Managing blocks

Where everything lives#

The activity log is deliberately several steps deep. It contains the names of gambling sites that were blocked, and that list should never be one glance away from the home screen.

Blocking something Haven missed#

Guard → Personal filters → add.

Two kinds:

• A website. Enter the domain. It covers subdomains automatically, so one entry for a site is enough.
• A keyword. Blocks any domain containing that word. Powerful, and easy to overdo — very short keywords are refused because they over-block badly.

Both take effect immediately.

When Haven blocks something it should not#

First, check Guard → Personal filters to see whether the domain is listed there.

If you added it yourself#

Remove it with a single tap. It was your rule; undoing it is your decision.

If Haven learned it automatically#

The entry shows why it was caught — the score and the specific signals. You have two options:

• "This is not a gambling page" — a correction.
• "I know what it is and I want in anyway" — a decision.

Both cost the same. Haven cannot tell which is true, and if one were the fast lane, that is simply the button an impulse would press. What differs is the record: the two read differently afterwards, and your accountability partner sees which one you chose.

How long it takes depends on how strong Haven's own evidence was:

A release you asked for and never collected expires after a week, so a decision you made a month ago cannot be sitting there pre-approved.

You can block anything again instantly. Strengthening protection never has a waiting period.

Full detail: Self-learning and personal filters.

When an app is flagged wrongly#

Guard → App-check. A heuristically-flagged app can be released, and the release can be undone later.

An app that belongs to a known gambling operator cannot be released. That would not be a correction, it would be a door.

Keeping the rules current#

Haven pulls signed rule updates roughly twice a day by itself. Guard → sync now does it on demand and reports the result, so you can verify the mechanism rather than take it on faith.

Updates are cryptographically signed and verified on the phone. A tampered or replayed update is rejected, and a failed update leaves your current rules working.

Reading the activity log#

Settings → Advanced → Diagnostics → Activity.

Each entry shows the host, when it was blocked, and which layer and rule caught it. If the page-content watch caught it, the reasons are attached.

This log is on your phone and nowhere else. It is never uploaded.

Turning individual layers off#

The Guard screen shows each layer and whether it is actually active. If a layer is inactive because a permission is missing, Haven says which one rather than implying coverage it does not have.

Turning protection itself off is a different matter, with its own page: Turning protection off.