What “zero upload” means here
The available PDF tools read your selected file using browser File and ArrayBuffer APIs, process the bytes on your device, and create a Blob for download. They do not send the selected document to a conversion endpoint.
Visiting a website still downloads HTML, JavaScript, styles, fonts and other resources. Hosting can receive normal connection details such as the requested URL and IP address; zero document upload is not a claim that visiting the site creates no network traffic. Emailing support is a separate action covered by the privacy policy.
Check a tool in Chrome, Edge or Firefox
- Open a tool such as Merge PDF. Use a synthetic PDF containing an unusual test phrase, rather than a confidential document.
- Open Developer Tools, choose Network, keep all request types visible and enable the Method column. Clear the current list after the page has loaded.
- Select the sample file and run the operation. Keep the Network panel open until the result is ready.
- Inspect new requests, their method, initiator and request payload. Static GET requests for scripts, workers, fonts, WASM or OCR language models are expected. A request carrying your document bytes to an upload or conversion endpoint would contradict the claim.
- Check the result download. A browser-created blob: URL is a local output reference; it is not an uploaded server job. Compare the downloaded pages or text with your input.
- Repeat on the specific tool and browser you intend to use. A source audit and a successful sample do not establish the behavior of browser extensions or a future modified deployment.
Why OCR can download data before it works
OCR loads the Tesseract.js worker, a recognition core and the chosen English/Hindi model from this website. PDF.js may also load fonts, CMaps and WASM resources. Those dependency downloads are different from transmitting your selected PDF; inspect their methods and payloads rather than judging by the presence of traffic alone.
The real libraries behind the tools
- @cantoo/pdf-lib 2.9.1: page operations, supported forms, metadata and encryption.
- PDF.js 5.6.205: text extraction, PDF reading and raster rendering.
- Tesseract.js 7.0.0: local OCR with bundled English and Hindi data.
- JSZip 3.10.1: packaging multiple outputs into a ZIP.
- SheetJS 0.20.3, Mammoth 1.8.0, docx 9.6.1 and PptxGenJS 4.0.1: the text, row and page-image Office workflows described on each tool page.
- qrcode-generator, JsBarcode, fontkit and Noto fonts: encoding, barcode generation and supported text rendering.
These libraries are self-hosted, and their license files are bundled with the assets. You can inspect the current processing-engine source; a visitor’s public tools do not require a backend job queue.
What the sample tests establish
The automated suite exercises all 53 available tools with synthetic files and checks real outputs. Native-browser Merge and Split downloads were independently inspected on 19 October 2026. Other native-browser checks were interrupted by a testing-environment timeout, so those results are not claimed here.
Read the testing and availability notes, keep your originals, and inspect each output. PDF/A and translation are unavailable integrations, rather than successful local conversions.
Source and test notes reviewed 2026-10-19. Developer and technical contact.