Bitmagnet is a self-hosted BitTorrent indexer that crawls the BitTorrent DHT itself, classifies what it discovers and serves it back as a searchable catalogue with a browser UI. Rather than depending on public trackers or third-party index sites, it harvests info hashes from the DHT, fetches the metadata behind each one, then enriches the result with content type, language, resolution, source and data pulled from The Movie Database.

It runs as a single Docker container alongside PostgreSQL and exposes three interfaces at once: an Angular web UI, a GraphQL API with an embedded playground, and a Torznab-compatible endpoint, so the Servarr stack can query a personal instance exactly like any other indexer. Upstream is explicit that the project is alpha software.

Key Features

• DHT crawler — a native protocol implementation that walks the DHT network for info hashes and requests metadata for each one, so no external tracker or indexer is required.
• Generic BitTorrent indexer — indexes torrents from any source, not only the DHT; today that means the /import endpoint, for example to ingest the RARBG backup.
• Customisable content classifier — recognises Movies, TV Shows, Music, Books, Software, Games and more, and enriches each record with language, resolution, source (BluRay, webrip and similar) and TMDB metadata.
• Torrent search engine — filterable full-text search across everything the instance has indexed.
• GraphQL API — a search query today, with an embedded playground available at /graphql.
• Torznab endpoint — plugs into Prowlarr, Sonarr and Radarr as an indexer; bitmagnet can be added to Prowlarr directly.
• Responsive, multilingual web UI — an Angular front end with a dashboard for monitoring and administration.

Why Use It

• No dependency on public trackers. The DHT crawler is the feature nothing else quite replicates: the index is built from the network itself, so it keeps working when indexers go dark.
• Metadata, not just magnets. Classification plus TMDB enrichment turns raw info hashes into records you can actually search by title, year, quality and language.
• Drops into an existing media stack. The Torznab endpoint means Sonarr, Radarr and Prowlarr can consume it with no glue code.
• One container plus Postgres. Little to operate, with a dashboard and observability hooks built in.

Use Cases

• Building a private, self-contained torrent index that does not rely on any hosted search site.
• Adding a DHT-derived indexer to a Servarr media automation stack.
• Exploring what is actually being announced and crawled on the DHT at any moment.
• Classifying and enriching a large torrent collection with clean, queryable metadata.

Deployment

Distributed as the Docker image ghcr.io/bitmagnet-io/bitmagnet with PostgreSQL, using the upstream Docker Compose file. There is no Debian or distro package; the supported deployment path is containers. The web UI is accessed from any browser.

Licence

Open source — MIT licence. Upstream marks the project as alpha, so expect bugs and occasional API and database schema changes before a future 1.0 release. Recognising that, it is genuinely useful today and easy to trial on a spare machine or VM.