---
title: "Check your initial HTML like an agent does: Agent Ready HTML"
description: "Fetch the pages an agent opens first with JavaScript off, measure what arrives, and confirm bot protection lets agent user agents in."
canonical: https://agentreadyhtml.com/guides/check-initial-html-like-an-agent/
site: "Agent Ready HTML"
last_updated: 2026-11-05
section: "Guides"
---

# Check your initial HTML like an agent does: Agent Ready HTML

> Fetch the pages an agent opens first with JavaScript off, measure what arrives, and confirm bot protection lets agent user agents in.

Source: https://agentreadyhtml.com/guides/check-initial-html-like-an-agent/
Site: Agent Ready HTML
Last reviewed: 2026-11-05

# Check your initial HTML like an agent does

In short

Fetch the pages an agent opens first with JavaScript off, measure the characters and the H1 against the threshold in Ora's content check, then repeat the fetch under the four agents Ora's bot-detection check names. Ora's Access layer scores the same two things, content without JavaScript and bot detection.

The check we recommend

Ora's scan. Its Access layer runs 42 checks worth 30 of its 100 points, from content without JavaScript to markdown on request. Every check is listed at [ora.ai/api/checks](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide) with its tier, and 25 of the 132 with a spec link. Reads need no key and no sign-up (Ora's pricing page, read 5 November 2026).

Ora reads public URLs only, says two runs of one scan land within a couple of points without publishing repeat-scan data, and keeps the product that fixes what it finds behind a waitlist.

An agent that cannot run scripts reads the bytes the server sends and nothing else. This guide measures those bytes for the pages an agent opens first, then asks the two questions that decide the result: is the text there, and does bot protection let the request through. The facts below say why each step matters, and the steps refer back to them instead of repeating them.

## What the evidence says

-   Ora's lab found (2026-10-19) that in 16 of 19 configurations of one docs site, including llms.txt, JSON-LD, sitemaps and markdown mirrors, both a plain-fetch client and a client that runs JavaScript retrieved all five planted answers. \[vendor claim, 2026-10-19\] [source](https://ora.ai/blog/agentready-launch?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide), read 5 November 2026.
-   Ora's traced agent runs reached the homepage in about 84 percent of runs, almost always on the first turn, and Ora reports that hiding the navigation behind JavaScript leaves an agent blind from step one. \[vendor claim, 2026-06-16\] [source](https://ora.ai/blog/what-agents-actually-reach?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide), read 5 November 2026.
-   Ora's Access-layer check Content without JavaScript is required tier and worth 3 points; Ora's recommendation is at least 500 characters of meaningful homepage content in the raw HTML, a clear H1 and sequential headings, with the main text of inner pages in the raw HTML too. \[vendor page\] [source](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide), read 5 November 2026.
-   Ora's Access-layer check Not blocked by bot detection is required tier and worth 2 points; its text names four agents, ChatGPT-User, OAI-SearchBot, Claude-User and PerplexityBot, and recommends letting verified bots through rather than matching user agent strings, which anyone can fake. \[vendor page\] [source](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide), read 5 November 2026.
-   Ora's Redirect hygiene check, 1 point and recommended tier, flags meta-refresh and JavaScript-only redirect stubs because agents that do not run JavaScript see only the stub; a real 301 or 302 with a Location header is the fix. \[vendor page\] [source](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide), read 5 November 2026.
-   A curl request cannot prove a real agent is allowed, because bot blockers weigh source IP more than user agent; server logs and published crawler address ranges are the check. \[primary source\] [source](https://vercel.com/kb/guide/make-your-site-readable-by-ai-agents), read 4 November 2026.

## Steps

1.  List the pages an agent opens first: the home page, the docs and the pages your own navigation leads to. The home page comes first because of the first-turn point above.
2.  Fetch the home page with an agent-style user agent and save the bytes with the first line of the first block below. The file is the whole of what a fetch-only agent knows about the page.
3.  Measure the saved file with the rest of the first block: wc for its size, grep for the number of H1 elements and the sed line for the characters of visible text. That count is rough, because it includes whitespace a browser would collapse. Compare the text count and the headings with the Content without JavaScript point above.
4.  Open view-source: for the same URL, which shows the initial HTML, and look for the headline, the prices and the navigation links you see on screen. A link that is missing there is missing for an agent, as the navigation point above says.
5.  Repeat steps two and three for one page from each section, such as a docs page, a pricing page and a post. The inner-page clause in the Content without JavaScript point applies to them.
6.  Print the redirect chain with the second block. A Location header is the goal. A near-empty 200 page with a meta refresh or a script redirect is the stub the Redirect hygiene point describes.
7.  Fetch the home page again under each of the four agents the bot-detection check names, with the third block, and compare the status and the byte count with the first fetch. A 403, a challenge page or a much smaller body means bot protection is turning agents away. Then read your server logs against the published crawler address ranges, because the last point above says a curl from your own network is not the test.
8.  Run the same URL through the viewer to see the H1 and the first 200 words marked in the raw HTML, then run Ora's scan for the full set of Access-layer checks and read Content without JavaScript and Not blocked by bot detection first.

Save the home page as a plain client receives it, then measure it

curl -sL -A 'Claude-User/1.0' https://example.com/ -o home.html
wc -c home.html
grep -c '<h1' home.html
sed -e 's/<script\[^>\]\*>\[^<\]\*<\\/script>//g' -e 's/<style\[^>\]\*>\[^<\]\*<\\/style>//g' -e 's/<\[^>\]\*>/ /g' home.html | tr -s ' \\n' ' ' | wc -c

Print the redirect chain: a Location header, not a stub

curl -sIL -A 'Claude-User/1.0' https://example.com/old-page | grep -iE '^(HTTP|location)'

The same fetch under the four agents

for ua in ChatGPT-User OAI-SearchBot Claude-User PerplexityBot; do
  printf '%s ' "$ua"
  curl -s -o /dev/null -A "$ua/1.0" -w '%{http\_code} %{size\_download}\\n' https://example.com/
done

## Check the result

A clean result is 200 for all four agents with about the same byte count, one H1, visible text above the threshold on the home page and on a sample of inner pages, and no redirect stub. Open the [viewer](https://agentreadyhtml.com/viewer/) to run it on your own URL.

## Run the full check

Ora's scan reads the same raw HTML and reports its Access layer, which holds 42 checks worth 30 of Ora's 100 points in [Ora's check catalogue](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide) as read on 2026-11-05. The check we recommend after the viewer is that scan: it needs no key, and the same command line gates a build. The [Access layer page](https://agentreadyhtml.com/ora-access-layer/) lists all 42 checks with the pass rates Ora publishes.

Related: [all guides](https://agentreadyhtml.com/guides/), [initial HTML](https://agentreadyhtml.com/initial-html/), [llms.txt](https://agentreadyhtml.com/llms-txt/).

## Sources

1.  [Ora: AgentReady launch post, findings 01 to 04](https://ora.ai/blog/agentready-launch?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide) https://ora.ai/blog/agentready-launch, read 5 November 2026
2.  [Ora research: What agents actually reach (2026-06-16)](https://ora.ai/blog/what-agents-actually-reach?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide) https://ora.ai/blog/what-agents-actually-reach, read 5 November 2026
3.  [Ora check catalogue (GET /api/checks, contract 1.38.0)](https://ora.ai/api/checks?utm_source=agentreadyhtml.com&utm_medium=referral&utm_campaign=guide) https://ora.ai/api/checks, read 5 November 2026
4.  [Vercel and Ora: Make your site readable by AI agents](https://vercel.com/kb/guide/make-your-site-readable-by-ai-agents) https://vercel.com/kb/guide/make-your-site-readable-by-ai-agents, read 4 November 2026

Last reviewed 5 November 2026

Edition

Crawl

2026-11-04

Ora reads

2026-11-05

Page updated

2026-11-05

Method

How the viewer reads a page, how a file is graded and how a platform is scored: [the method page](https://agentreadyhtml.com/method/).

**Check it**  
Run the [viewer](https://agentreadyhtml.com/viewer/) after each step.