Google Lighthouse adds AI‑agent resource discovery audit – what it means for your site
Lighthouse 13.5 introduces an Agentic Resource Discovery audit that looks for an ARD manifest. Here’s what the change is and how you can get ready.
Google’s Lighthouse 13.5 adds an audit for Agentic Resource Discovery (ARD). The audit checks whether a site publishes a catalog that lets AI agents locate tools, services or callable resources. It looks for a pointer in robots.txt, a link tag, an HTTP Link header or a default file at /.well-known/ai-catalog.json. If none is found, Lighthouse reports “Not Applicable”. The audit will appear in Chrome 156 DevTools and in PageSpeed Insights within two weeks of the release, according to Search Engine Journal.
The ARD audit is part of Lighthouse’s experimental “Agentic Browsing” category, separate from traditional SEO scores. The release notes state it is not tied to Google Search.
How the ARD audit works today
- Lighthouse first scans robots.txt for an Agentmap line that points to a catalog file.
- If no robots.txt directive, it looks for a <link rel="ai-catalog"> tag in the HTML head.
- It also checks the HTTP Link response header for the same relation.
- A valid catalog must conform to the ARD schema; schema errors cause a failure.
- If no catalog is found at any location, the audit is marked “Not Applicable”.
What you should verify on your site
- Confirm whether you already publish an ARD manifest (either ai-catalog.json or ard.json) under /.well-known/.
- If you have a manifest, ensure it is reachable at the expected path and returns a successful response.
- Validate the JSON against the ARD schema – Lighthouse will flag any schema violations.
- Add an Agentmap line to robots.txt pointing to your manifest, e.g. Agentmap: /.well-known/ard.json.
- Optionally, add a <link rel="ai-catalog" href="/.well-known/ard.json"> tag in the page head and/or an HTTP Link header with rel="ai-catalog".
Practical steps for your business
- Audit your current site with the latest Lighthouse 13.5 build (available in Chrome DevTools) or run PageSpeed Insights once the update rolls out.
- If the audit shows “Not Applicable”, you may consider whether an ARD catalog is appropriate for your business. Creating a catalog of your public APIs or services can help align with emerging AI‑agent standards.
- Work with your development team or a web‑development partner to create ard.json that lists your public endpoints, tools or skill definitions per the ARD v0.91 schema.
- Deploy the file to /.well-known/ard.json and test accessibility (e.g., curl https://yourdomain.com/.well-known/ard.json).
- Update robots.txt and HTML headers as described above, then re‑run Lighthouse to confirm a pass.
What to do now
Start a quick internal audit: open Lighthouse 13.5, run the “Agentic Resource Discovery” check, and note the result. If you get “Not Applicable”, map out any public APIs, chatbot endpoints or other services you want AI agents to discover, then create a minimal ARD manifest and publish it. Once the file is live and referenced in robots.txt or a link tag, re‑run the audit to verify compliance. This low‑effort work positions your site for the next wave of AI‑driven search experiences without impacting your current SEO scores.
The services behind this post.
Frequently asked questions.
No. The ARD audit lives in Lighthouse’s separate “Agentic Browsing” category and does not affect traditional SEO metrics. It simply adds a new compliance check for AI‑agent discoverability.
Talk to the team that does this every day.
Tell us what you're trying to grow. You'll get a plan, not a pitch.
Contact Us