ohaodha

I find it difficult to be impressed by "prompt injection" attacks that require the victim to enter the malicious prompt themselves --- like, really? If you tell Rovo to exfiltrate your data, it'll do it?

Obviously, there should be URL protection rules to control what it can access, but this requires a very specific and unlikely set of circumstances to exploit.

I have a similar project for speed-reading in the terminal: https://github.com/0hAodha/ogma

It works on plaintext, but you can define pre-processing hooks to convert, e.g., a PDF or webpage into readable content using the config file.