Methodology

Every tool is exercised against real inputs before it ships and after any change: genuine PDFs from scanners and office suites, photos straight off phones (including HEIC), messy CSVs, oversized images, malformed JSON, and the awkward edge cases that break naive implementations — encrypted PDFs, zero-byte files, enormous inputs, mixed page orientations. We test across the current versions of Chrome, Edge, Firefox and Safari, on both desktop and mobile viewports, because a browser-only tool lives or dies by cross-browser behaviour.

Where a tool makes a specific promise — "bookmarks are preserved", "the output opens in Microsoft Word", "HEIC is auto-detected" — that promise is verified against the real target application, not assumed.

"Fast" is easy to claim and easy to fake. We use it to mean two concrete things. First, work happens on your device, so there is no upload-and-wait round trip — a 50 MB file isn't sent anywhere before processing can begin. Second, the heavy lifting runs in WebAssembly or a Web Worker so the interface stays responsive while it works. We don't publish synthetic benchmark numbers dressed up as universal truth, because real speed depends on your device and file; we describe the architecture that makes it fast and let you feel the result.

The central privacy claim — your files never leave your browser — is not something you have to take on trust. It's verifiable in about thirty seconds, and we verify it the same way:

• Open your browser's developer tools and switch to the Network tab.
• Use any tool — merge a PDF, compress an image, decode a token.
• Watch the request list. The file you processed generates no upload request; the only network traffic is the page's own assets, loaded once.
• For the strongest proof, disconnect from the network entirely after the page loads. The tools keep working, because there was never a server in the loop.

The same property is what makes SnapToolz safe for confidential material, and it's why our security tools (password generation, AES encryption) are honest places to handle secrets — there is no server that could log them.

Technical claims in our guides are sourced from primary specifications and authorities, as set out in our editorial policy. Pages carry a last-reviewed date; when a standard changes, a browser ships a relevant feature, or we find an error, we update the page and the date rather than leaving stale guidance in place. The product itself ships changes openly — you can follow them on the changelog.