# Privacy — Adamancia Vault website

Updated 2026-10-04 for official dictionaries, creator contributions and alias-based email support.

This describes customblocker.com. The apps’ data practices are explained in their
own policies and manuals. Browsing this website does not connect it to your
Vault configuration, Classifier evidence, API keys or Activity records.

## Browsing and downloads

The website has no accounts, analytics trackers, download tracking or public
statistics. Routine proxy access logs are disabled. Service error and lifecycle
logs remain for operating the service; hosting and network providers process
requests to deliver the website.

The App/Extension demos and landing charts use fictional sample data. Demo edits
are temporary browser memory and are discarded on reload/reset. The demos do
not control your apps or record your browsing.

## Email support

The Contact authors section provides [hello@customblocker.com](mailto:hello@customblocker.com).
Email me opens your email app; Copy address copies that address to your clipboard.
The website does not submit or store reports or screenshots. Information you
choose to email is processed by your mail provider, SimpleLogin’s forwarding
service and the recipient’s Gmail service, and received by the creator. Replies
arrive by email.

## Retired accounts and reports

Website sign-in and on-site conversations are retired. Old browser session
tokens are removed when the current website loads. The owner approved deletion
of stored account/report records after private database/upload backups. Those
backups remain private disaster-recovery material and are not served by the site.

## Product data

For the software, read the [Mac user manual](../manuals/mac.md),
[extension user manual](../manuals/extension.md) and each product’s policies in
its source repository. Optional app web research and AI tools have their own
data-sharing choices; they are separate from this website.

## Official dictionaries and creator contributions

Vault checks for official term and creator dictionary updates at startup. You choose when to download updates. Creator lookup defaults to a 10,000-entry cache with adjustable capacity; a cache miss sends the public, platform-scoped creator ID to our server. Full-download creator lookup is local. Dictionary downloads and lookups do not require your own AI API key. User-written and user AI-generated definitions remain local and take priority over official definitions. Optional web/AI research uses its separate provider settings.

Help improve the creator dictionary is ON by default. On first launch, a screen explains the feature and lets you disable it before any contribution is sent. You can turn it off at any time in Settings → Classifier. When enabled, the app samples missing public creator IDs and their displayed public subscriber/follower counts (which may be rounded); missing counts are sent as unknown. It sends no term names, titles, browsing history, personal definitions, API keys or persistent user/device identifiers. Submissions are deduplicated locally, sampled one in four IDs per day, and capped at 50 per day. Turning it off cancels pending contribution requests and stops future submissions; it does not recall requests already received.

The server retains candidate creator IDs, public counts and last-seen timestamps for seven days after the latest submission. An in-memory, temporary address hash limits abuse and expires hourly; the contribution database stores no IP addresses. Hosting and network providers necessarily process requests. These candidate submissions help prioritize the project-maintained public dictionary; they are distinct from routine cache-miss lookup. Official descriptions are maintained separately and may remain published in future versions.