Would you consider open-sourcing the client-side code? For privacy-focused PDF tools, that seems like the easiest way to make the “no upload” claim more trustworthy.
[dead]