Test Your Site for Agentic Readiness with Lighthouse
▼ Summary
– A new Lighthouse report in Chrome Canary checks a website’s readiness for AI agents, covering discoverability, WebMCP integration, and LLMs.txt file evaluation.
– The report provides a ratio of passed agentic readiness checks rather than a percentage score, as shown by Google’s own documentation failing some checks.
– A well-formed accessibility tree is crucial for agents to locate buttons and elements, and agent-friendliness may become a ranking factor for agent recommendations.
– WebMCP is a proposed standard with declarative and imperative types to expose structured tools for agents to use website functionality.
– The LLMs.txt file helps agents understand a site at inference time with instructions and important information, distinct from Search ranking factors.
Google has rolled out a fresh tool to help website owners prepare for the age of AI agents. The new Lighthouse report is now available, and anyone can run it directly from their Chrome browser without needing third-party software. This built-in feature checks whether your site is discoverable by AI agents, whether you have WebMCP integration set up, and perhaps most notably, it evaluates your LLMs.txt file for agent readiness.
How to Access the Agentic Web Report
At the moment, you won’t find this report in the standard version of Chrome. You need Chrome Canary, the beta version that previews upcoming features. Once installed, right-click any page, select “Inspect Page,” and navigate to the Lighthouse tab at the top. There, you’ll see a new category labeled “Agentic Browsing.”
Walk Through the New Lighthouse Report with Me
I tested this on Google’s own documentation page about the agentic browsing report. Surprisingly, even Google’s site showed issues that could hinder agents. Instead of a score out of 100, the report provides a ratio of how many agentic readiness checks your site passes. It’s a straightforward pass-fail breakdown rather than a traditional performance grade.
3 Key Areas to Watch for Agentic Readiness
Here are the main topics I’ve been discussing with clients as this shift unfolds.
1. AI Accessibility and the Accessibility Tree
Agents can interpret your pages in three ways: through vision (screenshots), HTML, and the accessibility tree. Originally designed for screen readers, the accessibility tree actually reveals where buttons are and which elements matter most. If your accessibility tree is poorly structured, agents will struggle to navigate your site. I believe that agent-friendly pages will eventually become a ranking factor for whether agents recommend your content.
2. Understanding WebMCP
WebMCP is a proposed web standard that helps you build and expose structured tools for AI agents. Think of it as a way to teach agents how to use your website’s functionality. There are two types: declarative and imperative. Declarative involves simple code wrapped around a form, while imperative allows back-and-forth interaction between the agent and your site. If your site offers tools that people will use with agents, WebMCP will be critical.
3. The LLMs.txt File
This might seem confusing because Google recently stated that you don’t need an LLMs.txt file for ranking in AI features of Search. However, this report is about agents using your site, not Search. The LLMs.txt file (similar to robots.txt) provides markdown information that helps agents understand your site during inference. It lets you give specific instructions about what agents are allowed to do and where to find important data. You likely won’t need an LLMs.txt file unless you have elements specifically designed for agent use. Remember: LLMs.txt is for agents, not for Search.
I highly recommend setting aside time to test your own site in Chrome Canary. Most of us don’t need these files right now, but staying aware of them is essential as our websites become increasingly agentic.
(Source: Search Engine Journal)
