Google’s Lighthouse 13.5 introduces an audit for Agentic Resource Discovery (ARD), which is a projected spec for how AI agents find the tools and services offered by an organization. The audit compares a site’s ARD catalog alongside the schema defined by the ARD project. If the location does not provision a catalog pointer, Lighthouse defaults to /.well-known/ai-catalog.json. The most latest spec, dated Aug. 26, moved this document to ard.json, alongside the old filename remaining optional for application that says it.
The Lighthouse squad expects this update to rotate out in Chrome 156 DevTools, and in PageSpeed Insights inside two weeks.
This audit is part of Lighthouse’s experimental Agentic Browsing category, distinct from SEO audits, and it is not tied to Google Search anyplace in the release notes. The reports volition display an ARD outcome or Not Applicable if no catalog is found.
What The Audit Checks
In the 13.5 origin code, Lighthouse checks the robots.txt document for an Agentmap line, which is a directive from the ARD specification pointing to a catalog. It afterward looks for a nexus tag alongside the ai-catalog connection and additionally checks the HTTP Link header for the identical relation. If none of the three is available, it requests /.well-known/ai-catalog.json.
If there is no pointer and the well-known way does not assist the file, the audit is marked as Not Applicable. Schema errors logic failure, as does a pointer to an unloadable catalog. The pull request describes ard-schema as a “schema conformance audit.”
The identical publish groups the new audit alongside the existing llms.txt audit under a heading called Agent Discoverability. Google’s docs say the Agentic Browsing category doesn’t create a 0-100 mark and shows a continue proportion instead, since “the standards for the agentic web are motionless emerging.”
The Spec Has Since Renamed The File
ARD v0.91 now specifies /.well-known/ard.json as the manifest location and mentions ai-catalog.json as the before path. When I covered ARD’s initiate in June, the spec was a v0.9 draft, alongside ai-catalog.json as the document for organizations to publish.
In v0.91, application study these records is required to fetch ard.json and can additionally inspect the older path. The spec cautions that a document lone accessible at the old way “may not be found.”
I didn’t discover any references to ard.json or the ard nexus connection in the 13.5 origin or in the chief branch of the project as of Sept. 21. The spec additionally mentions finding through in-page JSON-LD and DNS records, but the 13.5 code doesn’t inspect either. Lighthouse built its validator according to the ARD project’s own conformance tests.
The specification is motionless a recommendation and credits three authors: Junjie Bu from Google, R.V. Guha from Microsoft, and Shaun Smith from Hugging Face.
How ARD Differs From llms.txt And WebMCP
ARD is all concerning discovering resources. It includes MCP tools, A2A agents, skills, and another callable services, so AI systems can discover them through search.
An llms.txt document summarizes your site’s satisfied for agents. WebMCP, which I shared concerning in August, allows a leaf to recommendation organized actions that an delegate can call formerly it’s on your site.
Why This Matters
Based on Lighthouse 13.5 origin code, if a website uses /.well-known/ard.json alongside rel="ard", doesn’t have an Agentmap line, and doesn’t use the older ai-catalog names, it would be classified as Not Applicable.
This merely method Lighthouse didn’t discover a catalog in its current hunt areas. But it doesn’t necessarily average the location is missing a current ARD manifest.
As I shared in May, whenever Lighthouse introduced its llms.txt audit, recall that Google Search’s direction and Lighthouse’s delegate checks are looking at distinct things.
Looking Ahead
Lighthouse expects 13.5 to attain PageSpeed Insights inside two weeks of the release.
The publish additionally adds a weekly automated check that flags whenever the ARD project’s schema or conformance test changes upstream. The places Lighthouse looks for a catalog are set separately, in the code that fetches it.
Featured Image: Golden Dayz/Shutterstock
