I have a lot of records. Discogs is where I track them. But browsing Discogs to look at your own collection is slow, and the interface is cluttered.
So I built a local version.
What it actually does
It imports your entire Discogs collection into a local SQLite database and caches the cover art. Once that is done, browsing is instant. Full-text search, filter by artist, genre, decade, or format, and statistics across the whole collection. No rate limits. No waiting.
There is a rate-aware HTTP client that reads Discogs’ own response headers and backs off when needed. Cover images are throttled to 1 request per second, with a daily cap of 1,000, which keeps you well within their ToS.
The interesting parts
I added a barcode matcher that cross-references your records against the Apple Music Developer API. When a match exists, you get an embedded player right in the collection view. This works better than you would expect; a surprising percentage of records have a clean barcode match.
There is also a Claude-powered recommendation layer. It reads your collection stats, what you own a lot of, what eras you lean toward, what you have rated highly, and suggests records you do not have yet. It is genuinely useful. It surfaces things I would actually want to own, not just popular records in the same genre.
The static export
This is the part I am most pleased with. You can export your entire collection as a standalone static website — no server, no database, just files. A version of this lives on the projects page here.
Why local-first
The Discogs API is great, but it is someone else's infrastructure. I wanted something that runs on my machine, stores data I control, and works offline. This does that.
The repo is on GitHub if you want to adapt it to your own collection.