Spotify Library Browser

An iTunes-style column browser for your Spotify library. Pick an album artist, see their albums, see the tracks — configurable Miller columns over a local index of everything you've saved.

Native .NET with Avalonia 12. No bundled browser engine, no webview.

Setting it up

You need your own Spotify app registration — the app doesn't ship a Client ID.

1. Create an app at the Spotify Developer Dashboard. Under "Which API/SDKs are you planning to use?" tick Web API and nothing else. Web Playback SDK is browser-only and unused here, and Android, iOS and Ads API aren't needed.
2. Add http://127.0.0.1:5543/callback as a redirect URI. It has to be the explicit loopback IP; Spotify no longer accepts localhost.
3. Copy the Client ID and paste it into the app on first run. The Client Secret isn't needed, since sign-in uses PKCE.

Spotify Premium is effectively required. Development-mode apps need the app owner to hold an active subscription, and Connect playback control requires Premium regardless. Since the February 2026 API changes removed product from the user object, the app can't check this up front — a non-Premium account simply falls back to handing playback to the desktop client.

Development mode allows five users, which is plenty for personal use. No extended-quota application is needed.

Running it

dotnet run --project src/SpotifyLibraryBrowser.App

On first run, paste your Client ID, sign in, then hit Quick sync to build the index.

• Quick sync walks newest-first and stops at the first thing it already knows about.
• Full rebuild walks everything, which is what notices removals.

Your data lives in %APPDATA%\SpotifyLibraryBrowser: the SQLite index, the settings, and the OAuth token (DPAPI-protected on Windows, owner-only file elsewhere).

Using it

Each column has a dropdown to choose what it groups by — Artist, Album Artist, Album, Year, Decade or Playlist. Add and remove columns with the + and − buttons; the layout persists.

• Selecting in a column filters everything to its right, and the track list below.
• Multi-select within a column widens the filter rather than narrowing it.
• The All row at the head of each column clears that column's filter.
• Double-click a track to play it in its album's context on the selected Connect device.

Shortcut	Action
Ctrl+L	Like the selected tracks
Ctrl+Shift+L	Unlike the selected tracks
Ctrl+Z	Undo the last unlike
F5	Quick sync

Liking writes straight through to Spotify. A bulk unlike is undoable rather than confirmed up front, and liked state re-checks itself against Spotify whenever the window regains focus, so a like made on your phone shows up here.

How it's built

src/SpotifyLibraryBrowser.Core/   UI-free: auth, sync, index, browse engine, playback, liking
src/SpotifyLibraryBrowser.App/    Avalonia 12 desktop UI
tests/                            xUnit, in-memory SQLite

Browsing never touches the Web API — every query runs against the local SQLite index, which is what keeps the columns instant. Because columns are user-configurable, the grouping is data-driven: each criterion is described once in CriterionRegistry as a SQL key/display/join/sort fragment, and BrowseQueryBuilder composes the active set into a single query.

Deliberate constraints

The app reads only fields that are current in the Web API — nothing deprecated. That rules out genres, popularity, followers, label, preview_url, available_markets, external_ids and linked_from. Two consequences worth knowing:

• There's no Genre column. Spotify's only genre data is the deprecated artist-level genres field. Year and Decade, from the non-deprecated release_date, cover that ground instead.
• No per-artist API calls happen at all. Artists arrive embedded in album, track and followed-artist responses. This matters because GET /artists (batch) was removed in February 2026, so anything needing artist detail would cost one request per artist. The upshot is that artist images only exist for artists you follow — album art carries the visual weight.

Playlist contents are only readable for playlists you own or collaborate on, so Spotify-curated playlists are recorded by name and skipped. Unchanged playlists are skipped on re-sync via their snapshot_id, which is the single biggest saving on a repeat run.

Tests

dotnet test

The browse engine carries the real risk — a wrong join gives plausible but incorrect results — so the tests assert exact contents: compilations grouping under their album artist rather than their track artists, collaborations appearing under every credited artist, year-only release precision landing in the right bucket, undated albums staying reachable through an "Unknown" bucket, and many-to-many joins not inflating the track list. The like path is covered for optimistic rollback and for chunking selections above the API's 50-URI cap.