Skip to content

Privacy policy

Last updated: 19-09-2026

This page describes what this site collects, why, and how long it keeps it, based on what the software actually does rather than on what a generic privacy policy usually says.

The short version: you can search without an account and without being identified. If you make an account, we store an email address, a hashed password, and the searches you choose to save. There is no analytics, no tracking, and nothing about you is shared with anyone.

Separately, the crawler behind this site handles the IP addresses of people using BitTorrent, in memory only. That's covered further down, under "Other people on the BitTorrent network".

Browsing without an account

If you only search and browse, nothing about you is stored.

  • No analytics, no tracking pixels, no fingerprinting, no advertising.
  • No third-party requests. Every script, style, font, and image comes from this site itself; your browser is not asked to contact anyone else.
  • No cookies, unless and until you sign in. Nothing is written to your browser's local or session storage at any point.
  • Searches you type are not logged and not associated with you. Search results are briefly cached on the server for a minute or so, keyed by the search text itself, so that several people running the same search cost the instance one query. That cache holds results, not identities.

If you create an account

An account exists so you can save searches as feeds and use the Torznab endpoint. It holds:

WhatWhyNotes
Your email addressIdentifies the account at sign-inStored lowercased and trimmed. Not verified, not used for anything else.
Your password, hashedSo you can sign inHashed with scrypt and a random salt. The password itself is never stored and cannot be recovered from the hash.
The date you registeredHousekeepingA timestamp, nothing more.
A hash of your Torznab API keyTo check the key when a client presents itOnly if you generate one. The key itself is shown to you once and never stored.

That is the whole account record. No name, no IP address, no login history, no device information, no profile.

We have no way to email you: this app has no mail integration at all. So there are no newsletters, no marketing, and no automated password reset. If you lose your password, write to support@torrents.observer.

Sessions

Signing in sets one cookie, named session. It holds a random token and nothing else. No personal data is encoded in it.

  • It is HttpOnly (scripts can't read it), SameSite=Lax, and Secure on any HTTPS deployment.
  • It lasts 30 days.
  • Only a SHA-256 hash of the token is stored on the server, next to your user id and an expiry time. The token itself only ever exists in your browser.

Logging out deletes the session record on the server and clears the cookie. Sessions that simply expire stop working immediately, but their record (a hash and a timestamp, nothing readable) may remain in the database until it is cleared out manually. Like everything in the database, session records are also copied into the daily backups described under "Where it's stored".

Saved feeds

When you save a search, we store the name you gave it, the search text, the filters you selected, when you created it, and a SHA-256 hash of that feed's access token. The token itself, like the API key, is shown once and never stored.

Deleting a feed deletes that record from the database. Rotating its token replaces the stored hash, and the old feed URL stops working. Earlier copies of the record stay in the encrypted backups for up to 30 days before they are deleted too.

IP addresses

Your IP address is used, in memory only, as the key for rate limiting, so one client can't overwhelm the instance behind this site. Those counters are held in the running process, swept once a client is back under the limit, and never written to the database. They are not linked to your account.

Server logs

Neither the application nor the reverse proxy or hosting environment in front of it writes a request log, so there is no per-visit record of who asked for what. The application logs errors to the container's output so the operator can see when something is broken. Those error messages deliberately exclude session cookies, feed tokens, API keys, and captcha payloads, though technical details of a failed request can appear in them. The only thing retained beyond that is the anonymized aggregate metrics described below, if the operator has enabled them.

Metrics

If the operator has configured a metrics collector, the app pushes aggregate counts to it: how many requests arrived per broad area of the site, how many returned an error, how many were refused by a rate limiter, how many captcha checks passed or failed, how many known crawlers were seen by name, and totals for registered accounts, unexpired sessions, and saved feeds.

These are numbers, not records. No IP address, email address, user id, feed name, or search text is included in them, and nothing in them can be traced back to an individual.

The bitmagnet crawler

The bitmagnet crawler that answers searches runs on separate hardware in the operator's homelab, on the same private network as this site's server. When you search, or when a saved feed or Torznab client asks for results, only the search itself (the query and its filters) is sent to it. Your IP address, cookies, and other request headers are never passed along, so the crawler does not learn who asked.

To match torrents to films and series, the crawler looks titles up on TMDB (The Movie Database) with the operator's own API key. Those lookups come from the crawler, not from your browser, and contain nothing about visitors.

Other people on the BitTorrent network

This section is for people who use BitTorrent, whether or not they ever visit this site. To build its catalogue, the crawler takes part in the DHT like any other node, and so it handles the IP addresses and ports of other people on the network:

  • DHT nodes that contact the crawler, or that it contacts,
  • peers that other nodes announce to it for a torrent,
  • peers it connects to in order to fetch a torrent's metadata.

The crawler holds these addresses in memory only, and loses them when it restarts. They are not written to its database, and none of them reach this site, so search results, feeds, and the Torznab API contain no IP addresses. As the DHT protocol requires, the crawler passes some of these addresses on to other nodes that ask for them, the same as every other DHT node does. It writes addresses to its logs only when debug logging is switched on, and the operator runs it with debug logging off.

Where the GDPR or UK GDPR applies, an IP address can be personal data. We process these addresses on the basis of our legitimate interest in running the crawler that builds this index, which can't take part in the DHT without exchanging them (Article 6(1)(f)). The processing is limited to what the protocol exchanges anyway, and none of it is kept. We get these addresses from the network rather than from you, and have no way of contacting the people behind them, so this public page is how we give you the information Article 14 requires. Because we don't store anything that would let us pick out your address, we usually can't find any data about you to act on a request (Article 11). You can still write to support@torrents.observer, and you can complain to a supervisory authority for the country you live in.

Anti-bot check

When you register, or if you search unusually fast, you may be asked to solve a short proof-of-work check. Your browser computes it locally. Nothing is sent to a third-party captcha service, no such service is contacted at any point, and the check needs no account and sets no cookie.

Copyright notices and other complaints

If you send a copyright notice, a dispute, or another complaint under the copyright and takedown policy, we keep your message and our replies. They usually contain your name, email address, and any other contact details you included. We keep them as a record of why a record was hidden or unhidden, so we can handle disputes and follow-up notices, and so we can respond if a notice turns out to have been sent in bad faith. They arrive by email and are not stored in this app's database. There is no automatic deletion of them.

When we hide an info hash, the app's database stores the hash, the time it was hidden, and an optional note written by the operator. That note is for our own record-keeping and can include details taken from the notice, such as the complainant's name or a reference to their message. It is never shown on the site. The hash, time, and note are kept for as long as the hash stays hidden, and are deleted from the database when the operator unhides it. There is no other time limit. Once it is unhidden, copies of the hash, time, and note, including any name in the note, remain in the encrypted backups for up to 30 days.

Who we share data with

Nobody. There are no third-party processors, no advertising or analytics partners, and no data sharing or selling of any kind, because the service has no integration through which it could do so. The backups described below go to storage the operator runs on their own hardware, so no outside company holds them either. On the BitTorrent side, the crawler passes contact addresses on to other DHT nodes as the protocol requires, as described under "Other people on the BitTorrent network", but nothing about visitors to this site goes with them. Otherwise, the only exception would be a legally binding demand from a competent authority, which the operator would have to comply with like anyone else.

Where it's stored

Account data and the list of hidden info hashes live in a single SQLite file on the server torrents.observer runs. There is no external database, and no third-party cloud storage service is involved. Copyright notices and other correspondence are kept as email, outside the app.

The crawler's catalogue lives in its own PostgreSQL database, on the crawler's machine in the operator's homelab. It holds torrent metadata only. Nothing about visitors or account holders is in it, and neither are the IP addresses of BitTorrent users.

Once a day, a full copy of that database file is taken as a backup, so the service can be recovered if the server's disk fails. Each copy is encrypted on the server before it leaves, and is then stored on a separate machine that the operator runs on their own hardware. A backup contains everything in the database at the moment it was taken: account records, session records, saved feeds, and the hidden info hashes with their notes. Each backup is deleted automatically 30 days after it was taken.

How long we keep it

The periods below are for the live database. Anything stored in the database also stays in the encrypted daily backups for up to 30 days after it is deleted or changed there.

  • Account record: until you ask us to delete it, plus up to 30 days in backups.
  • Saved feeds: until you delete them, or your account is deleted, plus up to 30 days in backups.
  • Sessions: they stop working after 30 days, or when you log out. The database record can outlive that (see "Sessions" above), plus up to 30 days in backups.
  • Backups: each daily copy of the database is deleted automatically 30 days after it was taken.
  • Rate-limiting counters: seconds to minutes, in memory, lost on restart.
  • Result cache: up to three minutes, in memory, lost on restart.
  • IP addresses of BitTorrent users handled by the crawler: in memory only, never written to a database, lost on restart.
  • Hidden info hashes and their notes: as long as the hash stays hidden. Deleted from the database when it is unhidden, plus up to 30 days in backups.
  • Copyright notices and related email: no fixed period and no automatic deletion.

Legal basis (for anyone in the EU/UK)

Where the GDPR or UK GDPR applies, we process your email address and password hash to perform the service you asked for, giving you an account and the feeds attached to it (Article 6(1)(b)). Rate limiting, the anti-bot check, and aggregate metrics rest on our legitimate interest in keeping a small, free service available and not overwhelmed (Article 6(1)(f)). Keeping copyright notices, disputes, and the notes stored with hidden info hashes rests on our legitimate interest in handling those complaints and being able to show how we handled them (Article 6(1)(f)). Keeping encrypted backups of the database rests on our legitimate interest in being able to recover the service after data loss (Article 6(1)(f)). No cookie beyond the sign-in session is set, which is why you aren't asked for cookie consent.

Your rights

You can ask us to show you what we hold about you, correct it, or delete it, and you can object to or ask us to restrict how we use it. Because the account record is so small, in practice this means: your email address, when you registered, and your saved feeds. If you sent us a copyright notice or other complaint, it also covers that correspondence and any note that mentions you.

There is no self-service delete button yet. Write to support@torrents.observer and the operator will handle it by hand.

When we delete something at your request, it is removed from the live database straight away. Backups can't be edited one record at a time, so older copies of it remain in the encrypted backups until those backups are deleted, which happens within 30 days. Until then, backups are only used to restore the service after data loss.

Where the GDPR or UK GDPR applies you can also complain to a supervisory authority for the country you live in.

Children

This service isn't directed at children, and we don't knowingly hold data about them. An account requires only an email address, so we have no way to tell. If you believe a child has registered, tell us at support@torrents.observer and we'll remove the account.

Changes

If this policy changes, the version on this page, with its date, is the one in force. Changes that affect what is collected will be noted here rather than made silently.

Contact

support@torrents.observer

About this document

This is practical boilerplate written from the application's actual code, not a template. It has not been reviewed by a licensed attorney and it is not legal advice. If you need it to satisfy a specific regulator or a specific jurisdiction's rules, have a lawyer there review it.