Tanay Roy

Every ad blocker I have used tells you it blocked 47 things and never tells you which 47, or why. When a page breaks, your only real option is to switch the whole thing off and hope.

JSK Shield shows the request it blocked, the exact filter rule that matched, and the list that rule came from. If one blocked request is what broke the page, you allow that single request on that single site and keep everything else. Guided recovery walks the fixes least-destructive-first: unblock the one request, then turn off element hiding, then pause on that site — in that order.

Honest about what this is not: uBlock Origin blocks more, and its filter engine is years ahead of mine. If raw blocking power is what you want, use uBO — it is excellent and free. This is for people who want to see the mechanism and make per-request decisions rather than per-site ones.

Everything runs on your device. No telemetry, no account needed to block, no paid allowlist. GPL-3.0, source on GitHub. Blocking stays free forever; an optional Family plan adds parental controls.

Happy to answer anything, including the awkward questions.

Coming at this from the other side of the fence - I build an ad blocker, so I spend a lot of time looking at what one-snippet widget bundles actually do on a page.

The claim I would most want nailed down is "trackers stay blocked until consent". Consent scripts sit on the annoyance lists in most blockers, which means the script enforcing that rule is itself a likely block target. What is the failure mode when your snippet never loads? If the site falls back to no-consent-no-tracking, that is genuinely strong and worth saying out loud. If the trackers fire because the gatekeeper was not there, that is a compliance hole that shows up precisely for the subset of visitors most likely to care about it.

Second, on the 13-month consent log: is it written client-side, or reconciled on your server? A client-written log has the same problem - the evidence goes missing for exactly the visitors whose consent you would most need to prove.

Nice touch that the security scan ships a fix prompt written not to break a working site. For leaked keys though, I would hope the copy says rotate first - a prompt can strip the key from the repo, but not from whoever already has it.