<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: OnePassApps</title>
    <description>The latest articles on DEV Community by OnePassApps (@onepassapps).</description>
    <link>https://dev.to/onepassapps</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4222432%2F971441a4-25b9-4a1f-9985-337cdf95d706.png</url>
      <title>DEV Community: OnePassApps</title>
      <link>https://dev.to/onepassapps</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/onepassapps"/>
    <language>en</language>
    <item>
      <title>The Card Title Is the Target Guide's Own Heading: 39 of 39, Word for Word</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 09:19:55 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-card-title-is-the-target-guides-own-heading-39-of-39-word-for-word-466d</link>
      <guid>https://dev.to/onepassapps/the-card-title-is-the-target-guides-own-heading-39-of-39-word-for-word-466d</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com. This is the third post on the "Related guides" section that closes all 22 of our app guides. The first measured the section's shape, the second measured the graph its links make. This one reads what is written on the card.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I measured
&lt;/h2&gt;

&lt;p&gt;2026-11-12: all 22 guide pages and all 22 app pages of app.onepassapps.com, fetched over plain HTTPS, signed out, no cookies, no JavaScript, GET only. Every page answered HTTP 200. On each guide page I parsed the Related guides section and, for every card in it, the tool name, the category label, the title link, the description paragraph and the Open button.&lt;/p&gt;

&lt;p&gt;The section holds 39 cards in total: 17 guides list two other tools and 5 list one.&lt;/p&gt;

&lt;h2&gt;
  
  
  A card carries five written fields, and all five come from the target
&lt;/h2&gt;

&lt;p&gt;Here are three of the cards, with every field each one carries:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;On this guide&lt;/th&gt;
&lt;th&gt;Tool name&lt;/th&gt;
&lt;th&gt;Category label&lt;/th&gt;
&lt;th&gt;Title link text&lt;/th&gt;
&lt;th&gt;Description&lt;/th&gt;
&lt;th&gt;Open button&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;newsflow&lt;/td&gt;
&lt;td&gt;Browser Teleprompter&lt;/td&gt;
&lt;td&gt;Writing / Speech / Local utility&lt;/td&gt;
&lt;td&gt;How to read a script with a browser teleprompter&lt;/td&gt;
&lt;td&gt;Paste a script, adjust scrolling speed and text size, mirror...&lt;/td&gt;
&lt;td&gt;Open Browser Teleprompter&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;movie-tracker&lt;/td&gt;
&lt;td&gt;Screen &amp;amp; Cinema Planner&lt;/td&gt;
&lt;td&gt;Personal / Media / Planning&lt;/td&gt;
&lt;td&gt;How to compare screen dimensions and viewing angles&lt;/td&gt;
&lt;td&gt;Compare two TV or cinema screen sizes using diagonal, aspect...&lt;/td&gt;
&lt;td&gt;Open Screen &amp;amp; Cinema Planner&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;paperkit&lt;/td&gt;
&lt;td&gt;CV Builder&lt;/td&gt;
&lt;td&gt;Writing / Documents / Careers&lt;/td&gt;
&lt;td&gt;How to build a CV and print or save a résumé as PDF&lt;/td&gt;
&lt;td&gt;Enter your own résumé details, choose Classic, Modern or Edi...&lt;/td&gt;
&lt;td&gt;Open CV Builder&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Three things in those rows are worth pulling apart.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The title link text is the target guide's own heading, word for word.&lt;/strong&gt; For every one of the 39 cards, the text of the title link is byte-identical to the &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; of the guide it points at: 39 of 39. It is also a prefix of that guide's document title in 39 of 39 cases, because every guide title is its own heading followed by the same suffix - a space, an em dash, a space and the word OnePassApps.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The name and the button come from the target app page.&lt;/strong&gt; The card's name is the head of the target app page's own &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; in 39 of 39 cards, and the button text is "Open " plus that same name in 39 of 39, pointing at the same tool as the title link in 39 of 39.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The description is shared by every card that names the same tool.&lt;/strong&gt; The 39 cards name 18 distinct tools, and each of those tools carries one description string wherever it appears - 18 distinct description strings across 39 cards. The description is written once per tool and repeated on every card that names it.&lt;/p&gt;

&lt;h2&gt;
  
  
  This corrects what I wrote on 8 November
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://dev.to/onepassapps/the-related-guides-graph-39-links-between-22-guides-one-hub-and-four-guides-nothing-points-at"&gt;An earlier post here&lt;/a&gt; measured this same section and made the graph it makes. It concluded, from two examples, that "the sentence on a card is written for that card, not pulled from a shared field on the target page", and the two sentences it quoted were "How to merge and organise PDF pages in your browser" and "How to build a CV and print or save a résumé as PDF".&lt;/p&gt;

&lt;p&gt;Both of those strings are the card's title link, and both are the target guide's own &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt;. So the title link is not written for the card - it is the heading the target page already prints, reused as the link text.&lt;/p&gt;

&lt;p&gt;The description under it is the other field on the card, and that one is shared by every card naming the same tool, which is why the two descriptions in a reciprocal pair differ from each other: they are two different tools' descriptions, not two different sentences written for two cards.&lt;/p&gt;

&lt;p&gt;The graph itself is unchanged and that post remains the canonical write-up for it: 39 edges across 22 guides, 26 of them reciprocated, in 13 pairs. This post is about what the words on the card say, not about how the links connect.&lt;/p&gt;

&lt;h2&gt;
  
  
  The app links on a guide page
&lt;/h2&gt;

&lt;p&gt;Each guide page also carries exactly two links to its own app page - 44 across the 22 guides - plus one Open button per card, so guide pages carry 83 links to app pages in total. A guide page links to its own guide 0 times.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not do
&lt;/h2&gt;

&lt;p&gt;One build, one day, one network, plain HTTPS, GET only, signed out, no cookies and no JavaScript, so this measures the documents the server sends and not the rendered pages. I did not click a link and I did not open an app, so I am claiming only that the served documents carry these fields and these words. A card or a label added by JavaScript would not appear in this count.&lt;/p&gt;

&lt;p&gt;If you want to walk the apps themselves, they are here: &lt;a href="https://app.onepassapps.com/apps/label-maker?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=card-text-layer&amp;amp;utm_content=label-maker-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/apps/label-maker?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=card-text-layer&amp;amp;utm_content=label-maker-link&lt;/a&gt; - and the guides that carry the cards are here: &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=card-text-layer&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=card-text-layer&amp;amp;utm_content=guides-link&lt;/a&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>html</category>
      <category>seo</category>
      <category>documentation</category>
    </item>
    <item>
      <title>The Signature Preview Is a Mock Message: To, Subject, and a Dark Background That Only Approximates Real Mail</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 07:19:28 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-signature-preview-is-a-mock-message-to-subject-and-a-dark-background-that-only-approximates-21p1</link>
      <guid>https://dev.to/onepassapps/the-signature-preview-is-a-mock-message-to-subject-and-a-dark-background-that-only-approximates-21p1</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com. This is a measurement of the Email Signature Builder's preview - the part of the panel where you see the signature before you copy it. I measured it in the visible Chrome on 12 November 2026, signed out, on one build.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I measured
&lt;/h2&gt;

&lt;p&gt;I opened &lt;a href="https://app.onepassapps.com/apps/email-signature" rel="noopener noreferrer"&gt;https://app.onepassapps.com/apps/email-signature&lt;/a&gt; and read the preview area as it came up, signed out, with the example details the app loads.&lt;/p&gt;

&lt;h2&gt;
  
  
  The preview is a message, not a signature
&lt;/h2&gt;

&lt;p&gt;The preview area prints, in this order: &lt;strong&gt;Signature&lt;/strong&gt;, &lt;strong&gt;Light message&lt;/strong&gt;, &lt;strong&gt;Copy signature&lt;/strong&gt;, &lt;strong&gt;New message&lt;/strong&gt;, then &lt;strong&gt;To&lt;/strong&gt; and &lt;strong&gt;Subject&lt;/strong&gt;, then the line &lt;strong&gt;Example loaded. Add your details.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;So the preview is not just the signature on a blank page. Reading the panel's own text in order, the preview area prints the signature, a message header of New message, To and Subject, and a Copy signature button next to it.&lt;/p&gt;

&lt;p&gt;Three tabs sit above that preview area: &lt;strong&gt;Preview&lt;/strong&gt;, &lt;strong&gt;Plain text&lt;/strong&gt; and &lt;strong&gt;HTML source&lt;/strong&gt; - the same three copies the export step produces, reachable without leaving the preview.&lt;/p&gt;

&lt;h2&gt;
  
  
  The dark background is labelled as an approximation
&lt;/h2&gt;

&lt;p&gt;There is a control called &lt;strong&gt;Message background&lt;/strong&gt; with two choices, &lt;strong&gt;Light&lt;/strong&gt; and &lt;strong&gt;Dark&lt;/strong&gt;. Choosing Dark is not a real dark-mode rendering, and the panel says so itself, in one sentence:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Dark approximates how dark-mode mail apps recolour a message. Send yourself a test to be sure.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;That last clause is the instruction the app gives you instead of a guarantee: send yourself a test message and look at what actually arrives. It is the honest thing for a tool to say, because a mail client is free to rewrite the HTML you pasted into it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the panel says about privacy, in one line
&lt;/h2&gt;

&lt;p&gt;Under the help text the panel prints &lt;strong&gt;Nothing leaves this tab.&lt;/strong&gt; The example contact details are editable and nothing is saved or uploaded.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not do
&lt;/h2&gt;

&lt;p&gt;One build, one browser, one signed-out session, one day, 12 November 2026. I read the preview as it loaded; I did NOT press Copy signature, so nothing here is a claim about what lands on the clipboard, and I did not send myself a test message, so I am not claiming what any mail client would do with the signature. I did not compare the Light and Dark views against a real inbox. The app's own guide says the examples help with setup but do not promise identical rendering in every client.&lt;/p&gt;

&lt;h2&gt;
  
  
  The tool and its guide
&lt;/h2&gt;

&lt;p&gt;The Email Signature Builder is here: &lt;a href="https://app.onepassapps.com/apps/email-signature?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=signature-preview&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/apps/email-signature?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=signature-preview&amp;amp;utm_content=app-link&lt;/a&gt; - and the guide that spells out the setup steps is here: &lt;a href="https://app.onepassapps.com/guides/email-signature?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=signature-preview&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/guides/email-signature?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=signature-preview&amp;amp;utm_content=guides-link&lt;/a&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>email</category>
      <category>html</category>
      <category>documentation</category>
    </item>
    <item>
      <title>Two Hash Tabs in One Studio, and One Number That Means Two Different Things</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 06:18:10 +0000</pubDate>
      <link>https://dev.to/onepassapps/two-hash-tabs-in-one-studio-and-one-number-that-means-two-different-things-4c9a</link>
      <guid>https://dev.to/onepassapps/two-hash-tabs-in-one-studio-and-one-number-that-means-two-different-things-4c9a</guid>
      <description>&lt;p&gt;I work on OnePassApps, so this is our own product and every line below was read off the live File Studio today, in the browser, on the panel itself.&lt;/p&gt;

&lt;p&gt;The File Studio has two hash tabs next to each other. One of them will give you a SHA-512 checksum. The other one will not give you a SHA-512 anything, and it will not tell you that either - it just names one algorithm and moves on.&lt;/p&gt;

&lt;h2&gt;
  
  
  The two tabs, read off the live panel
&lt;/h2&gt;

&lt;p&gt;Both live under &lt;code&gt;app.onepassapps.com/studios/files&lt;/code&gt;. I opened each tab and read the copy.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;File hash&lt;/strong&gt; reads:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;01 Source
02 Algorithm
SHA-256
SHA-512
SHA-256 is the common choice for published checksums.
03 Result
The checksum appears on the stage, ready to copy and compare.
Nothing leaves this tab.
Calculate SHA-256
One file of any type, up to 2 GiB · read on this device
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;Find duplicates&lt;/strong&gt; reads:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;01 Source
0 files · 0 B. Only the files you choose are compared.
02 Comparison
Files
0 of 500
Total size
0 B of 2 GiB
Compared by
SHA-256 of the content
03 Result
Groups of identical files are listed on the stage. Nothing is deleted.
Nothing leaves this tab.
Find exact duplicates
Up to 500 files and 2 GiB in total · compared by their content, never moved or deleted
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  The same number, two different ceilings
&lt;/h2&gt;

&lt;p&gt;Both tabs say &lt;strong&gt;2 GiB&lt;/strong&gt;. They do not mean the same thing by it.&lt;/p&gt;

&lt;p&gt;In &lt;strong&gt;File hash&lt;/strong&gt; the 2 GiB is a ceiling on one file: "One file of any type, up to 2 GiB". The panel takes a single file and returns a single checksum.&lt;/p&gt;

&lt;p&gt;In &lt;strong&gt;Find duplicates&lt;/strong&gt; the 2 GiB is a ceiling on the whole set: "Up to 500 files and 2 GiB in total". The panel takes up to 500 files, and 2 GiB is the ceiling on them added together, not on each one.&lt;/p&gt;

&lt;p&gt;So on the second tab you can compare 500 files, but only while their combined size stays under 2 GiB. That is about 4 MiB per file if you use the full file count, and it is a ceiling on the batch, not a per-file allowance you can spend 500 times.&lt;/p&gt;

&lt;h2&gt;
  
  
  The algorithm difference nobody is told about
&lt;/h2&gt;

&lt;p&gt;This is the part I would have missed without opening both panels.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;File hash&lt;/strong&gt; gives you a choice. Its second step is a picker with two options, &lt;code&gt;SHA-256&lt;/code&gt; and &lt;code&gt;SHA-512&lt;/code&gt;, plus a line explaining the default: "SHA-256 is the common choice for published checksums." The action button even changes with the selection - it reads &lt;strong&gt;Calculate SHA-256&lt;/strong&gt; when SHA-256 is picked.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Find duplicates&lt;/strong&gt; gives you no choice and no explanation. Its second step is called &lt;strong&gt;Comparison&lt;/strong&gt;, not &lt;strong&gt;Algorithm&lt;/strong&gt;, and its third line reads "Compared by SHA-256 of the content". There is no SHA-512 option in that panel, and there is no sentence saying why there is not one.&lt;/p&gt;

&lt;p&gt;So if the checksum you need to match is a SHA-512, the duplicate finder cannot produce it. It groups by SHA-256 or it does not group at all. The panel does not refuse you, does not warn you and does not mention SHA-512 - it simply states the one algorithm it uses.&lt;/p&gt;

&lt;p&gt;That is a real difference between the two tools sitting in the same tab strip, and the copy on the page is the only place you would learn it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the duplicate finder does and does not claim
&lt;/h2&gt;

&lt;p&gt;Two lines in that panel are worth reading carefully, because they are narrower than the tab name suggests.&lt;/p&gt;

&lt;p&gt;"Groups of identical files are listed on the stage. Nothing is deleted." - identical, and nothing is deleted.&lt;/p&gt;

&lt;p&gt;"Up to 500 files and 2 GiB in total · compared by their content, never moved or deleted" - compared by content, so a renamed copy still groups, and never moved or deleted, so the tool only reports.&lt;/p&gt;

&lt;p&gt;Both of those are the tool's own words, and they are the reason the tab is safe to run on a folder you care about: it reads, it groups, it lists, and it leaves your files exactly where they were. What it does not do is decide which copy you should keep - that is still your call, and for files that only look alike rather than being byte-identical, it will not group them at all.&lt;/p&gt;

&lt;h2&gt;
  
  
  The one thing I did not test
&lt;/h2&gt;

&lt;p&gt;I read both panels in the live app today and I did not run a SHA-512 hash or a multi-file comparison in this run, so everything above is the panel copy and the structure of the two tools, not the output of a completed run. The hashing and grouping behaviour itself is covered in our earlier write-up of the Find duplicates tab.&lt;/p&gt;

&lt;p&gt;If you need SHA-512 for a published checksum, use the File hash tab and pick it. If you need to know which of several files are byte-identical, Find duplicates does that by SHA-256 and will not do SHA-512 at all.&lt;/p&gt;

&lt;p&gt;You can try both in the &lt;a href="https://app.onepassapps.com/studios/files?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=two-hash-tabs&amp;amp;utm_content=file-studio-link" rel="noopener noreferrer"&gt;File Studio&lt;/a&gt;. Every app we make has a guide page as well, and they are all &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=two-hash-tabs&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;listed here&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>security</category>
      <category>files</category>
      <category>javascript</category>
    </item>
    <item>
      <title>The Link From an App Page to Its Own Guide: 22 of 22 Use the Guide's Own Title</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 05:28:28 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-link-from-an-app-page-to-its-own-guide-22-of-22-use-the-guides-own-title-4a41</link>
      <guid>https://dev.to/onepassapps/the-link-from-an-app-page-to-its-own-guide-22-of-22-use-the-guides-own-title-4a41</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com - so this is our own site, and every count below was measured on it today, 12 November 2026, signed out, over plain HTTPS, with no cookies and no JavaScript.&lt;/p&gt;

&lt;p&gt;We publish two pages for every app: an &lt;code&gt;/apps/&lt;/code&gt; page and a &lt;code&gt;/guides/&lt;/code&gt; page. The app page is the tool itself; the guide page is the written walkthrough of the same tool.&lt;/p&gt;

&lt;p&gt;The app page carries exactly one link to its guide, and the text of that link is built from the guide's own title. I checked all 22 apps today, so 22 app pages and 22 guide pages.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;App pages that link to their own guide exactly once: &lt;strong&gt;22 of 22&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;App pages whose link text contains their guide's own &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; word for word: &lt;strong&gt;22 of 22&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;App pages whose link text is exactly &lt;code&gt;Read the &amp;lt;guide title&amp;gt; guide&lt;/code&gt;: &lt;strong&gt;22 of 22&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Links from an app page to any &lt;em&gt;other&lt;/em&gt; guide: &lt;strong&gt;0&lt;/strong&gt; across the 22 app pages.&lt;/li&gt;
&lt;li&gt;Links from an app page into &lt;code&gt;/guides/&lt;/code&gt; in total: &lt;strong&gt;22&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The guide's &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; is reused as the link text, not reworded. Here is what that looks like on two of the 22:&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;ambient-noise-mixer&lt;/strong&gt; - the guide's &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; is "How to mix white, pink and brown noise with a timer" and the app page's link reads "Read the How to mix white, pink and brown noise with a timer guide".&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;paperkit&lt;/strong&gt; - the guide's &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; is "How to merge and organise PDF pages in your browser" and the app page's link reads "Read the How to merge and organise PDF pages in your browser guide".&lt;/p&gt;

&lt;h2&gt;
  
  
  How I read it
&lt;/h2&gt;

&lt;p&gt;For each of the 22 app pages I took every anchor whose &lt;code&gt;href&lt;/code&gt; points into &lt;code&gt;/guides/&lt;/code&gt;, and for each of the 22 guide pages I took the &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; element and the &lt;code&gt;&amp;lt;title&amp;gt;&lt;/code&gt; element out of the served HTML.&lt;/p&gt;

&lt;p&gt;A link counts as naming its guide when the guide's &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; string appears inside the link's visible text. The comparison is exact - the same characters, the same order - so a reworded or truncated title would not count.&lt;/p&gt;

&lt;p&gt;On the 22 guide pages, the document &lt;code&gt;&amp;lt;title&amp;gt;&lt;/code&gt; starts with that same &lt;code&gt;&amp;lt;h1&amp;gt;&lt;/code&gt; string on &lt;strong&gt;22 of 22&lt;/strong&gt; pages, with the site name appended after it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the numbers look like
&lt;/h2&gt;

&lt;p&gt;The link text runs from &lt;strong&gt;55 to 74 characters&lt;/strong&gt; across the 22 apps, and the guide titles it is built from run from &lt;strong&gt;40 to 59 characters&lt;/strong&gt;. The link is the shortest path to the guide, and it carries the guide's own name rather than a summary of it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why it is worth knowing
&lt;/h2&gt;

&lt;p&gt;A reader who lands on an app page and clicks through to the guide meets the same name for the same tool on both pages. That is the one place the two pages hand off, and the handoff does not rename anything.&lt;/p&gt;

&lt;p&gt;If we ever rewrote one of these link texts - "ambient-noise-mixer's app page calling its guide something else" - the guide's own title would stop matching the way people arrive at it, and the two pages would disagree about the name of the same tool. This measurement is the check for that.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not do
&lt;/h2&gt;

&lt;p&gt;I read the served HTML of the 22 app pages and the 22 guide pages. I did not open the interactive app, and I did not click the links. I measured what the server sends, not what a reader sees after JavaScript runs, and the byte counts of a build move.&lt;/p&gt;

&lt;p&gt;The guide pages have their own in-page anchors - &lt;code&gt;#features&lt;/code&gt;, &lt;code&gt;#steps&lt;/code&gt;, &lt;code&gt;#limits&lt;/code&gt;, &lt;code&gt;#questions&lt;/code&gt; - which a &lt;a href="https://dev.to/onepassapps/the-twin-pages-of-apponepassappscom-22-app-pages-22-guide-pages-and-the-214648-bytes-between-4hkp"&gt;previous post of mine&lt;/a&gt; measured. This post is about the link that crosses from the app page to the guide, not about the anchors inside the guide.&lt;/p&gt;

&lt;p&gt;Try it yourself - the ambient-noise-mixer app page and the ambient-noise-mixer guide are a pair you can compare directly:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://app.onepassapps.com/apps/ambient-noise-mixer?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=guide-link-text&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;ambient-noise-mixer app page&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://app.onepassapps.com/guides/ambient-noise-mixer?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=guide-link-text&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;ambient-noise-mixer guide&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://app.onepassapps.com/apps?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=guide-link-text&amp;amp;utm_content=apps-link" rel="noopener noreferrer"&gt;The guide link text on all 22 app pages&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>webdev</category>
      <category>html</category>
      <category>seo</category>
      <category>documentation</category>
    </item>
    <item>
      <title>Seven Studio Pages, No Limits of Their Own: The Numbers Are in the Tools, Not the Menu</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 05:12:17 +0000</pubDate>
      <link>https://dev.to/onepassapps/seven-studio-pages-no-limits-of-their-own-the-numbers-are-in-the-tools-not-the-menu-35dd</link>
      <guid>https://dev.to/onepassapps/seven-studio-pages-no-limits-of-their-own-the-numbers-are-in-the-tools-not-the-menu-35dd</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com. Seven pages on our site are Studio landing pages, one per Studio. Each one lists the tools inside its Studio and then hands you over to them, and this post is about the thing they do NOT tell you: not one of the seven states a single limit of its own. Measured on 12 November 2026 from outside the browser, signed out, with no cookies and no JavaScript.&lt;/p&gt;

&lt;h2&gt;
  
  
  The handoff, in the page's own words
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;"Processing stays in the current browser tab; export your results before leaving."&lt;/li&gt;
&lt;li&gt;"Supported formats and device limits are shown by each tool."&lt;/li&gt;
&lt;li&gt;"No account is required for browser-local processing."&lt;/li&gt;
&lt;li&gt;"Enable JavaScript to use this Studio."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That is the deal these pages make. They say where the work happens and they say the formats and the limits belong to the individual tools - and then they keep the promise completely. The line is on all seven pages, not one:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Processing stays in the current browser tab - on 7 of the 7 Studio pages&lt;/li&gt;
&lt;li&gt;Supported formats and device limits are shown by each tool - on 7 of the 7 Studio pages&lt;/li&gt;
&lt;li&gt;No account is required for browser-local processing - on 7 of the 7 Studio pages&lt;/li&gt;
&lt;li&gt;Enable JavaScript to use this Studio - on 7 of the 7 Studio pages&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What the seven pages actually contain
&lt;/h2&gt;

&lt;p&gt;I searched each served page for a limit stated in numbers - a file size, a page count, a file count, a character count - and for the words that usually come with one. Both searches run over the page's visible text, with the script and style blocks removed, so the counts below are what a reader sees and not what the inline code happens to contain:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Studio page&lt;/th&gt;
&lt;th&gt;Served bytes&lt;/th&gt;
&lt;th&gt;Limits stated as numbers&lt;/th&gt;
&lt;th&gt;Cap words&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;audio&lt;/td&gt;
&lt;td&gt;17,114&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;colour&lt;/td&gt;
&lt;td&gt;17,201&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;data&lt;/td&gt;
&lt;td&gt;17,367&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;files&lt;/td&gt;
&lt;td&gt;17,224&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;icons&lt;/td&gt;
&lt;td&gt;17,286&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;image&lt;/td&gt;
&lt;td&gt;17,208&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;text&lt;/td&gt;
&lt;td&gt;17,150&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Across all seven Studio pages that is 0 limits stated as numbers. The one cap word on each page is the 'limits' in the promise line above - seven in total, one per page. The pages name the idea of a limit and then leave the numbers to the tools.&lt;/p&gt;

&lt;p&gt;For contrast, the same search over the 22 guide pages found a limit stated in numbers on 11 of 22 of them. The guides are where our numbers live; the Studio landing pages are where you find out which tool to open. We have written the guides' own limit sentences up separately - that write-up is the canonical one: &lt;a href="https://dev.to/onepassapps/every-app-page-repeats-its-guides-limits-word-for-word-44-sentences-across-22-routes-7pl"&gt;https://dev.to/onepassapps/every-app-page-repeats-its-guides-limits-word-for-word-44-sentences-across-22-routes-7pl&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Why that is the right shape, and where it bites
&lt;/h2&gt;

&lt;p&gt;It is a clean split: one page per Studio to choose from, one guide per app for the numbers. The catch is the order of reading. A Studio page cannot tell you that an import is capped or that an export has an output-size limit, because it does not know yet which tool you will open - so if you read the Studio page and then the tool, you have to read the tool's own limits before you commit. The seven Studio pages are not summaries of the tools; they are menus.&lt;/p&gt;

&lt;p&gt;We have published the shell of these seven pages before - the tool names and the fact that the pages carry no links are already written up: &lt;a href="https://dev.to/onepassapps/seven-studio-shells-49-tool-names-and-zero-links-what-apponepassappscomstudios-sends-before-1plh"&gt;https://dev.to/onepassapps/seven-studio-shells-49-tool-names-and-zero-links-what-apponepassappscomstudios-sends-before-1plh&lt;/a&gt;. This post is about the one thing on those pages that is not a name and not a link: the limits they decline to state.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not do
&lt;/h2&gt;

&lt;p&gt;One host, one day, one network, plain HTTPS, GET only, signed out, no cookies and no JavaScript, so this is a measurement of what the server sent, not of the rendered page, and no tool panel was opened. A limit that only appears after JavaScript runs would not show up here. The byte counts move with the build. The search patterns are mine, not the app's: a page that stated 'under 25 megabytes' in words rather than digits would be counted by my cap-word search but not by my number search, which is why both columns are in the table.&lt;/p&gt;

&lt;p&gt;If you want the numbers, the guides are where they are: &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=studio-limits-handoff&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=studio-limits-handoff&amp;amp;utm_content=guides-link&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;And the Colour Studio is the one to open first: &lt;a href="https://app.onepassapps.com/studios/colour?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=studio-limits-handoff&amp;amp;utm_content=studio-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/studios/colour?utm_source=devto&amp;amp;utm_medium=blog&amp;amp;utm_campaign=studio-limits-handoff&amp;amp;utm_content=studio-link&lt;/a&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>browser</category>
      <category>documentation</category>
      <category>design</category>
    </item>
    <item>
      <title>Every App Page Repeats Its Guide's Limits Word for Word: 44 Sentences Across 22 Routes</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 04:54:14 +0000</pubDate>
      <link>https://dev.to/onepassapps/every-app-page-repeats-its-guides-limits-word-for-word-44-sentences-across-22-routes-7pl</link>
      <guid>https://dev.to/onepassapps/every-app-page-repeats-its-guides-limits-word-for-word-44-sentences-across-22-routes-7pl</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com - so this is our own site, and every sentence below was measured on it today, 12 November 2026, signed out, over plain HTTPS, with no cookies and no JavaScript.&lt;/p&gt;

&lt;p&gt;We publish two pages for every app: an &lt;code&gt;/apps/&lt;/code&gt; page and a &lt;code&gt;/guides/&lt;/code&gt; page. Both carry a limits section - the app page calls it &lt;strong&gt;Limitations&lt;/strong&gt;, the guide calls it &lt;strong&gt;Limits and privacy&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;A &lt;a href="https://dev.to/onepassapps/the-twin-pages-of-apponepassappscom-22-app-pages-22-guide-pages-and-the-214648-bytes-between-4hkp"&gt;previous post here&lt;/a&gt; measured the pair itself - the headings, the meta descriptions and the bytes between them. This post measures one thing the earlier one did not: whether the two limits sections say the same thing.&lt;/p&gt;

&lt;p&gt;I checked all 22 apps, so 44 pages, and counted the sentences in those two sections.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;App pages: 44 limitation sentences, across 22 routes.&lt;/li&gt;
&lt;li&gt;Guide pages: 44 limit sentences, across 22 routes.&lt;/li&gt;
&lt;li&gt;Sentences that appear in &lt;strong&gt;both&lt;/strong&gt; places, word for word: &lt;strong&gt;44&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Sentences the app page says that its own guide does not repeat: &lt;strong&gt;0&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Sentences the guide says that its own app page does not repeat: &lt;strong&gt;0&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;In other words: for every one of the 22 routes, every limitation sentence on the app page is repeated in that route's own guide. There is not one app-only warning on the site, and not one guide sentence is missing from its app page either.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I counted
&lt;/h2&gt;

&lt;p&gt;One list item or one paragraph per sentence, taken from the served HTML of each page, and I only counted a block when it was longer than 40 characters so that stray labels could not pass as a sentence. Matching was done on the text with punctuation removed and the case lowered, so a sentence counts as identical only when the words are the same and in the same order.&lt;/p&gt;

&lt;p&gt;The count is 44 sentences per side and 44 of them match.&lt;/p&gt;

&lt;h2&gt;
  
  
  A worked example
&lt;/h2&gt;

&lt;p&gt;Here are 2 of the limits sentences from the Label &amp;amp; Card Maker app page - each one is repeated, unchanged, in the guide for the same app:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;Inputs are limited to 1 MiB, entries to 1,000 characters, and output to 500 labels including copies. Unreadable text overflow blocks output.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Printer drivers and non-printable regions vary. The preview cannot guarantee stock alignment, validate postal addresses or generate barcodes. Save output before leaving or reloading.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;And the same sentences appear, unchanged, in the guide for the same app.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why that is worth knowing
&lt;/h2&gt;

&lt;p&gt;A reader who lands on an app page and a reader who lands on the guide are reading the same set of boundaries. If we ever want the two pages to say different things - a shorter warning on the app page, a longer one in the guide - this measurement is the check that says the two have drifted apart.&lt;/p&gt;

&lt;p&gt;Right now they have not: 0 app-only sentences and 0 guide-only sentences across 22 routes.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it yourself
&lt;/h2&gt;

&lt;p&gt;The Label &amp;amp; Card Maker app page and the guide for the same route are a pair you can compare directly:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://app.onepassapps.com/apps/label-maker?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=limits-pair&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;Label &amp;amp; Card Maker app page&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=limits-pair&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;All 22 app guides&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Every claim here is a count over the pages as they are served today. If a page changes, the count changes with it.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>documentation</category>
      <category>browser</category>
      <category>seo</category>
    </item>
    <item>
      <title>Split a PDF in the Browser: One File per Page in a ZIP, and What the App Does Not Do</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 04:02:28 +0000</pubDate>
      <link>https://dev.to/onepassapps/split-a-pdf-in-the-browser-one-file-per-page-in-a-zip-and-what-the-app-does-not-do-27eb</link>
      <guid>https://dev.to/onepassapps/split-a-pdf-in-the-browser-one-file-per-page-in-a-zip-and-what-the-app-does-not-do-27eb</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com - so this is our own app, and every limit below was read off our own live guide and app page today, 12 November 2026.&lt;/p&gt;

&lt;p&gt;Splitting a PDF by hand means opening it, working out which pages you want, and saving them one at a time. Doing it online usually means handing the file to a server first. There is a third way: the browser already has a PDF engine in it, so the page you are reading can do the work on the file that never leaves your machine.&lt;/p&gt;

&lt;p&gt;That is what our PDF Organiser does. Its own guide says it plainly:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;PDF Organiser works with PDFs already on your device.&lt;/p&gt;

&lt;p&gt;Merge, split, reorder, and rotate existing PDF pages locally.&lt;/p&gt;

&lt;p&gt;Your device originals remain unchanged, and you do not need an account to complete the workflow.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h2&gt;
  
  
  What it does
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Combine pages from multiple PDFs into one document in the order you choose.&lt;/li&gt;
&lt;li&gt;Select, move, rotate or remove pages using controls that also support keyboard operation.&lt;/li&gt;
&lt;li&gt;Download the arranged PDF or split selected pages into a ZIP with one PDF per page.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The third bullet is the one this post is about: the split export writes &lt;strong&gt;one PDF per page into a ZIP&lt;/strong&gt;, so a 30-page selection comes back as a ZIP of 30 single-page PDFs.&lt;/p&gt;

&lt;h2&gt;
  
  
  The three steps
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;01&lt;/strong&gt; Choose one or more PDF files.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;02&lt;/strong&gt; Select the pages to keep, reorder them and adjust rotation.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;03&lt;/strong&gt; Download the combined PDF or split-page ZIP, then open the output to verify it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The limits, as published
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Imports allow up to 20 files, 25 MiB per file, 50 MiB combined and 300 pages.&lt;/li&gt;
&lt;li&gt;Split exports allow at most 50 pages and have an output-size limit.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Those two lines matter more than they look. The split export has its own cap, and it is &lt;strong&gt;half the import page budget&lt;/strong&gt;: 300 pages in, 50 pages out of one split. And the size of the ZIP is not quoted, only bounded - the guide says it has an output-size limit without naming the number, so do not plan a split around a specific archive size.&lt;/p&gt;

&lt;p&gt;The app also states its own category on the page - Documents / PDF / Local utility - and the guide is explicit that it is a page organiser, not an editor:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;This is page organisation: it does not provide OCR, content editing, compression or secure redaction.&lt;/p&gt;

&lt;p&gt;Bookmarks, forms and digital signatures are not guaranteed to survive export.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;So there is no OCR, no text editing, no compression and no redaction here, and bookmarks, forms and signatures are not guaranteed to survive an export. It does not take photographs either: The input is existing PDF documents.&lt;/p&gt;

&lt;h2&gt;
  
  
  Your file stays in the tab
&lt;/h2&gt;

&lt;blockquote&gt;
&lt;p&gt;Document bytes stay in the app's memory.&lt;/p&gt;

&lt;p&gt;PDF Organiser runs in your browser with no account or payment details, and it never uses OnePassApps credits.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;There is no account to create and nothing to upload, so the usual privacy question - what happens to the file after I close the tab - has a short answer: nothing, because it never left.&lt;/p&gt;

&lt;p&gt;We wrote up the merge side of the same app earlier, with the page-size arithmetic: &lt;a href="https://dev.to/onepassapps/merge-pdf-pages-in-the-browser-reorder-rotate-and-export-without-uploading-56ik"&gt;https://dev.to/onepassapps/merge-pdf-pages-in-the-browser-reorder-rotate-and-export-without-uploading-56ik&lt;/a&gt; .&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;The PDF Organiser is one of the apps under one membership, and it runs in your browser: &lt;a href="https://app.onepassapps.com/apps/paperkit?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=pdf-split-zip&amp;amp;utm_content=pdf-split-link" rel="noopener noreferrer"&gt;Open the PDF Organiser&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Every app has a plain-language guide with its limits and its tool list: &lt;a href="https://app.onepassapps.com/guides/paperkit?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=pdf-split-zip&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;the OnePassApps guides&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;If you have used a split-PDF site and had to upload the document to get the pages back, this one does the same job without that step.&lt;/p&gt;

</description>
      <category>pdf</category>
      <category>javascript</category>
      <category>browser</category>
      <category>productivity</category>
    </item>
    <item>
      <title>A Collision in Our Own Per-Request Key: One of Our URLs Returned the Same Bytes Twice in 30 Fetches</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 01:56:01 +0000</pubDate>
      <link>https://dev.to/onepassapps/a-collision-in-our-own-per-request-key-one-of-our-urls-returned-the-same-bytes-twice-in-30-fetches-16j9</link>
      <guid>https://dev.to/onepassapps/a-collision-in-our-own-per-request-key-one-of-our-urls-returned-the-same-bytes-twice-in-30-fetches-16j9</guid>
      <description>&lt;p&gt;I work on OnePassApps - the team behind app.onepassapps.com - so this is our own site and our own build. Every number below was measured on it today, 12 November 2026, signed out, over plain HTTPS, with no browser and no JavaScript.&lt;/p&gt;

&lt;p&gt;On 10 November I published a post here saying that one of our URLs never returns the same bytes twice. I need to correct that. Today I fetched one of our pages thirty times in a row and two of those thirty responses were byte-for-byte identical.&lt;/p&gt;

&lt;p&gt;The 10 November post is the canonical write-up on what moves between two fetches. This post is about the exception I missed.&lt;/p&gt;

&lt;h2&gt;
  
  
  The part that still holds
&lt;/h2&gt;

&lt;p&gt;The mechanism is exactly as described then. I fetched all 56 URLs in our sitemap twice today. All 56 came back with the same byte length on the second fetch. Between the two copies of each page, 4,914 bytes differed across 404 runs - and every one of those bytes sat inside a run of long hexadecimal characters. None sat outside. So the differences live in one place, and that place is the encoded copy of our contact address.&lt;/p&gt;

&lt;p&gt;What I got wrong was the word "never".&lt;/p&gt;

&lt;h2&gt;
  
  
  The key is one byte wide, so it repeats
&lt;/h2&gt;

&lt;p&gt;Each of our pages carries our support address in Cloudflare's email-protection encoding: 42 hexadecimal characters, which is 21 bytes. The first byte is a key. The remaining 20 bytes are the address with every character XORed against that key - and 20 characters is exactly the length of the address we print in the footer.&lt;/p&gt;

&lt;p&gt;There are 256 possible keys. That is the whole supply. It is not a fresh value per response, it is a draw from a set of 256.&lt;/p&gt;

&lt;h2&gt;
  
  
  The collision
&lt;/h2&gt;

&lt;p&gt;I fetched one of our guide pages thirty times, a moment apart. The thirty responses produced 28 distinct SHA-256 digests - two digests each appeared twice. The key byte repeated twice across the thirty fetches, and in both cases the response that repeated a key was byte-for-byte identical to the earlier one with that same key.&lt;/p&gt;

&lt;p&gt;Thirty fetches give 435 pairs, and each pair has roughly a one-in-256 chance of drawing the same key, so a key spread evenly over those 256 values would be expected to collide about 1.7 times. I saw 2. The page in question stayed exactly 28,749 bytes long across all thirty fetches.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why it matters to anyone measuring a site
&lt;/h2&gt;

&lt;p&gt;A hash comparison between two responses can tell you the bytes differ. It cannot tell you the page was rebuilt - but the reverse case matters: two responses with the same digest may be two draws of the same key rather than a page that stayed still. Any pipeline watching our pages for changes will occasionally see a flat line, and it will not be able to tell a quiet page from a repeated key.&lt;/p&gt;

&lt;p&gt;So if you record a page size or a page hash once and quote it later, you are quoting one sample from a range - and occasionally the range collapses to a single value.&lt;/p&gt;

&lt;h2&gt;
  
  
  The page that did not move at all
&lt;/h2&gt;

&lt;p&gt;In the 56-page sweep, 55 pages came back with a different digest and one - the same guide page - came back byte-for-byte identical. That is the collision above, caught on a double-fetch by chance.&lt;/p&gt;

&lt;h2&gt;
  
  
  The one page I did not explain
&lt;/h2&gt;

&lt;p&gt;One page in the sweep changes far more than the rest: 1,449 bytes across 96 runs, against 42 or 84 bytes on 45 of the 56 pages. Those bytes are still all inside long hexadecimal runs, so they come from the same mechanism, but I did not analyse them and I am not claiming what they are.&lt;/p&gt;

&lt;h2&gt;
  
  
  What this does not say
&lt;/h2&gt;

&lt;p&gt;One build, one day, one network, signed out, one GET per fetch, plain HTTPS, no browser, no JavaScript, no cookies. I did not test from a second network, did not test a signed-in request, and I make no claim about how any cache or browser treats these responses. The 10 November post measured a double-fetch of the same 56 pages and found 0 that were byte-identical; a double-fetch is a small sample, and today one of them was.&lt;/p&gt;

&lt;p&gt;You can see the site at &lt;a href="https://app.onepassapps.com?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=key-collision&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=key-collision&amp;amp;utm_content=app-link&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;And the guide page where the collision showed up is at &lt;a href="https://app.onepassapps.com/guides/paperkit?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=key-collision&amp;amp;utm_content=paperkit-link" rel="noopener noreferrer"&gt;https://app.onepassapps.com/guides/paperkit?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=key-collision&amp;amp;utm_content=paperkit-link&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The 10 November post is the canonical write-up on the per-request encoding: &lt;a href="https://dev.to/onepassapps/the-same-url-never-returns-the-same-bytes-56-pages-of-apponepassappscom-and-one-per-request-key-j36"&gt;The Same URL Never Returns the Same Bytes: 56 Pages of app.onepassapps.com and One Per-Request Key&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;I work on OnePassApps - the team behind app.onepassapps.com. Corrections are better than praise.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>performance</category>
      <category>http</category>
      <category>cloudflare</category>
    </item>
    <item>
      <title>The Description Layer of Our Own llms.txt: 47 of 55 Repeat the Page's Meta Description, and 39 Bullets Carry No Link</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 01:23:48 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-description-layer-of-our-own-llmstxt-47-of-55-repeat-the-pages-meta-description-and-39-jjd</link>
      <guid>https://dev.to/onepassapps/the-description-layer-of-our-own-llmstxt-47-of-55-repeat-the-pages-meta-description-and-39-jjd</guid>
      <description>&lt;p&gt;A plain-text file at app.onepassapps.com/llms.txt is the version of our site that a language model reads. I have written about the file twice before: on 4 November, when I measured it at 6,065 bytes across 62 lines with 15 linked entries, and on 8 November about how its URL set lines up with our sitemap. This time I went through the descriptions instead of the links.&lt;/p&gt;

&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com, so everything below is a measurement of our own live site on 12 November 2026, fetched over plain HTTPS with no JavaScript and no signed-in session.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the file is now
&lt;/h2&gt;

&lt;p&gt;/llms.txt answers HTTP 200 with the content type text/plain; charset=utf-8 and 17,965 bytes across 122 lines. It carries 8 headings and 94 bullets. 55 of those bullets are markdown links - 22 app pages, 23 guide pages, 7 Studio pages and 3 platform pages - and 39 are plain text with no link in them at all.&lt;/p&gt;

&lt;p&gt;Those 55 linked URLs are exactly the 55 of our 56 sitemap URLs that are not the home page. That comparison is the 8 November post and this is not a repeat of it; the canonical write-up is &lt;a href="https://dev.to/onepassapps/the-55-of-56-what-our-sitemap-and-our-llmstxt-agree-on-and-the-one-url-only-one-of-them-names-5f6l"&gt;The 55 of 56&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The description layer: 47 of the 55 are the page's own meta description
&lt;/h2&gt;

&lt;p&gt;I fetched each of the 55 targets and read its &lt;code&gt;&amp;lt;meta name="description"&amp;gt;&lt;/code&gt; out of the served HTML, then compared it with the description llms.txt prints beside the link.&lt;/p&gt;

&lt;p&gt;47 of the 55 are byte-identical. Split by group:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;app pages: 22 of 22 identical&lt;/li&gt;
&lt;li&gt;guide pages: 22 of 23 identical&lt;/li&gt;
&lt;li&gt;Studio pages: 0 of 7 identical&lt;/li&gt;
&lt;li&gt;platform pages (plans, terms, vote): 3 of 3 identical&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So for three quarters of the file, llms.txt is not writing anything new. It repeats, word for word, the description each page already serves to a search engine.&lt;/p&gt;

&lt;h2&gt;
  
  
  The eight that differ
&lt;/h2&gt;

&lt;p&gt;Seven of them are the Studios, and they differ in one direction - llms.txt says more than the page does. The page's meta description stops at the first sentence:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Convert, compress, resize, crop and prepare images on this device.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;llms.txt keeps going:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Convert, compress, resize, crop and prepare images on this device. Tools: Convert, Compress, Resize, Crop, Remove background, Remove metadata, Images to PDF.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;That "Tools:" tail is the only place in the file where a description lists what a tool can do, and it is the whole reason the seven Studio lines do not match their pages. Across the seven Studios the tail carries 49 tool names - 7 for Image, 4 for Audio, 7 for File, 8 for Text, 11 for Data, 8 for Icon, 4 for Colour - which is the same 49 the studio registry post names, so the file carries that registry forward in prose form.&lt;/p&gt;

&lt;p&gt;The eighth goes the other way. "All app guides" says less in llms.txt: "Supported tasks, steps, formats, and limitations." against the page's own "Explore practical OnePassApps app guides, supported formats, step-by-step instructions, and tool limitations."&lt;/p&gt;

&lt;h2&gt;
  
  
  The prose layer: 39 bullets, no URLs
&lt;/h2&gt;

&lt;p&gt;The other half of the file is not links. The "Access and cost" section is 23 bullets with no URL in any of them: one for each of the 22 apps, plus one line for Browser-local Studios as a group.&lt;/p&gt;

&lt;p&gt;Twenty of those 23 are the same sentence with the app's name swapped into it - the name, then "Yes, it is free.", then the name again, then "runs in your browser with no account or payment details, and it never uses OnePassApps credits." Only three are written by hand: NewsFlow (briefs prepared at no credit cost, guests reading the public Inbox within shared capacity), MovieMatcherAI (browsing open without an account, film analysis and fresh title lookups drawing on the plan's credits) and the Studios line.&lt;/p&gt;

&lt;p&gt;The remaining plain-text bullets are the three plan lines, the twelve terms-of-use section names and the support contact.&lt;/p&gt;

&lt;h2&gt;
  
  
  Nothing points at it
&lt;/h2&gt;

&lt;p&gt;I left a promise on 4 November: the file is not linked from our own pages. Checking that again today, 0 of the 56 pages in our sitemap contain the word "llms" anywhere in the served HTML, robots.txt does not name it, sitemap.xml does not list it, and the home page carries no link to it. It is reachable by guessing the path and by nothing else we publish. The canonical write-up for the file itself is &lt;a href="https://dev.to/onepassapps/what-apponepassappscom-tells-ai-tools-a-measured-look-at-our-llmstxt-2lko"&gt;What app.onepassapps.com Tells AI Tools&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limits
&lt;/h2&gt;

&lt;p&gt;One build, one day, plain HTTPS, no cookies, no JavaScript, no signed-in session, 12 November 2026. I compared only the 55 URLs llms.txt names. I did not test how any crawler or model reads the file, and I make no claim about whether it is read at all.&lt;/p&gt;

&lt;p&gt;The studios themselves are at &lt;a href="https://app.onepassapps.com/studios/image?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=llms-description-layer&amp;amp;utm_content=studio-link" rel="noopener noreferrer"&gt;app.onepassapps.com/studios/image&lt;/a&gt;, and the guides it links are at &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=community&amp;amp;utm_campaign=llms-description-layer&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;app.onepassapps.com/guides&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>llmstxt</category>
      <category>seo</category>
      <category>webdev</category>
      <category>documentation</category>
    </item>
    <item>
      <title>The Feature Layer of Our Own 22 Guides: 66 Bullets, Three per Guide, and Not One of Them Says Upload</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Thu, 12 Nov 2026 01:06:38 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-feature-layer-of-our-own-22-guides-66-bullets-three-per-guide-and-not-one-of-them-says-upload-5hdc</link>
      <guid>https://dev.to/onepassapps/the-feature-layer-of-our-own-22-guides-66-bullets-three-per-guide-and-not-one-of-them-says-upload-5hdc</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com.&lt;/p&gt;

&lt;p&gt;Every one of the 22 app guides we publish opens with the same short list: three sentences that say what the tool does. I have already published three other layers of those guides - the Limits and privacy section of all 22 (article 4994613), the 66 numbered steps in "How to use this app" (article 5009926) and the answer grammar of the Common questions section (article 5008397). The "What you can do" list was the one layer nobody had measured, so I fetched all 22 guide pages over plain HTTPS today, 12 November 2026, parsed that list out of each one and read all 66 sentences.&lt;/p&gt;

&lt;p&gt;Four things came out of it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Three bullets on every guide, and not one of them repeated
&lt;/h2&gt;

&lt;p&gt;All 22 guides carry exactly three bullets under "What you can do". The histogram is flat: no guide has two, none has four. Across the fleet that is 66 bullets and 1,166 words, with the shortest bullet at 10 words and the longest at 26.&lt;/p&gt;

&lt;p&gt;Every bullet is different. 66 bullets, 66 sentences, no repeats - the sameness is in the shape, not in the words, which is the same result the step layer gave for the same reason.&lt;/p&gt;

&lt;h2&gt;
  
  
  Forty opening verbs, and Download wins
&lt;/h2&gt;

&lt;p&gt;Take the first word of each bullet and count. There are 40 distinct opening words across the 66 bullets, and the most common one is not a technical verb: Download opens 6 of them, Choose opens 5, Print and Set open 3 each. The step layer underneath is narrower, where Choose opened 16 of the 66 steps. A bullet has to say what comes out as well as what goes in.&lt;/p&gt;

&lt;h2&gt;
  
  
  The third bullet is usually the output
&lt;/h2&gt;

&lt;p&gt;In 19 of the 22 guides the third bullet names an output - download, export, print, copy or save. The three exceptions are NewsFlow, MovieMatcherAI and the Focus Timer. Their last bullets read "After signing in, manage the supported source catalogue", "Record private ratings, reviews and explicit reactions as signals for future recommendations" and "Hear a chime generated in the browser, optionally get a browser notification, and see the time left in the tab title". Those three guides do not end with a file: their last bullet is about reading, recording or listening instead.&lt;/p&gt;

&lt;h2&gt;
  
  
  The privacy words are not in the list
&lt;/h2&gt;

&lt;p&gt;The word upload appears in none of the 66 bullets. It appears on 11 of the 22 guide pages, but always somewhere else - in the app description or in the limits text. The word account appears in none of the bullets either, although 20 of the 22 pages tell you somewhere that no account is needed. The word browser appears in just 6 of the 66.&lt;/p&gt;

&lt;p&gt;So the feature list describes what a tool does, and it leaves the privacy promise to the other two sections. If you are reading a bullet to decide whether a file leaves your machine, the bullet will not tell you; the limits section will.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not do
&lt;/h2&gt;

&lt;p&gt;I read the served HTML of the 22 guide pages, not the running app, and I did not open the app pages themselves. The bullets come from the ordered list inside the "What you can do" heading on each page. The step sentences from "How to use this app" are a different layer, and none of the 66 bullets is identical to any of the 66 steps.&lt;/p&gt;

&lt;p&gt;The guides themselves are at &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=guide-feature-layer&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;the app guides&lt;/a&gt;, and every app behind them is at &lt;a href="https://app.onepassapps.com?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=guide-feature-layer&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;app.onepassapps.com&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>documentation</category>
      <category>guides</category>
      <category>seo</category>
    </item>
    <item>
      <title>The Same Tags Block a WebVTT and Pass an SRT: Our Subtitle Workbench Validates the Format, Not the Markup</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Wed, 11 Nov 2026 20:49:45 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-same-tags-block-a-webvtt-and-pass-an-srt-our-subtitle-workbench-validates-the-format-not-the-578d</link>
      <guid>https://dev.to/onepassapps/the-same-tags-block-a-webvtt-and-pass-an-srt-our-subtitle-workbench-validates-the-format-not-the-578d</guid>
      <description>&lt;p&gt;I work on OnePassApps, so this is our own app and our own build, and every number below was measured in it today, signed out, in one browser, in the visible Chrome on display :1: &lt;a href="https://app.onepassapps.com/apps/subtitle-workbench?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=subtitle-format-gate&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;Subtitle Workbench&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;The Subtitle Workbench takes subtitle files you already have and tells you what is wrong with them before you export. Its own guide promises the check is about information: unsupported markup can block conversion, and converting positioning-bearing WebVTT to SRT is refused when it would lose information.&lt;/p&gt;

&lt;p&gt;Here is what the app actually does, measured cue by cue: the markup gate is a WebVTT rule, not a markup rule. The same tags that stop a WebVTT file sail through an SRT file unchanged, all the way into the exported file.&lt;/p&gt;

&lt;h2&gt;
  
  
  The same tags, two formats
&lt;/h2&gt;

&lt;p&gt;I pasted one SRT with three cues (198 B). Cue 1 carries a voice tag, cue 2 carries a bold tag and a colour tag, cue 3 is plain:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;1
00:00:01,000 --&amp;gt; 00:00:03,000
&amp;lt;v Roger Bingham&amp;gt;Voice tagged line

2
00:00:04,000 --&amp;gt; 00:00:06,000
&amp;lt;b&amp;gt;Bold&amp;lt;/b&amp;gt; and &amp;lt;font color="#ff0000"&amp;gt;coloured&amp;lt;/font&amp;gt;

3
00:00:07,000 --&amp;gt; 00:00:08,000
Plain line
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The panel read 3 cues · 0:01–0:08, then 3 cues, then No issues. Export SRT was enabled, I pressed it, and the confirmation line read Downloaded SRT for pasted-1.srt. The file that landed on disk was 198 bytes - the same 198 bytes, with the markup byte for byte intact:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight html"&gt;&lt;code&gt;&lt;span class="nt"&gt;&amp;lt;v&lt;/span&gt; &lt;span class="na"&gt;Roger&lt;/span&gt; &lt;span class="na"&gt;Bingham&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;Voice tagged line

&lt;span class="nt"&gt;&amp;lt;b&amp;gt;&lt;/span&gt;Bold&lt;span class="nt"&gt;&amp;lt;/b&amp;gt;&lt;/span&gt; and &lt;span class="nt"&gt;&amp;lt;font&lt;/span&gt; &lt;span class="na"&gt;color&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s"&gt;"#ff0000"&lt;/span&gt;&lt;span class="nt"&gt;&amp;gt;&lt;/span&gt;coloured&lt;span class="nt"&gt;&amp;lt;/font&amp;gt;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;No error, no warning, nothing stripped. The file list calls that source pasted-1.srt 3 cues · 198 B · OK.&lt;/p&gt;

&lt;p&gt;Then the same tags in a WebVTT file (221 B) are refused, one message per problem:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Error: WebVTT NOTE blocks are unsupported. Export is blocked to preserve their meaning; download the original instead.
Error: Rich WebVTT markup (voice, class, ruby, or inline timestamps) is unsupported. Download the original to preserve it.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The header counted 3 cues · 221 B and 3 errors, and the panel added a line of its own: Resolve source errors before exporting. The original file is still available. The NOTE block gets its own sentence, separate from the rich-markup one, and the rich-markup sentence names four kinds by name - voice, class, ruby, or inline timestamps. The voice tag and the ruby tag each produced one of those errors.&lt;/p&gt;

&lt;h2&gt;
  
  
  The cue that produced no error at all
&lt;/h2&gt;

&lt;p&gt;The positioned cue was the third case, and it was the interesting one, because it raised nothing. Its WebVTT cue carried line:85% align:center position:50%, and the app printed the cue and its positioning settings on a single line in the cue list: Positioned cue line:85% align:center position:50%. The file list read 2 cues · 133 B · OK and the cue count read No issues.&lt;/p&gt;

&lt;p&gt;The guide's sentence about positioning is conditional - refused when it would lose information - and this cue produced no refusal and no error. With the export format set to SRT, the button read Export SRT and the confirmation was Downloaded SRT for pasted-2.vtt. I did not read that file's bytes, so I am not claiming what it contains. What I can say is what the panel did: it raised nothing, and it showed the positioning joined to the cue text.&lt;/p&gt;

&lt;h2&gt;
  
  
  Nothing left the tab
&lt;/h2&gt;

&lt;p&gt;I captured every request the tab made at the browser's own network layer, before I typed anything. The only requests were GETs and OPTIONS to app.onepassapps.com, api.onepassapps.com and clerk.onepassapps.com - the app's own runtime and feature-flag reads - and there was no POST and no request body anywhere.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I did not test
&lt;/h2&gt;

&lt;p&gt;One build, one browser, one signed-out session, one day, 11 November 2026. I pasted subtitle text rather than using the file picker, so the 5 MB per file and 20 files per batch limits are the app's own wording and not something I measured against a file that size. I edited no cue by hand and left every timing repair alone - the shift field read +0 ms throughout. The rich-markup error names four kinds; I exercised voice and ruby, so class and inline timestamps are untested. The positioning cue is the one case I did not read the exported bytes for.&lt;/p&gt;

&lt;h2&gt;
  
  
  The earlier write-ups
&lt;/h2&gt;

&lt;p&gt;The timing and repair layers of this same app are already covered: &lt;a href="https://dev.to/onepassapps/subtitle-timing-in-the-browser-shift-cues-repair-overlaps-and-export-srt-or-webvtt-l7"&gt;Subtitle Timing in the Browser: Shift Cues, Repair Overlaps and Export SRT or WebVTT&lt;/a&gt; and &lt;a href="https://dev.to/onepassapps/the-repair-order-in-the-browser-subtitle-editor-four-checkboxes-one-sequence-and-a-measured-shift-5gfb"&gt;The Repair Order in the Browser Subtitle Editor: Four Checkboxes, One Sequence, and a Measured Shift&lt;/a&gt;. The reference-media path is covered by &lt;a href="https://dev.to/onepassapps/attach-a-video-to-the-subtitle-workbench-what-changes-and-what-the-export-still-gives-you-461m"&gt;Attach a Video to the Subtitle Workbench: What Changes, and What the Export Still Gives You&lt;/a&gt;, and the guide's own limits wording is quoted in &lt;a href="https://dev.to/onepassapps/what-each-onepassapps-app-says-it-will-not-do-the-limits-section-on-all-22-guides-4ob7"&gt;What Each OnePassApps App Says It Will Not Do: The Limits Section on All 22 Guides&lt;/a&gt;. The markup gate itself - which format gets checked, and what each format does with the same tags - is what this post is about. The &lt;a href="https://app.onepassapps.com/guides/subtitle-workbench?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=subtitle-format-gate&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;guide for this app&lt;/a&gt; is where the limits are stated.&lt;/p&gt;

</description>
      <category>subtitles</category>
      <category>webvtt</category>
      <category>srt</category>
      <category>javascript</category>
    </item>
    <item>
      <title>The Four Origins a Cold Load Reaches, and the Two the Document Never Names</title>
      <dc:creator>OnePassApps</dc:creator>
      <pubDate>Wed, 11 Nov 2026 20:33:19 +0000</pubDate>
      <link>https://dev.to/onepassapps/the-four-origins-a-cold-load-reaches-and-the-two-the-document-never-names-864</link>
      <guid>https://dev.to/onepassapps/the-four-origins-a-cold-load-reaches-and-the-two-the-document-never-names-864</guid>
      <description>&lt;p&gt;I work on OnePassApps, the team behind app.onepassapps.com. This is our own site, and every number below was measured on it today, 11 November 2026 - plain HTTPS from outside the browser for the document layer, and one real cold browser load with the cache disabled for the runtime layer.&lt;/p&gt;

&lt;p&gt;Two earlier posts on this account are the canonical write-ups for the parts they already own: the preloaded chunk set (&lt;a href="https://dev.to/onepassapps/two-of-our-three-preloaded-chunks-rotated-what-changed-in-the-asset-filenames-on-11-november-98"&gt;Two of Our Three Preloaded Chunks Rotated&lt;/a&gt;), the per-route script inventory (&lt;a href="https://dev.to/onepassapps/23-of-56-pages-ship-no-javascript-the-guide-fleet-is-static-the-app-fleet-is-not-1dkm"&gt;What Each Route Asks the Browser to Fetch&lt;/a&gt;) and the link-tag census (&lt;a href="https://dev.to/onepassapps/the-link-layer-of-apponepassappscom-491-link-tags-two-shapes-and-one-preload-set-44d"&gt;The Link Tags of 56 Pages&lt;/a&gt;). A fourth covers what a visitor pays to open a URL that does not exist (&lt;a href="https://dev.to/onepassapps/what-a-404-costs-on-apponepassappscom-114-requests-12-mb-and-a-title-that-never-changes-3hlm"&gt;The Cost of a Wrong URL&lt;/a&gt;). This post is about the layer none of them covers: which hosts a page actually reaches, and what the document says about them before any of it runs.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I measured
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;All 56 URLs in our own sitemap.xml, fetched over plain HTTPS - signed out, no cookies, no JavaScript&lt;/li&gt;
&lt;li&gt;The link tags in each served document, and the response headers on every answer&lt;/li&gt;
&lt;li&gt;One cold load of an app page, &lt;code&gt;/apps/focus-timer&lt;/code&gt;, and one of a guide page, &lt;code&gt;/guides/focus-timer&lt;/code&gt;, in desktop Chrome with the cache disabled, with every request captured&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A cold app page reaches four origins
&lt;/h2&gt;

&lt;p&gt;The app page made 125 requests, to four different origins.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Origin&lt;/th&gt;
&lt;th&gt;Requests&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;app.onepassapps.com&lt;/td&gt;
&lt;td&gt;109&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;clerk.onepassapps.com&lt;/td&gt;
&lt;td&gt;10&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;api.onepassapps.com&lt;/td&gt;
&lt;td&gt;5&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;static.cloudflareinsights.com&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The guide page is a completely different shape. It boots no app bundle, so it made 10 requests to two origins: nine of ours and the same single Cloudflare beacon.&lt;/p&gt;

&lt;p&gt;So the bundle is what pulls in the other two origins. The sign-in host and the API host are both reached because our JavaScript asks for them after the document has parsed, not because the document asked for them.&lt;/p&gt;

&lt;h2&gt;
  
  
  The link tags: 491 of them, and not one hint
&lt;/h2&gt;

&lt;p&gt;Counted across all 56 served documents, there are 491 link tags: 112 stylesheet, 112 icon, 99 modulepreload, 56 apple-touch-icon, 56 manifest and 56 canonical. That census is the subject of the earlier post, so I am not repeating the arithmetic.&lt;/p&gt;

&lt;p&gt;What is new is what is missing from it. There is no preconnect tag on any of the 56 pages. No dns-prefetch tag. No prefetch tag. And not one plain preload relation either - all 99 are modulepreload, and the three files they point at are the same three on every page that carries them.&lt;/p&gt;

&lt;h2&gt;
  
  
  And no Link header either
&lt;/h2&gt;

&lt;p&gt;The HTTP-level version of the same hint is absent as well. Not one of the 56 responses carries a Link header, so nothing about the two origins we depend on is pushed by the server on the first answer either. The connection to them is set up purely by code that runs later.&lt;/p&gt;

&lt;h2&gt;
  
  
  The served documents never name either origin
&lt;/h2&gt;

&lt;p&gt;This is the part I did not expect. I looked for both hostnames inside all 56 served documents: 0 pages name api.onepassapps.com, and 0 name clerk.onepassapps.com. The Cloudflare beacon is the opposite - all 56 pages carry it.&lt;/p&gt;

&lt;p&gt;So the markup names one origin in advance, and that one belongs to Cloudflare, not to us. The two origins our app genuinely needs are found by running our JavaScript.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the API host is asked for
&lt;/h2&gt;

&lt;p&gt;Five requests, three routes, and the first two are asked for twice.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;/api/runtime&lt;/code&gt; - twice, 471 bytes each&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;/api/config/feature-flags&lt;/code&gt; - twice, 1,771 bytes each&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;/api/auth/config&lt;/code&gt; - once&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The sign-in host answers a redirect first
&lt;/h2&gt;

&lt;p&gt;Of the 10 requests to the sign-in host, two are for unpinned major-version paths. Each one answers HTTP 307 with a Location header pointing at the pinned version - 6.42.3 for the browser bundle, 1.42.4 for the UI bundle - and it is the pinned file that is actually served.&lt;/p&gt;

&lt;p&gt;That pinned browser bundle is 327,846 bytes and carries &lt;code&gt;public, max-age=31536000, s-maxage=31536000, immutable&lt;/code&gt;, so once the version is pinned it is cached for a year at the edge. The redirect itself is the cheap part: a 307 with no body.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limits
&lt;/h2&gt;

&lt;p&gt;One build, one day, one browser, one signed-out session, one app route and one guide route measured in the browser. The 125 and the 10 are for those two pages, not for every page in the fleet. The document layer was measured over plain HTTPS with the redirect handler disabled, so the 308 answers are counted as answers and not followed. The host counts come from the request URLs as the browser reported them, and the byte counts from a separate fetch of the same paths from this server. I did not measure a second browser, a second network, or a warm cache.&lt;/p&gt;

&lt;h2&gt;
  
  
  The tools themselves
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://app.onepassapps.com/apps/focus-timer?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=origin-hints&amp;amp;utm_content=app-link" rel="noopener noreferrer"&gt;the Focus Timer app&lt;/a&gt; and &lt;a href="https://app.onepassapps.com/guides?utm_source=devto&amp;amp;utm_medium=post&amp;amp;utm_campaign=origin-hints&amp;amp;utm_content=guides-link" rel="noopener noreferrer"&gt;the app guides&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>performance</category>
      <category>http</category>
      <category>javascript</category>
    </item>
  </channel>
</rss>
