anfragment/zen

Implement differentiation between "trusted" and "untrusted" filters

Closed this issue · 1 comments

anfragment commented

Description

Some scriptlet rules (prefixed by trusted-) should only work in "trusted" filters, as they have a potential to disrupt user privacy and security. Thus, we need:

  1. A documented and well thought-through algorithm for deciding which lists should get a "trusted" status.
  2. A code implementation of the concept.
anfragment commented

Which lists get a "trusted" status

  1. This problem should be approached in a manner similar to the principle of least privilege. This means that only lists that use trusted scriptlets (and use scriptlets at all) should be granted a trusted status.
  2. We should keep the number of trusted filter lists to a minimum. I suggest setting a limit of 5 for now.
  3. Trusted filter lists should either be open-source and distributed via a repo-linked CDN (such as GitHub), or maintained by a trusted, community driven organization.

Considering the lists currently included in our default configuration, I propose granting trusted status to the following two lists:

  1. AdGuard Base filter
  2. AdGuard Spyware filter
  • 👍1