1 · Watch the network panel
- Open any Bindery tool with a harmless sample PDF — the built-in samples work, or use the dashboard’s try the sample files link.
- Open your browser’s developer tools (F12) and select the Network panel.
- Run the job — compress, merge, redact, anything.
- Inspect the requests: you will see application assets load (scripts, fonts, icons), and nothing containing your document’s bytes. There is no upload request because there is no upload endpoint.
2 · Pull the plug
Load Bindery once, then disconnect from the network entirely — airplane mode, Wi-Fi off, cable out. Keep working. Merging, splitting, compressing, OCR and every other tool continue to function, because the processing engine is already on your device. A tool that uploaded your files for processing could not do this.
The installable app makes the same point permanently: install Bindery from your browser menu and it runs as an offline-capable application.
3 · Audit the engine
The transforms behind the web tools are published as the bindery-pdf package on npm — the CLI ships the same ops.js and vendored pdf-lib build the site runs, copied verbatim at packaging time. Reading that package is reading the product. The About page documents how product behavior is tested, and the offline PDF tools guide explains what browser-local processing can and cannot do.
What this proves — and what it doesn’t
These checks prove document bytes stay on the device during processing. To be complete about the rest: ordinary website telemetry (cookieless page analytics), optional account data (identity, subscription state, a daily job count, saved recipes and presets) and payment processing through Stripe all exist and are described in the privacy policy. None of them ever contain your documents.