes WebMCP as a browser-driven protocol that lets agents invoke site-hosted tools or APIs directly rather than relying on screenshots or ad-hoc scraping, which could reduce brittle parsing and increase deterministic interactions for agents.

Editorial analysis: Observers covering both Search and developer tools note this is an example of divergent guidance produced by different product teams inside large platforms. Reporting frames the Search team's guidance as focused on ranking and generative-feature visibility, while Lighthouse's experimental audit targets browser-driven agents and developer ergonomics. Comparable tensions have appeared previously where search-ranking advice differs from developer-tooling checks; industry commentary in the scraped coverage places this instance in that ongoing pattern.

Editorial analysis: Practitioners and tool builders should track three indicators:

• •whether browser and agent vendors beyond Chrome adopt llms.txt or WebMCP
• •any formal specification moves or registration activity at llmstxt.org
• •whether Search Central updates its guidance to reference WebMCP or agentic readiness signals. Reporting so far cites Lighthouse behavior and Search Central documentation but does not provide a published, cross-team Google standard that reconciles both positions

Editorial analysis: For teams building agent-facing APIs or working on site interoperability with browser-driven assistants, the coverage suggests prioritizing deterministic, tool-like interfaces such as WebMCP where feasible and treating llms.txt as an experimental, low-effort signal that some developer tools will flag. For those whose primary concern is Search visibility, the documented position in Google Search Central (as summarized by Search Engine Journal) does not require llms.txt for generative search features.

The Lighthouse documentation language quoted in reporting is: "Without llms.txt, agents may spend more time crawling the site to understand its high-level structure and primary content," as cited by Search Engine Land.

All factual claims about product behavior and public commentary are drawn from Search Engine Journal, Search Engine Land, Lucid Media, MarketingAgent.blog, and related coverage of Lighthouse 13.3 and Google Search Central guidance.