I built it twice. Didn't place twice. Then Google and Vercel shipped it.

A you.com rep said it at a hackathon kickoff in March. The web has a second audience now. The first one is people. The second is machines. When you ask ChatGPT or Perplexity something, the retrieval crawler behind it fetches your page and reads the markup that comes back. Not the site as a browser would draw it. The markup.

That second audience idea really stuck with me. Then it fired me up. So I went and built a thing about it. I knew I could improve on it, so I built the thing again, better. Both times I submitted to a hackathon. Neither one placed.

On May 7, Google shipped an Agentic Browsing category in Lighthouse. On August 21, Vercel and Ora launched is-agentic.com. That makes four tools now, counting my two, all pointed at the same question: can an agent use this site? Same question, three answers, and we do not agree on what counts as an agent seeing your page.

Hermes was a CLI wearing an agent costume

My first build, Hermes Clew, was for the GitLab Duo Agent Platform Challenge. A Python engine that walked the HTML, JSX, and TSX files in a repo, scored six categories out of 100, and let Duo Chat reason over the findings. It worked. I still like the bones of it.

It also missed the assignment in two ways I did not see at the time.

It read files, not pages. A repo is not what an agent meets. An agent meets a URL.

And it told you how to fix things. Ranked fixes, effort estimates, code samples. I wrote that in because it felt more helpful. I did not place, and by the time I understood why the fixes bothered me, I was already rebuilding.

Findings are the first floor, not the whole building

My second attempt, Agentis Lux, runs on the Perseus Clew engine and takes a live URL instead of a repo. Six deterministic checks, raw HTML, no JavaScript executed, because that is what a retrieval crawler gets. I wrote about the build in My website has two audiences now and the shipped version in Can retrieval agents read your website.

Hermes Clew told you how to fix things. Agentis Lux does not. That is the change I care about most, and it is the one that reads like a missing feature.

I think findings sit at the bottom of a pyramid.

A finding is a fact about what an agent could not do. A fix is a decision about someone's codebase, made by someone who has never seen it, with no idea what the constraints were or who is on call for the regression. Shipping the first fix a model produces seems reckless to me. Maybe I am too cautious. I am also aware enough to know I am not the expert on your code.

I run my own sites through it. It tells me a heading is unreadable to a crawler, and then it stops, and part of me wants the paragraph that says what to type instead. I shipped it without that anyway. I got the same lesson from the opposite direction when Kiro and Claude delivered exactly what I asked for and it wasn't what I wanted.

Confident output is not the same as correct output as I keep learning over and over!

Google decided not to give you a number

Lighthouse 13.3 shipped on May 7, 2026 with a category called Agentic Browsing. Four checks: an llms.txt file, which is a proposed text file that tells AI agents what your site is about, WebMCP, which lets a page advertise tools an agent can call, the agent-relevant slice of the accessibility tree, and cumulative layout shift.

The interesting part is the scoring, or the refusal of it. The category does not produce a weighted average from 0 to 100. Because the standards for the agentic web are still emerging, the stated focus is gathering data and providing actionable signals rather than a definitive ranking. You get a pass ratio. Google marked the whole category experimental and under development.

That is a restrained choice from the company that made everyone care about a 0 to 100 performance score, and I respect it.

The measurement model underneath is a different animal from mine. Lighthouse calls the Chrome DevTools Protocol WebMCP domain to watch tool registration, verifying both declarative tools defined in HTML and imperative ones defined in JS, and it filters a subset of accessibility audits that matter for machine interaction, because agents rely on the accessibility tree as their primary data model.

That is a requirements list. It needs Chrome. It needs JavaScript to run. It needs the page to register tools at runtime. That is an agent driving a browser.

A retrieval crawler does a fetch and parses what comes back. No Chrome, no hydration, no registered tools. So Lighthouse and Agentis Lux can look at the same URL and produce two defensible readings, because we are not looking at the same artifact. Lighthouse reads the rendered page. I read the response body.

There is a second split hiding in those four checks. llms.txt and WebMCP are things you add. A file you publish, a tool you register. They measure whether you opted into a proposal. Semantic HTML and the accessibility tree are things your site already is or is not. Two of the four checks reward adoption. Two reward construction. DebugBear says llms.txt is not currently widely adopted by AI services and calls the practical benefit of adding one doubtful. The evidence underneath that is an Ahrefs study of 137,210 domains: 28% published a valid llms.txt, and 97% of those files got zero requests in May 2026. No AI bot went looking for one that did not exist.

I build for the second kind. Not because adoption is wrong, but because I can only measure what is there.

That is also why the Agentis Lux engine, Perseus Clew, does not run JavaScript. Two reasons, and I wrote them into the architecture doc under a NEVER list before I wrote the scanner. The product reason is that a retrieval crawler does not run it either, so rendering the page would mean scoring an artifact my reader never sees. The engineering reason is plainer: a headless browser in a Lambda means cold starts and memory I did not want to pay for on a free public scan.

What I did do was leave the door open. The score is not a bare field in the report. It sits under a render-mode label, scoredViews.rawHtml, so a future version that renders the DOM can emit a second mode next to it and show the delta between what a JS-capable agent sees and what a fetch-only one sees. The check modules take parsed output and do not care how the HTML arrived, so rendering is an addition at the fetch seam, not a rewrite.

One nesting level, spent up front, to keep a decision from becoming permanent.

Vercel decided to give everyone a number, in public

Vercel and Ora launched is-agentic.com on August 21. Audits with 100+ checks, visualizations of agents using your site, one-click prompts to fix problems, and a CLI for agents. Ora's own post describes the ranker underneath as 127 checks across four layers, discovery, access, usability, and payments, with a relevance pass that sets aside what does not apply so a free tool is never marked down for missing payment rails. Another writeup counted 118. The numbers differ by source, which is what I would expect from two products sharing one engine.

The scope is wider than mine by a lot. Payments and transaction surfaces are a lane I never entered.

Two observations, and I will leave them as observations.

The reports are public. Ora says vercel.com scores 85 on the ranker, second of 310 companies in Infrastructure and DevOps, in the top 1% of the 16,000+ domains they scan. Agentis Lux stores a scan result under an opaque id for 24 hours and then it is gone. Neither is the correct answer. They are different beliefs about who a report belongs to.

The other one I keep turning over. Vercel published a document called the Agent Readability Spec, and within weeks other people's tools were checking sites against it by name, in CLIs, in a GitHub Action, in an MCP server. A hosting company wrote a standard and the ecosystem picked it up. No standards body was in the room.

That is the adoption split again, one level higher. llms.txt asks whether you added a file. The spec asks whether you followed a document a vendor published. Neither one measures what your site is. Both measure whether you signed on.

The row I keep coming back to is whether it runs your JavaScript

Researched with AI assistance from public documentation in September 2026. Every claim links to a primary source below. These tools move fast, so check the source before you quote me.

The row that decides the rest is "runs your JavaScript," and the one under it, reader or actor, is just that answer restated. Read the page a crawler gets and you are measuring construction. Render it in Chrome first and you are measuring what a browser-driving agent can do. Same URL, two artifacts.

Primary sources: Lighthouse agentic browsing scoring · is-agentic about page · Ora on the Vercel partnership · Vercel's launch post · DebugBear on the new category · Hermes Clew · Perseus Clew, the Agentis Lux engine

I did not place. Twice.

Not with Hermes, not with Agentis Lux.

I sat with that for a while before I could say anything useful about it, because I was bummed. I had an idea, I executed it, and to me all my builds are winners, lol. That is not a judging criterion, unfortunately.

The first read was that the idea did not resonate. I do not think that is what happened.

A hackathon judges a demo. A demo rewards a visible before and after, a novelty you can see in one glance, and a tight fit to the prompt. "Findings only, no fixes, no ranking" is a position. Positions do not demo. Five DynamoDB tables with two TTLs, 770 passing tests, and a tool that scans its own site to a 96 and publishes the one finding it has left, none of that shows up in a three minute video.

Vercel's launch leads with one-click fix prompts. That is the demoable thing. It was always the demoable thing. I removed it on purpose and then submitted to a format that pays for it.

So: the idea was fine, the packaging was not, and I built the same tool twice without noticing I had optimized it away from the room I kept walking into. That is a submission problem. It is a useful one to name, because it is fixable and "nobody liked my idea" is not.

Agentis Lux is shelved for now. It is live, it is Apache 2.0, the benchmark data and the known limitations are in the repo, and there is a STATUS file explaining where I stopped so future me does not have to reconstruct it.

I am not sorry I removed the fixes.

I am also not certain I am right, and I would like to hear from people who landed somewhere else. Vercel shipped one-click fix prompts and the ecosystem picked up the name within weeks. I shipped findings only. If you ran one of these on your own site, which would you want back, the fact or the fix? I have been wrong about my own tool before.

Quick context if you are new here. I work in the California courts, running court operations for the county. I started building with AI in July 2025 and I have been learning in public ever since. I do not write the code. I direct, the agents generate, I validate and decide. I build the Clew Suite, a set of civic tech tools for making complex systems easier to inspect. That is the lens I am writing from.

AI Assisted. Human Approved. Powered by NLP.