Quality & test lab
Know what changed before you trust the result.
Browser-based document tools are convenient, but no converter can preserve every feature in every file. This page explains what the service checks, what can change, and the short inspection every visitor should perform before sharing or relying on an output.
Synthetic test files
Automated checks use generated, non-sensitive fixtures for page operations, conversion, file limits, cancellation, and error handling.
Privacy boundaries
Tests preserve the local-first design. Ordinary files remain in browser memory; Secure Link is tested as the documented encrypted-storage exception.
No blanket guarantee
Passing tests cannot certify every PDF, browser, font, device, assistive technology, or unusual document feature.
What we check—and what you should still inspect
| Task | Core checks | Common limitations |
|---|---|---|
| Merge, split, rotate and reorder | Page count, selected order, rotation, and ability to reopen | Bookmarks, links, annotations, forms, and unusual attachments may not behave exactly like the source |
| PDF to Word or text | Extracted text, paragraph sequence, and whether the output opens | Complex columns, tables, handwriting, scans, fonts, and exact page layout are not recreated |
| Images to PDF | Image order, orientation, page count, and readable output | HEIC decoding depends on browser support; images may be resized to fit PDF pages |
| Compression | File opens, every page remains, and visual quality is acceptable | Already-optimized PDFs may shrink very little or can become larger |
| Fill, edit, redact and type a signature | Every intended mark appears in the downloaded copy | A typed mark is not identity verification; visual redaction needs careful inspection |
| Secure Link | Ciphertext upload, recipient opening, expiry, and optional one-view behavior | This feature temporarily stores browser-encrypted data and is not intended for regulated records |
A 60-second result check
- Keep the original file unchanged until you have approved the result.
- Open the downloaded file in a trusted viewer—not only the browser preview.
- Confirm page count, order, orientation, text, images, marks, links, and form values that matter.
- Zoom in on small text and inspect the first, last, and any unusually complex pages.
- If the document will be printed, signed, filed, or relied on, test that exact next step with a non-sensitive copy first.
What “local processing” does and does not mean
For ordinary tools, selected document content is processed in browser memory and is not intentionally uploaded to the application server. Your browser still requests the public website, so hosting and security providers can receive ordinary network metadata. Downloads remain on your device until you remove them. Secure Link is different: it uploads only ciphertext encrypted in the browser, while the decryption key remains in the browser fragment; the feature expires access after about 15 minutes.
Current assurance boundary
The project uses automated regression, security, accessibility, browser-quality, payment-state, cleanup, and production-health checks. These are internal engineering controls—not an independent penetration-test certificate, cryptographic certification, accessibility conformance claim, legal opinion, or guarantee of perfect conversion. The service is not intended for personal health information or regulated legal, financial, government, child-related, or professional records.
Evidence behind these instructions
The release checks exercise generated PDFs and images rather than customer documents. They cover core PDF operations, ordered multi-image conversion, cancellation and file limits, secure-link cleanup, payment-state transitions, route-level advertising exclusions, keyboard-visible controls, production builds, and public health responses. Every tool guide is also checked against the application's real tool name and supported workflow so a guide cannot silently point to a missing feature.
This evidence is deliberately bounded. Automated checks can show that known examples behave as intended; they cannot establish that every PDF feature, font, browser, assistive technology, or low-memory device will behave identically. That is why the original-file and output-inspection steps remain part of every workflow.
. This page is updated only when testing practices, evidence, or material product limits change.
Try the workflow with a non-sensitive sample
Choose a supported tool, read its specific limitations, then inspect the output using the checklist above.