PdfNetStack parses untrusted input, so parser robustness is a security property.

Report privately through GitHub: Security > Report a vulnerability on this repository. Please do not open a public issue for crashes, hangs, memory exhaustion or anything that lets a crafted PDF escape the size and nesting limits.

Include a proof of concept file if you can, the options used, and the .NET version. You will get an acknowledgement within a week.

In scope: crashes or unhandled exceptions from crafted input, unbounded CPU or memory use beyond the configured limits, output that differs from the input while verification reports success.

Out of scope: high memory use that stays within the documented limits and scales with MaxDegreeOfParallelism, and files the optimizer correctly refuses.

• Keep MaxInputBytes and MaxDecodedStreamBytes at values your workers can afford.
• Run batch workers with an OS level memory limit (container limits or job objects).
• Use a CancellationToken with a timeout per document.