The proof
Every tool site on the web writes that it respects your privacy. We would rather show you how to check. There are three ways to do it, from the quickest to the most thorough, and none of them requires knowing how to program.
The quick method: the network inspector
The browser itself shows the complete list of what comes in and what goes out. It is the most direct proof, and it takes ten seconds.
- Open a tool on this site
The metadata viewer, for instance. Wait for the page to finish loading. - Open the inspector
The F12 key on Windows and Linux. On a Mac, Cmd + Option + I. You can also go through the browser menu, under “Developer tools” or “Inspect”. - Go to the “Network” tab
Then click the clear icon or “Clear” to empty the list: it must be empty before the test. - Drop a file in and run the tool
This is the moment that counts. Watch the list while the tool works and while you download the result. - See for yourself
The list stays empty. Not one line appears, not while the file is read, not while it is processed, not on download, because the download is built by your own browser out of the tab's memory.
An honest nuance: on the very first load of the page, the inspector does show lines. Those are the site's own files, the page, the stylesheet, the fonts, the scripts, all served from this same domain. What matters is what happens AFTER that load, during processing: that is where other sites upload, and that is where the list must stay empty.
The method with no comeback: cut your connection
A site that needs a server to work cannot work without a network. This one can.
- Visit the tool once, with a connection
The site then installs itself into your browser's storage, exactly like an application. - Cut the network
Aeroplane mode, or unplug the cable, or set the inspector's Network tab to “Offline”. - Reload the page and use the tool
Everything works: the page appears, the file drops in, the processing happens, the result downloads. A site that uploads cannot do that.
The thorough method: read the code
The slowest and the strongest. It needs no account and no special tooling.
- Show the page source
Ctrl + U, or right-click then “View page source”. No code on this site is minified or obfuscated, and that is a deliberate choice: unreadable code cannot be checked. - Open the tool's file
Each tool is ONE JavaScript file, named after itself, loaded from this domain. Inside you will find the file reading, the processing, and the building of the result. - Look for anything that could send something
In a browser there is only a small family of ways to send data out: the words to search for are fetch, XMLHttpRequest, sendBeacon, WebSocket, EventSource, and image or script tags built on the fly. You will find none of them in a tool's code.
And one proof we owe you: the machine's
The three methods above belong to you: they let you check without us. The fourth one binds us.
Before every release of the site, a program launches a real browser, opens every tool page, drops a real test file in, runs the processing, and watches the network at the level of the browser itself, not at the level of the page's code. If it sees a single request go out to another server during processing, publication is refused. Not flagged: refused.
It is the only lock in the project that cannot be worked around by changing a sentence. It lives in the project folder under the name “validate-localite”, and it runs every single time.