Security
How the product is built, and how to tell us if we got something wrong.
Last updated 16 August 2026
In your browser
- Every file is checked against its real content signature — its “magic bytes” — as well as its extension, so a renamed file cannot reach a parser that does not expect it.
- The PDF renderer is configured with remote fetching, XFA and system font access disabled, and character maps and standard fonts are served from this site rather than a CDN.
- Rendering and heavy processing run off the main thread, so a malformed document degrades into an error message rather than a frozen tab.
- A strict Content Security Policy limits what the page itself is allowed to load.
On the server
The optional processing service exists only for conversions that genuinely need a document engine. When it is enabled:
- requests are rate limited per network address;
- only an allow-listed set of conversions is accepted;
- file names are sanitised and never interpolated into a shell command;
- conversion runs as an unprivileged user in a container with no network access;
- each job gets a random identifier and its own temporary directory, which is deleted when the job finishes and swept again by a periodic cleanup task;
- document contents are never logged.
Reporting a vulnerability
If you find a security problem, please email support@pdfnexly.com with enough detail to reproduce it, and give us a reasonable chance to fix it before publishing. We are grateful for the help.