Sources

Sources

How indexarr acquires releases: the native catalog, RSS/Atom, direct Torznab, Usenet/Newznab, eD2k and the sites with a dedicated parser.

Related: Compatibility, Definitions, Catalog, MirCrew, Migration.

Adding a source

Each source can have HTTP Basic credentials (HTTP username and HTTP password fields), independent of the Torznab API key. Next to each source there are Test and Run; Test all and Run all work on all enabled sources and write the outcomes to Statistics → Runs and the logs.

Catalog and adding

• Add indexer: choose a catalog definition (ExtTo, TorrentGalaxy, CorsaroBlu, Knaben, TorrentLeech, a generic one) or paste a Torznab endpoint.
• Quick add recognizes Torznab, Newznab, RSS and HTML listings.
• Discover feeds extracts the feeds declared by a page.
• Export/import JSON saves and reloads the public configuration of the indexers; secrets are never exported.
• Sources not ready yet stay queued and are not offered in the user list; the maintenance view /indexers/new?queue=1 also shows the composition of the queue.

As of 2026-10-29 the catalog contains 627 entries: 111 native definitions, 86 verified (85 torrent + 1 eD2k), 92 verified public hosts and 495 entries still to try, mostly private. Numbers by state in Catalog. Coverage of individual sites grows with the catalog; the generic Usenet path is already complete.

RSS / Atom

• Type: RSS / Atom feed
• URL: the feed
• Title, magnet, .torrent/.nzb enclosure, size, numeric category, date and any torznab:attr attributes are read.
• Entries without a magnet or download link are ignored.
• Default category assigns a Torznab category to feeds and listings that do not declare one.

Freeleech only

When the definition provides for it, the indexer form shows the Freeleech only checkbox (setting.freeleech). If the site accepts a filter in the search, the definition passes it as a parameter (for example free=1); if the site marks each row, the engine discards non-free rows before saving them and before answering the apps. The option appears only for definitions that declare it and also applies to the new-release list and to the live search.

Direct Torznab

• URL: full endpoint, for example http://indexer.local:8080/api
• Manager: Direct Torznab (or Auto)
• Categories: comma-separated Torznab ids (5000, 2000)

Usenet / Newznab

• Type: Newznab / Usenet
• URL and API key of the service
• Categories: standard Newznab ids
• The protocol is usenet; the payload is an .nzb and sending goes to SABnzbd or NZBGet.
• Native definitions for individual Usenet sites are not yet in the catalog: a source that exposes Newznab works immediately, the others stay queued.

eD2k / aMule

Definitions with protocol: ed2k publish ed2k://|file|name|size|hash|/ links. indexarr stores them, expands them per file and sends them to aMule through the Web API. The link can come from the page body or from an eD2k field; topics with multiple files produce one release per file. eD2k remains excluded from synchronization with the *arr apps.

HTML sites: ExtTo, TorrentGalaxy, CorsaroBlu

The three sites have native definitions with real fixtures and equivalence tests. The engine follows detail links to get the magnet, with limited concurrency and cache. CorsaroBlu is a listing-only source for guests: no username, password or cookie is needed to read its latest list and magnets. The site disables guest keyword search, so indexarr does not send a query. The CorsaroBlu (account) variant accepts a pasted session cookie for keyword search; it remains a draft until an account can verify it. In the UI, open Add indexer, find the CorsaroBlu card and follow Add with account login (optional) (or open /indexers/new?definition=corsaro-account). The username and password fields do not exist: the site's login is protected by reCAPTCHA, so paste the browser's Cookie request header into Session cookie, or leave it empty and press Create and import cookie: indexarr creates the indexer and shows a one-line curl command. Run that command on the machine with Firefox; it reads the local profile and sends the cookie. Since 29 October 2026 every request goes through a FlareSolverr session because a plain client receives 429/503. Limits: first page, at most 50 releases per source; if the layout changes or the site requires login, the test reports the error.

Private sources, cookies and forums

• Site access: every indexer form and overview shows the access method and its fields in a fixed order (username, password, cookie, API key, passkey). Public sources say they need no account; sources whose key is in the personal address (RSS key, Smartfeed) say so.
• HTTP Basic for private feeds and indexers.
• Session cookie instead of the login: every form-login definition run by the engine (MirCrew, ilDraGoNeRo, DDUnlimited, TorrentLeech, …) also accepts a cookie pasted from the browser. With a cookie the login form is skipped; when the site drops that session indexarr signs in with the username and password if they are set, otherwise the test reports that the pasted cookie has expired. Family connectors (NexusPHP, TorrentTrader, XBTIT) do their own login and do not offer the cookie.
• Import cookie for sites with a form login protected by Cloudflare Turnstile, such as IPTorrents: it shows a one-line command to run on the machine with Firefox; the helper reads the profile, sends the cookie and the browser User-Agent, and indexarr stores both encrypted. The indexer can be created first and the cookie imported afterwards.
• MirCrew and ilDraGoNeRo are phpBB forums with native definitions: search never thanks; the thanks to the author happens only at download time, when the magnet is unlocked. Details in MirCrew.
• DDUnlimited is an eD2k forum: search reads its alphabetical title indexes, every file of a topic becomes a release and the forum flood control is respected with a single search lane and a pause after a refusal.

Cloudflare / FlareSolverr

• Endpoint in Settings → Cloudflare / FlareSolverr.
• Recognizes cf-mitigated: challenge and Cloudflare pages on 403/503/200.
• Repeats the request through FlareSolverr for RSS and direct indexers.
• Persistent sessions are closed after session.idle_timeout (30 minutes by default); the browsers left by a stopped instance are reconciled 30 seconds after start and every two minutes, using only the ownership prefixes of indexarr.
• LAN addresses never go through the solver.
• .torrent files do not go through the solver: if the site requires clearance for the file too, use the magnet.

Jackett and Prowlarr

indexarr replaces them: they are not sources and there is no import that reads them. A URL that belongs to a manager is rejected with an explicit message. See Replacing Jackett and Prowlarr.

---

Part of the indexarr documentation.