Big fan of the idea of offline in-browser tools: works on any device, and "installing" the program is as simple as visiting a webpage or opening the saved HTML file. This one uses WebAssembly, but even that isn't necessary: I wrote a couple of PDF tools for my own use that simply use pdf-lib: https://shreevatsa.net/pdf-pages/https://shreevatsa.net/pdf-unspread/
I created something similar for pdf diffing since I needed “visual” diffing instead of “textual” diffing. It uses Mozilla’s PDF.js library as the rendering engine. It’s amazing how much you can do in-browser. https://parepdf.com/
I created a tool to take project plans from a tool that we use to model schedules and generate gantt diagrams from the data in a way that is better than the commercial tool. It's all just a zip file containing an html file and a couple of javascript files. Others in our company are using it now as it is easier and produces better output than the modeling tool.
When I decided on a local html for the app, I wanted something that did not require installing a language like python, did not require installing, and would work on Windows and Mac.
I also love the idea. However it's really hard to trust as a non IT person who understands that's possible and how to verify (via dev tools for example) that it's truly working offline.
I'm not really sure what the solution is, but I'd love to be able to navigate to a site like that and know that it's not uploading my files somewhere, without having to take your word for it.
Software that does not run in the browser can do even worse things (that are not possible in a browser): it could access (and upload) arbitrary files other than just the PDF file(s) you gave it access to, it could set up things that stay running even after you close it (like a virus or keylogger), etc — so you need to verify/trust the developer or run it inside a sandbox, and the same is true for in-browser software. At least with a browser, the tools needed for monitoring and disabling network access (dev tools) are built-in to the browser (e.g. don't have to separately install Little Snitch).
One thing that seems worse with software-on-a-webpage is the risk of malicious behaviour getting added later (this is also possible with non-browser software with automatic "updates"): you could mitigate it, if you trust or have verified one particular version of such a webpage, by saving the HTML file (put it in a Dropbox folder say) and opening the file (maybe with network disabled), whenever you need to use it. But then you lose the advantage of "don't have to install anything"; it's a trade-off.
