Methodology

Where your files are processed

The table below is generated from the tool registry: a category reads “In your browser” only when every published tool in it is marked as client-only in the code. CPU-heavy jobs such as PDF assembly and data parsing run in a Web Worker, which is still code running on your own device — a worker is not a server.

No tool on the site has a server-side processing path, and the application exposes no endpoint that accepts a file or pasted text. Because nothing is received, there is no retention period to state: an uploaded copy does not exist.

How a tool page is written

• Each page is built from catalog data, not from a template with the name swapped: a one-line answer, a short introduction, numbered steps, a key-facts block, concrete use cases, stated limitations and four questions.
• Limits on a page — file-size caps, page counts, entry limits — are read from the same code that enforces them, so a stated cap is the one the tool actually applies.
• Input and output formats are listed only where the tool genuinely accepts or produces them. A fact that is not known is left out rather than estimated.
• Pages avoid claims the project cannot support: no ranking promises, no accuracy percentages, and no comparisons with other products.

How pages are reviewed before they go live

1. A new page starts as a draft and is marked noindex, so it is reachable for review but cannot appear in search results or in the sitemap.
2. Automated checks run first: catalog validation, a duplicate-copy scan, linting, type checking, unit and browser tests, and a full production build.
3. A person then reviews the page: are the claims true to the tool, are the limits the real ones, do the examples work, and do the links point somewhere relevant.
4. Only after that review is a page switched to indexable and added to the sitemap. Until then it stays a draft.

How to report an error

There is no working contact channel configured for this site yet, and the tool pages carry a “Send feedback” link that is not wired up. That gap is recorded on the contact page, which also lists exactly what a useful error report should contain, so it can be sent the moment a channel exists.

What this page does not claim

Nothing here is a guarantee of accuracy, availability or search performance. Results depend on the input you supply, and a converted file or a calculated figure should be checked before it is relied on.