abox.tools VS https://open-gpt.app/

abox.tools's answer

None — abox.tools is free and has no accounts, no signup and no server, so there are no customer records.

abox.tools's answer

All tools do their work entirely inside your browser. The claim isn't "we promise not to look at your files" — it's that there is no code path that could send them anywhere. Every tool page ships a Content-Security-Policy of default-src 'none', and its connect-src does not include 'self', so the page cannot open a connection back to its own origin even if code were added that tried. The site is open source and the built output is committed to a branch, so anyone can diff what's actually served against what a build of the sources produces.

abox.tools's answer

The usual online converters — CloudConvert, iLovePDF, Smallpdf, the various online EXIF removers — upload your file to a server, process it there and hand back a download link. That buys you a queue, a file-size cap, usually an account, and a copy of your file on somebody else's disk.

abox.tools does the same work on your own machine instead: no upload, no account, no file-size limit, and it keeps working offline after the first visit because every tool registers a service worker. It's free, and there's nothing to install.

abox.tools's answer

People who need a one-off file job done and would rather not hand the file over: converting iPhone HEIC photos, stripping GPS and EXIF data before posting a picture, trimming a video or an audio clip, merging PDFs, making a favicon or a QR code.

Two groups over-index. Privacy-conscious users, who want the guarantee rather than the promise. And developers and IT staff, who can read the source and check the claim rather than take it on faith. It's also used by people handling material they're not permitted to upload at all — client documents, medical scans, anything under an NDA. The site is available in 11 languages.

abox.tools's answer

It started from a small observation: "free online converter" almost always means "upload your file to our server." Modern browsers can do nearly all of that work themselves — WebCodecs, Canvas, Web Audio, WebAssembly — so for most of these jobs the upload is a business decision rather than a technical necessity.

So it was built as a static site with no backend at all, which makes the privacy claim structural instead of a policy someone has to trust. It's deliberately kept small and readable: no dependencies, no lockfile, no JavaScript build step, so the code in the repository is byte-for-byte what the browser runs.

abox.tools's answer

• Static site generated by a Python 3.11+ script using only the standard library — no runtime, no database, no backend of any kind.
• Vanilla JavaScript ES modules. No framework, no dependencies, no lockfile, no bundler; tools//src/ is byte-for-byte what the browser executes.
• Browser APIs do the actual work: WebCodecs (VideoEncoder/VideoDecoder/AudioEncoder), Canvas and createImageBitmap, Web Audio, ImageDecoder, and Service Workers for offline use.
• One vendored WebAssembly engine, libheif, for HEIC decoding — it runs on the visitor's machine like everything else.
• Hand-written format code where the browser offers no API: GIF LZW encoder and decoder, MP4 reader and writer, PDF reader and writer, EXIF/TIFF parser, QR and barcode generation.
• Served as static files from GitHub Pages, with DNS at Cloudflare.