Milestone #160 Opt 2b (mood half of the era+mood option). A fourth taste
facet alongside artists + genre tags + eras: signed weights over canonical
mood buckets (melancholic / energetic / chill / …) derived from a track's
enriched folksonomy tags (#1490).
- internal/mood: shared vocabulary — Of(tags) maps folksonomy tags to
canonical mood buckets (synonyms collapse). Imported by both the taste
builder and the scorer so a track's mood is derived identically.
- Migration 0047: taste_profile_moods table + taste_tuning.mood_scale
(DEFAULT 0.5).
- Build side (internal/taste): Config.MoodScale ([0,1] damper, mirrors
EraScale); accumulate folds each play/like's mood buckets at
base*MoodScale; persist atomic-replaces the mood rows.
- Scorer (internal/recommendation): TasteProfile gains a mood term
(own tanh scale + additive 0.12 share, so it never weakens the existing
signal when a track has no mood tags). Match now takes the candidate's
mood buckets; loaded per candidate (ListTrackTagsForTracks → mood.Of) in
the primary similarity loader only — the near-whole-library fallback
pool passes nil (mood → 0) to avoid a full-library tag scan.
- Tuning lab: mood_scale threaded through recsettings + admin API + web
card ("Mood weight" row) + Go/web tests.
Coverage is partial (grows with tag enrichment; richer once Last.fm is
keyed), so mood is a supplement — neutral for tracks with no mood tags.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Milestone #160 Opt 5. A collaborative candidate arm: tracks by artists
co-played across the instance with the seed's artist.
Minstrel is a single shared-library, multi-user server (no per-user
library ACL — verified: no owner/share/group model), so the "household"
is the whole instance's user set; the rule #47 scoping is satisfied by
the shared-library boundary. Single-user servers produce no edges.
- No migration: source='user_cooccurrence' was pre-whitelisted in the
0009 similarity CHECK from day one.
- internal/db/queries/coplay.sql: Delete + Insert artist co-play edges.
Score = Jaccard of the two artists' distinct-player sets (controls for
globally-popular artists); >= 2 co-players AND Jaccard >= floor kept
(the floor also self-limits hub artists). Completed plays, 365d window.
- internal/coplay: periodic worker (6h) that atomic-replaces the
user_cooccurrence edge set from play_events — pure local SQL, no
external calls. Wired in main.go alongside the similarity worker.
- LoadRadioCandidatesV2: new coplay_artists arm (source='user_cooccurrence',
seed-artist based, 0.5 damp like similar_artists) + $11 limit;
CandidateSourceLimits.UserCoplay (default 20, For-You 40).
- Integration tests: perfect-overlap Jaccard=1.0 edge + single-user
empty-set gate.
Device axis and AcousticBrainz (Opt 4) are separately tracked; this
closes the milestone-#160 sequential options.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Milestone #160 Opt 3 (temporal half). A new additive scoring term that
boosts a candidate when its artist's play history concentrates in the
CURRENT daypart × weekday-type cell, in the user's local timezone.
- Migration 0046: recommendation_weight_profiles.context_time_weight
(per-profile scoring weight, DEFAULT 1.0).
- Query ListArtistContextPlayCountsForUser: per-artist completed-play
counts split by the current cell (daypart night[22,5)/morning[5,12)/
afternoon[12,17)/evening[17,22) × weekday-vs-weekend) via
started_at AT TIME ZONE users.timezone; 365-day window, skips excluded.
- internal/recommendation/context.go: LoadContextAffinity computes each
artist's shrunk cell-share minus the user's baseline share, clamped to
[-1,1]; sparse artists shrink toward baseline (pseudo-count 5), unknown
artists → 0 (cold-start neutral).
- Score() gains context_affinity_score · ContextTimeWeight; both
candidate loaders set it per candidate.
- Tuning lab: ContextTimeWeight threaded through recsettings + admin API
+ web card ("Time-of-day weight" row) + Go/web tests. Shipped 1.0 both
profiles (uniform start, re-bakeable).
Device-class axis deferred to #1551 (needs a client_id → device-class
mapping that doesn't exist yet).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Milestone #160 Opt 2 (era half). A third taste facet alongside artists
+ genre tags: signed weights over decade buckets ("1990s") derived from
albums.release_date, rebuilt daily and scored into the taste match.
- Migration 0045: taste_profile_eras table (mirrors taste_profile_tags)
+ taste_tuning.era_scale column (DEFAULT 0.5).
- Build side (internal/taste): Config.EraScale ([0,1] damper, mirrors
EnrichedTagScale), accumulate folds each play/like's decade at
base*EraScale, persist atomic-replaces the era rows.
- Scorer (internal/recommendation): TasteProfile gains an era term (own
tanh scale + additive 0.15 share so it never weakens the existing
artist/tag signal when a track is undated); candidate queries return
album release_date; decadeOf mirrors the builder helper.
- Tuning lab: era_scale threaded through recsettings + admin API + web
card (auto-renders the new row) + Go/web tests.
Mood facet deferred to #1534 (partial enrichment coverage + needs
candidate-side enriched-tag loading).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Promote the enriched-tag weight (#1490) from a taste.Config default into
the DB-backed tuning lab so operators can dial how much folksonomy tags
count vs raw ID3 genre (rule #25).
- Migration 0044: taste_tuning.enriched_tag_scale (DEFAULT 0.5, backfills
the existing row).
- recsettings: TasteTuning gains the field; seeded/read/updated through
reconcile + persistTaste; applyTastePatch validates it to [0,1]
(generic non-half-life clamp) and diffTaste audits it; TasteConfig maps
it into the profile build.
- API: tasteTuningResp exposes enriched_tag_scale.
- Web tuning card: a data-driven "Enriched tag weight" knob (0 = genre
only). Tests: recsettings persist+range, web fixture field.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Widen keyless coverage: when a track's recording is untagged or has no
recording MBID, fall back to the artist's tags (/ws/2/artist/{mbid}
?inc=tags), down-weighted 0.6 as a coarser signal. Recording tags still
win outright when present.
- ListTracksMissingTags returns a.mbid AS artist_mbid; TrackRef gains
ArtistMBID; the enricher threads it through.
- MB provider: recording-first, artist-fallback via a shared
fetchEntityTags helper. A transient error at the recording step is
returned (retry) rather than masked by the fallback.
Enricher, registry, and settings are untouched — the pluggable design
absorbs the wider lookup. Tests cover fallback-when-untagged,
artist-only-when-no-recording-MBID, and recording-preferred.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add tag_provider_settings to the truncation list and reset
tag_sources_meta.current_version to 1 in ResetDB, mirroring the cover-art
settings reset. Without it the migration-seeded musicbrainz/lastfm rows
leaked across tests and desynced the enabled-set signature, so a key-only
PATCH spuriously reported version_bumped=true
(TestAdminUpdateTagSource_KeyOnlyDoesNotBump).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Expose the tag-enrichment provider settings over the admin API, mirroring
the cover-sources surface so a Last.fm key can be pasted and sources
toggled from the web admin UI (rules #25/#27).
- GET /api/admin/tag-sources — list providers + version
- PATCH/api/admin/tag-sources/{id} — enable / set api_key
- POST /api/admin/tag-sources/{id}/test — test connection
- POST /api/admin/tag-sources/research — bump version, re-open
settled rows for re-enrich
- Thread tags.SettingsService through server.New (struct field, like
RecSettings) → Router → api.Mount → handlers.tagSettings; main.go sets
srv.TagSettings.
Handler tests mirror admin_cover_sources_test (list / flip-bumps-version /
key-only-no-bump / unknown-404 / non-admin-403 / not-testable-ok-false),
integration-tier (skip without MINSTREL_TEST_DATABASE_URL).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The taste recompute's tag facet now unions the cached track_tags
(MusicBrainz/Last.fm folksonomy tags) alongside raw ID3 genre, so a coarse
"Rock" gains "post-punk / shoegaze / melancholic".
- taste_profile.sql: ListPlayEngagementInputsForUser +
ListLikedTrackTasteInputsForUser now return track_id to key the
enriched-tag lookup.
- accumulate(): for each play, fold its track's enriched tags weighted by
engagement × tag.weight × EnrichedTagScale; for each liked track, by the
tag-like bonus × tag.weight × scale. A track with no cached tags
contributes genre only (graceful).
- New Config.EnrichedTagScale (default 0.5) — enriched tags augment the
ID3 signal without swamping it; 0 = genre-only. Flows through
recsettings.TasteConfig() (starts from DefaultConfig). Promoting it into
the admin tuning lab is a small follow-up.
Unit-tested the pure foldEnrichedTags helper (overlap accumulation +
scale=0 disable).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Construct the tag SettingsService + Enricher at boot (mirroring coverart:
reconcile providers, bump the sources version if the provider set changed
to re-open settled rows), then run a standalone background Worker that
drains tracks needing folksonomy tags on a periodic tick.
Standalone (not threaded through the file-scan chain like cover art)
because tag lookups need only DB fields — recording MBID / artist / title
— so it mirrors the ListenBrainz similarity worker instead: an initial
drain shortly after boot, then every 30 min, up to 200 tracks per tick.
MusicBrainz's 1 req/s ceiling is the real throttle.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Milestone #160 Option 1, Step 2. New internal/tags package that enriches
the taste profile's tag facet beyond raw ID3 genre, built so adding a
source later is "implement TrackTagProvider + Register()" — no enricher,
settings, or schema change (operator directive).
Mirrors the internal/coverart pattern:
- Provider interface + package registry (Register/AllProviders/ByID);
TrackTagProvider fetch capability + TestableProvider for the admin test.
- DB-backed SettingsService over new tag_provider_settings +
tag_sources_meta (migration 0043) — enable/key/version, boot
reconciliation, and a provider-hash bump that re-opens 'none' rows when
the compiled-in provider set changes.
- Slim self-contained httpClient (rate-limit + retry + User-Agent), kept
local so tag enrichment never depends on coverart internals.
Providers:
- MusicBrainz: keyless, default-ON, recording tags by MBID (rule #26 baseline).
- Last.fm: keyed, default-OFF, track.getTopTags by artist+track — opt-in
once a key is supplied.
Enricher uses MERGE semantics (differs from coverart's first-success-wins
for a single image): unions tags across every enabled provider, caps to
top-K by weight, and stamps tag_source musicbrainz|lastfm|mixed|none.
Writes are transactional (atomic replace of track_tags).
Unit-tested without a DB: registry mechanics, provider JSON parsing +
weight normalization via httptest, and the pure merge/top-K/source-label
helpers. Wiring (startup + on-scan), integration tests, and the taste
union (Step 3) + Settings UI (Step 4) come next.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Promote the best-performing surface ("Songs like {artist}", ~8% skip /
~86% completion) out of the shared Playlists carousel into its own Home
row on both Android and web, and widen the daily build from 3 to 6 mixes
so the dedicated row shows a wider spread.
Server (internal/playlists):
- PickSeedArtists candidate pool 5 → 12; pickSeedArtistsForDay now takes
songsLikeSeedCount (6) instead of a hardcoded 3. Graceful degradation
and daily rotation preserved.
Android (HomeScreen.kt):
- New songsLikeSection + buildSongsLikeRow; PlaylistsRow takes a title so
it renders both the "Playlists" and "Songs like…" rows. buildOnlineRow
/ orderedRealPlaylists no longer reserve the 3 songs-like slots.
Offline shows cached mixes (available-first), hides the row when none.
Web (+page.svelte):
- Dedicated "Songs like…" row from songsLikeRow; dropped the 3-slot cap
and removed songs-like from the Playlists carousel.
Tests: seed_selection_test.go, BuildPlaylistsRowTest.kt, page.test.ts.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The previous commit staged only track_tags.sql.go and left the rest of
the sqlc regen uncommitted, so HEAD referenced TrackTag / the new
tracks.tag_source columns without their definitions — a broken tree.
Adding tracks.tag_source + tag_sources_version to the tracks table
regenerated every generated file that returns/embeds the Track model
(models.go, tracks/events/history/likes/recommendation). Commit them all
together so dev HEAD compiles.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
First step of taste-profile fidelity via metadata enrichment (milestone
#160, task #1490) — no ML sidecar, operator's constraint.
The taste profile's tag facet is built purely from raw ID3 tracks.genre
(splitGenres in internal/taste/profile.go). This lands the data layer for
enriching it with track-level folksonomy tags:
- track_tags(track_id, tag, weight) — a global cache of style/mood tags,
top-K per track, weight = normalized folksonomy strength [0,1].
- tracks.tag_source / tag_sources_version — versioned enrichment
bookkeeping mirroring artists.artist_art_source (NULL = eligible,
provider name = found, 'none' = settled, version bump = re-process).
- Queries: ListTracksMissingTags (batch drainer), DeleteTrackTags +
InsertTrackTag (atomic per-track replace), SetTrackTagSource, and
ListPlayed/LikedTrackTagsForUser for the recompute to union enriched
tags into the tag facet alongside genre.
No consumer yet — the enricher (MusicBrainz + Last.fm providers,
track-level) and the taste-recompute integration land next.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The 2-week metrics review found Discover beating the manual baseline on
skip rate — which for a discovery surface means it plays it safe. On a
single-user server it's effectively dormant + crude-random, and the
per-user taste profile (taste_profile_tags, #796) went unused (#1254 gap).
Add a fourth Discover candidate bucket, ListTasteUnheardTracksForDiscover:
unheard / non-liked / non-quarantined tracks ranked by summed taste-tag
weight over tracks.genre (split like the radio tag_overlap arm), md5
tiebreak. Its picks are stamped pick_kind = 'taste_unheard' (migration
0041 widens the CHECK on playlist_tracks + play_events, rule #36).
Rebalance the slot allocation 40/30/30 → taste_unheard 35 / dormant 30 /
cross_user 20 / random 15, and lead the interleave with taste_unheard so
a track shared with another bucket keeps the taste stamp and the
targeted-novelty arm stays measurable. Metrics label "Taste-matched" +
order entry added to the single server-side pickKindLabels map, so web
and Android surface the new breakdown row with no client change.
Cold start (empty taste_profile_tags) yields an empty taste bucket that
redistributes to the survivors, so Discover still fills.
Scribe #1488. Companion review outcomes: Songs-like starvation already
fixed (#1255); For You v2 ratified as-is (fresh-injection cost disproven).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The verify half of the tune→verify loop (#1251), on the same admin
Tuning page as the knobs:
- RecommendationWeeklyTrends: weekly per-source outcomes aggregated
across all users (the knobs are global, so judging a turn needs
global outcomes — rows carry rates only, no track/user identity),
with a taste-hit count per bucket: plays whose track's artist has a
positive weight in the player's current taste profile. That's the
"cheap recompute" reading — retroactive over the whole window, at
the cost of profile drift.
- GET /api/admin/recommendation-trends?weeks=N (default 12, cap 52):
per-family weekly series (skip rate, sample-weighted completion,
taste-hit rate) plus the tuning-audit markers inside the window.
- Web: sparkline table under the tuning cards — skip rate per week on
a shared axis with dashed ticks at knob turns, latest-week columns,
window taste-hit rate, low-volume rows dimmed as anecdote, and a
plain-text list of the window's tuning changes.
Also fixes the revive unused-parameter lint on the tuning GET handler
that failed CI run 1903 on the previous commit.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
The recommendation scoring knobs move out of YAML (radio profile) and
out of the systemMixWeights hard-code (daily_mix profile) into
DB-backed settings with live effect (#1250) — the defaults-discovery
lab per decision #1247: the operator turns knobs to find good values,
which then get baked back into shipped defaults; end users and other
operators should never need the card.
- Migration 0040: recommendation_weight_profiles (radio / daily_mix,
8 weight columns), taste_tuning singleton (engagement half-life +
completion-curve points), recommendation_tuning_audit (one row per
change with a {field, old, new} diff — the trend view's markers,
#1251).
- internal/recsettings: boot reconcile seeds shipped defaults without
clobbering tuned rows (coverart SettingsService pattern), validates
patches (bounds, curve ordering), writes audit rows, and pushes
daily_mix weights + taste config into package playlists. No-op
patches write no audit row.
- playlists gains SetSystemMixWeights / SetTasteConfig swap points
under a RWMutex — no signature threading through the producers; the
scheduler's taste rebuild reads the pushed config.
- Radio reads its weight profile from the service per request; the 8
weight fields leave config.RecommendationConfig (YAML keeps only
RecentlyPlayedHours / RadioSize / RadioSizeMax).
- Admin API: GET/PATCH/reset under /api/admin/recommendation-tuning,
echoing current + shipped values.
- Web: new admin Tuning tab — two weight profiles side by side, taste
card, per-scope save (changed fields only) + reset, deviation dots
against shipped defaults.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Two approved composition changes (#1269), mechanism only — the
taste/fresh share stays data-decided (#1252) and pick_kind
attribution is unchanged.
Multi-seed blending: each day's build now seeds from up to 3 of the
user's top-5 tracks (pickDailySeeds, the generalized daily shuffle)
instead of one rotating anchor, so the mix spans neighborhoods within
a day and stops feeling bipolar as the rotation swings between
dissimilar seeds. Per-seed pools merge first-seen-deduped; the head
is filled best-first under 50/30/20 per-seed quotas (60/40 for two
seeds) so one neighborhood can't monopolize it, with thin-seed quota
spilling best-first.
Score-weighted fresh tail: the tail sample (rank 2*headN onward) was
uniform — the 380th-best candidate as likely as the 101st. It now
uses deterministic Efraimidis-Spirakis keys with weight halving every
50 ranks, so freshness keeps its "you'll probably enjoy this" half
while still rotating daily.
The retired single-seed picker's one other caller, You-might-like,
moves to pickDailySeeds(n=1) — a single neighborhood per day is right
for a short shelf, and the behavior note is inline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Both mixes move from a single hard eligibility rule to the tiered
ladder, with their tier stamped onto playlist_tracks.pick_kind via the
#1270 provenance pipeline.
New for you (#1267) — consume on play, degrade by stepping back:
- "Consumed" = any track attempted >=30s; played albums leave the mix
at the next build instead of crowding it until the calendar window
expires.
- Tier 1: unconsumed albums added <30d by direct-affinity artists.
Tier 2: unconsumed affinity albums from the wider 30-90d window —
added while you weren't looking. Tier 3: any unconsumed album added
<90d, newest first.
First listens (#1268) — track-level "attempted" threshold:
- A 2-second accidental brush no longer disqualifies a whole album;
"attempted" is duration_played_ms >= 30000 per track.
- Tier 1: albums with zero attempted tracks. Tier 2: barely-attempted
albums (<=25% of tracks reached 30s), minus the attempted tracks
themselves. The artist-affinity ordering signal also moves to the
>=30s definition so skip-only contact doesn't read as trust.
Producer plumbing: fetch adapters map the tier column onto pick kinds,
finishMix propagates PickKind into the persisted candidates, and
rotateForDay now rotates within contiguous same-pick-kind blocks so
daily rotation can't hoist tier-3 filler above tier-1's exact fits
(untiered pools are one block — original behavior).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Three discovery-mix defects from the intent audit (Scribe note #1254),
all sharing the same root pattern — skips treated as non-events:
- Deep Cuts (#1257): eligibility counted only unskipped plays, so a
track skipped twice with zero completed listens read as "barely
heard" and kept being re-offered. Tracks with >=2 skips no longer
qualify; a single accidental skip doesn't banish.
- Rediscover (#1258): a skip on a rediscover-sourced play — the user
explicitly declining the resurfacing invitation — changed nothing,
so declined tracks re-qualified the next day forever. Such tracks
now sit out 90 days.
- On This Day (#1256): day-of-year distance used plain ABS, so
Dec 28 vs Jan 3 read as 359 days apart and the window silently
gutted itself for ~3 weeks around every New Year. Now circular
(LEAST(d, 365-d)), anchored on the build-date parameter instead of
now() so it's testable and consistent with the mix's daily
determinism.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
PickSeedArtists had a hard 7-day window with no fallback: a week
without listening emptied the seed pool, produceSeedMixes returned
zero playlists, and the daily atomic-replace build deleted every
existing "Songs like X" mix until the user played something again
(#1255).
The query now falls back through widening engagement windows — 7d →
30d → all-time → liked artists — the same tiered shape that fixed the
identical vanish for For You's seeds (PickTopPlayedTracksForUser).
Like-boost scoring is preserved in every tier.
All returned rows share the winning tier, and produceSeedMixes maps it
onto the rule-#131 pick-kind ladder (7d = tier1 exact, 30d = tier2,
all-time/liked = tier3) and stamps the built tracks — the #1270
provenance pipeline then attributes plays and skips to seed freshness,
so the metrics card can say whether stale-seeded mixes actually
perform worse.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
The #1249 mechanism (stamp WHY a track is in the snapshot at build
time, freeze it onto the play at ingestion, break it down in metrics)
generalizes from a For You one-off to the standard for every system
mix (#1270):
- Migration 0039 widens both pick_kind CHECKs (drop + re-add in the
same change) to taste/fresh + Discover's dormant/cross_user/random
+ tier1-3 for the rule-#131 eligibility ladders.
- GetForYouPickKindForTrack becomes GetSystemPickKindForTrack
(user, variant, track); ingestion stamps any systemPlaylistSources
play from its own variant's live snapshot, live + offline paths.
- Discover stamps its candidate bucket on discoverTrack before the
interleave, making the 40/30/30 allocation measurable; dedup keeps
the taking bucket's stamp.
- Metrics replace the for_you special-case with one pick-kind
vocabulary — any family with attributed plays gets a breakdown,
future stamping mixes need no metrics change.
- Web: breakdown sub-rows are now toggled per surface (collapsed by
default) so eight stamping mixes don't swamp the card.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Milestone #127 step 2 (#1249). For You deliberately blends two
populations — a head of top-scored taste picks and a tail sampled from
deeper ranking (the freshness injection) — but the metrics judged it as
one blob, so its skip rate couldn't distinguish "the taste engine is
missing" from "the freshness tax is too high". That number decides the
exploration share before we tune it.
- Migration 0038: nullable pick_kind ('taste'|'fresh') on both
playlist_tracks (stamped at snapshot build) and play_events (frozen at
play-ingestion — the snapshot rebuilds daily, so attribution cannot be
reconstructed at read time).
- Builder: pickHeadAndTail marks head=taste / tail=fresh; the small-pool
fallback is all taste (top-N-by-score IS the taste mechanism). Other
variants persist NULL.
- Ingestion: for_you plays (live + offline replay) look the track up in
the user's current snapshot; not found → unattributed, never guessed.
- Metrics: For You's row gains a breakdown (taste / fresh / earlier
unattributed plays), parent row stays the sum; web card renders the
sub-rows indented with the same baseline deltas + low-data dimming.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
The recommendation metrics table was observable but not actionable: raw
source strings (album:<uuid> one-offs) drowned the stable surfaces, and
manual plays were excluded so skip rates had no control group.
- SQL: include NULL-source rows (the baseline) and carry completion_n
so family merges can weight avg_completion correctly.
- Handler buckets raw sources into stable families (radio:<uuid> →
Radio, album:/artist: → direct plays, etc.) grouped by surface
intent: go-to / discovery / direct — each band judged against its
job, since discovery mixes are expected to skip hotter. Families
under 20 plays are flagged low-confidence, not hidden.
- Settings card renders the baseline row and per-surface deltas in
percentage points vs baseline (worse-than-baseline deltas in danger
color), intent hint copy per group, low-data rows dimmed.
- Pure-unit test for the bucketing/merge; DB test updated to the new
contract (baseline included, radio:<uuid> collapse).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Confirmed against prod: exactly one track (17 plays, cold since May 21)
met the c>=5 + 30d-cold bar, and three process defects turned that into
a 1-track playlist instead of the locked placeholder.
- ListRediscoverTracks: collapse the two-tier UNION into one blended
pool. The old shallow-tier gate (WHERE NOT EXISTS deep) was
all-or-nothing — one 6-month row suppressed the entire 30-day tier —
and deep was a strict subset of shallow anyway. Eligibility drops to
>=3 non-skip plays (on a weeks-old history the >=5-play tracks are
precisely the ones still in rotation); ordering prefers >=6mo cold,
then >=5 plays, then raw count.
- Minimum viable mix floor for all five discovery mixes: below
minLen (15; 5 for the album-coherent NewForYou/FirstListens) the
variant is withheld so Home renders the 'listen more to unlock'
placeholder instead of a mix that reads as built-wrong.
- /api/events: clamp client-supplied 'at' to [user.created_at,
now+5m]. Unbounded client clocks could write arbitrarily old plays
and poison the 6-month ordering (prod data verified clean — no
scrub needed).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TsF3cNoKrqCYsU78cXC8U6
Relabel (#1204): route + player_state events fire for every output route,
not just UPnP — split them into a new 'playback' kind; 'upnp_sync' now
means genuinely UPnP/Sonos signal (drops, resync). Migration 0037 adds
'playback' to the kind CHECK; server whitelist, Android reporter labels,
and the web kind filter updated.
Web sort: the diagnostics list gains a Newest/Oldest-first sort (default
newest at top); export follows the displayed order.
Fix (#1205): OutputRoute.isConnected was derived from RouteInfo.connectionState,
which stays DISCONNECTED for local SYSTEM routes even when active — so a
connected Bluetooth device showed "Available" and reported connected:false.
The picker subtitle now uses isSelected (route == selected route); the dead
isConnected field is removed and the misleading `connected` field dropped
from the diagnostics route event (it only ever logs the active route).
Refs Scribe M9 (#119), tasks #1204#1205.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K55iTxn95BtshocgdE1shW
New diagnostic_events table + per-account users.debug_mode_enabled flag.
When an account's flag is on, its client(s) POST a batch timeseries of
connectivity / UPnP-sync / power / lifecycle events to /api/diagnostics
(no-op 204 when off, kind whitelist mirrors the CHECK constraint).
Admin surface: GET /api/admin/diagnostics (optional account/device/kind/
time-window filters, RFC3339-or-epoch-ms, export-sized paging) + a
/diagnostics/devices overview + PUT /api/admin/users/{id}/debug-mode to
flip an account remotely while a bug is live. debug_mode_enabled is now
exposed on /api/me (client gate) and the admin user views.
Retention: a 30-day gc-worker sweep (GcPruneDiagnostics), keyed on the
server clock so a skewed device clock can't keep rows alive.
Refs Scribe M9 (#119), tasks #1172#1173.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K55iTxn95BtshocgdE1shW
The daily 03:00 scheduler rebuild (and the manual refresh endpoint) replace
a user's system playlists + You-might-like rows but published no event, so a
client left open across the rebuild served yesterday's snapshot until a
manual reload — the stale-tab case behind #968. Add a user-scoped
playlist.system_rebuilt event (envelope {kind,user_id,data:{}}) from both the
scheduler (bus threaded into NewScheduler) and handleSystemPlaylistRefresh.
Clients consume it to invalidate home / system-playlist views and proactively
re-pull a stale active queue. Issue #968.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make a fresh install usable out of the box and document the first-run flow
for the public-facing repo.
- Default scan_on_startup to true (Default() + config.example.yaml, which is
the live config baked into the image). Previously false, so a fresh stack
came up with an empty library and no hint to scan. Scans are incremental
(mtime skip), so the per-restart cost is just a directory walk. Re-point
the env-override test to exercise the override against the new default.
- README: add a "First run" walkthrough (register -> scan -> integrations ->
install Android app -> invite users), each grounded in a real route.
- Add docs/screenshots/ with six captures, referenced via width-constrained
<img> wrapped in a link (shrink inline + click to open full size).
API token and invite token were cropped/redacted out of the captures
before commit so no live credential lands in the public history.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per-source play outcomes so the operator can see whether each recommendation
surface is landing and tune the now-operator-tunable taste weights.
Server:
- query RecommendationSourceMetricsForUser: groups the user's play_events by
source (system-playlist surface), reporting plays / skips / avg completion
over a window; NULL-source (library/radio) plays excluded.
- GET /api/me/recommendation-metrics?days=30 (default 30, capped 365) →
{window_days, sources:[{source, plays, skips, skip_rate, avg_completion}]}.
- handler test: 401 unauth; per-source aggregation + NULL-source exclusion +
skip_rate / avg_completion math.
Web:
- lib/api/metrics.ts: query + friendly source labels.
- settings page gains a "Recommendation metrics" card (table of surface / plays
/ skip rate / avg completion), with loading/error/empty states.
- settings tests mock the new query (manual subscribe-store, hoisting-safe).
Note: You-might-like plays aren't source-tagged (it's a Home row, not a system
playlist), so this covers For-You / Discover / the mixes. Tagging YML plays
would be a client follow-up.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2a re-ranks the existing pool by TasteMatch; this ensures taste-relevant tracks
ARE in the pool. Adds a 6th arm to LoadRadioCandidatesV2: in-library tracks by
the user's top positively-weighted taste-profile artists ($10 K, weight > 0,
deterministic weight-DESC,id order so it doesn't reintroduce same-day
nondeterminism). Pool-inclusion only (sim_score 0) — TasteMatch already scores
the fit. Empty for cold-start users (no profile).
- CandidateSourceLimits.TasteOverlap; default 20 (radio), 80 for For-You via
systemForYouSourceLimits.
- You-might-like deliberately sets TasteOverlap=0: it surfaces NOT-actively-
engaged artists, so flooding its pool with top-taste (mostly already-played)
artists would just feed the read-time dedup.
- Test: positive-weight artist's track enters via the arm; negative-weight one
is excluded (weight > 0). Existing pool tests unaffected (no profile seeded).
Deferred within 2b: profile-seeded For-You — marginal given the arm + TasteMatch
already inject taste broadly (top-played seed ≈ top-taste artist).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The fallback pulled artists only from explicit artist-likes (general_likes_artists),
but most users like albums and tracks far more than artists — so the artists row
still came up thin (a couple of tiles) even with a rich library, while the albums
row filled fine.
Broaden both fallbacks to "entities you've shown affinity for":
- artist fallback = explicit artist-likes ∪ artists of liked albums ∪ artists of
liked tracks.
- album fallback = explicit album-likes ∪ albums of liked tracks.
New dedicated queries (ListYouMightLike{Artist,Album}FallbackForUser) replace the
narrow Rediscover-fallback reuse; same projection so the Go layer still converts
directly. (Aliased + fully-qualified the UNION arms — sqlc merges UNION scopes,
so unqualified user_id was ambiguous across the three like tables.)
Test: 12 liked TRACKS by distinct artists, no artist-likes → the artist row now
fills from their artists (was empty before).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
golangci-lint v2 (CI-only; local is v1) flagged the field-by-field struct
literals — the fallback and you-might-like row types are identical, so convert
directly instead.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The taste roll-up surfaces top-similar albums/artists, which for a heavy
listener are mostly ones they already play — so the read-time dedup (vs Most
Played + Rediscover + Last Played) can strip the section down to a single tile
(reported on the artists row). The code was sound; the section was just starved.
Adds a read-time fallback: when a You-might-like row comes up short after dedup,
top it up from the user's LIKED artists/albums — a far larger pool than the
12-entity similarity roll-up, so the same exclusions still leave plenty. Reuses
the existing Rediscover-fallback queries (no new SQL), applies the same
exclusions (already-shown + Rediscover + Most/Last Played) so it never
duplicates a tile or suggests an actively-played entity, and is best-effort
(a query error leaves the section as-is). Takes effect immediately — no rebuild.
A cold-start user with no likes gets nothing from the fallback, so the
new-user-empty behaviour is preserved (test still passes).
Test: 20 liked artists, none played → Rediscover fills 10, You-might-like
fallback fills the other 10, disjoint.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The profile built in phase 1 now changes what gets surfaced. Adds a TasteMatch
term to the weighted-shuffle score so candidates are re-ranked by their fit to
the user's learned taste (positive draws toward it; negative reflects passive
avoidance; 0 at cold start).
- recommendation/score.go: ScoringInputs.TasteMatchScore ([-1,+1]) +
ScoringWeights.TasteWeight + the term in Score.
- recommendation/taste.go: LoadTasteProfile reads the taste_profile_* tables;
TasteProfile.Match blends the candidate's artist weight (0.7) and avg genre-tag
weight (0.3), each tanh-squashed by a fixed scale so one outlier artist can't
compress the rest. Unknown artist/tags and empty profiles → 0 (neutral).
- candidates.go: both candidate loaders set TasteMatchScore per candidate, so
every Score caller (system playlists incl. You-might-like, radio) becomes
taste-aware automatically.
- weights: systemMixWeights.TasteWeight = 1.5 (daily mixes are the primary
taste surface); config.RecommendationConfig gains taste_weight (default 1.0,
lighter — radio is seed-directed) wired into the radio handler.
- tests: pure (Match curve incl. saturation/clamp/empty-neutral, Score term
add+subtract) + DB round-trip (seed taste rows → Match positive). All green
vs real Postgres; existing playlist/radio tests unaffected (empty profile →
zero taste effect).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
golangci-lint v2 (CI-only; local is v1) flagged two unused-parameter issues:
- BuildTasteProfile's `now` was genuinely dead — decay/windowing are computed
DB-side via now(), so no Go-side timestamp is threaded. Removed it (a
phase-3 context model that needs a pinned reference time would re-add it);
updated the scheduler call site.
- the degenerate-params engagement test ignored t; reworked it to assert the
result stays in [-1,1], which also strengthens the test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Build a persistent, decaying model of each user's taste, recomputed daily,
that later phases consume across every recommendation surface. Phase 1 only
BUILDS the object — no behaviour change to what's surfaced yet.
Core mechanic — graded engagement (replaces binary was_skipped for learning;
was_skipped stays for History): a play's completion ratio maps to a signal in
[-1,+1] via two linear ramps (instant-skip → -1, ~0.30 neutral, ≥0.90 → +1).
Time-decayed (half-life ~75d) so recent behaviour dominates and the profile
tracks drift.
Per operator constraints:
- No explicit dislike button — negatives come only from passive behaviour
(early skips). Nothing recorded to regret or opt out of.
- Negatives are track-scoped; artist/tag weight is the decayed SUM of their
tracks' engagement, so one skip nets out against many good plays (a
DB test asserts a liked artist stays positive despite an early-skipped
track). A floor clamp bounds how negative any single entity can get.
- migration 0035: taste_profile_artists / taste_profile_tags (signed weight,
indexed by (user, weight DESC)).
- internal/taste: engagement.go (pure curve + decay) + profile.go
(accumulate plays + like bonuses, floor damping, size caps, atomic-replace).
- scheduler: rebuildUserDaily recomputes the profile before the playlist
build (so phase 2 can read it), best-effort — a taste failure never blocks
playlist building. Wired into the daily job + startup catch-up only (not
manual/lazy rebuilds).
- tests: pure (engagement curve, decay, ranking, floor, genre split) +
DB-backed (positive/negative weights, aggregation-protects-artist, like
bonus, atomic replace). All green vs real Postgres.
Config knobs live in taste.DefaultConfig() for now; wiring them into the
server RecommendationConfig is a later follow-up.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
scoreAndSortCandidates drew per-candidate jitter by slice position, but
the candidate query (LoadRadioCandidatesV2) has ORDER BY random() arms and
no stable outer ordering, so DB row order varies call-to-call. When the
recency spread between candidates is smaller than the ±jitter (small or
recency-clustered libraries), two same-day rebuilds assigned jitter to
different tracks and reordered near-ties — so the build was not actually
deterministic-within-a-day as documented.
Pre-existing latent flake in TestBuildSystemPlaylists_DailyNonceDeterminism
(passed in isolation / by luck in CI; deterministically reproduced when the
system-build tests run in sequence). Confirmed independent of the
You-might-like change by neutralizing buildYouMightLike — the flake
persisted.
Fix: sort the candidate slice by track id before assigning jitter, so the
jitter for a track is a function of (track, day) alone, independent of DB
return order. Verified: full playlists package green 4/4 and the build-test
sequence green 5/5 (was 0/4 before).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Surface in-library albums/artists the listener doesn't actively spin but
is predicted to enjoy, derived from the same similarity + like-weighted
candidate engine that powers For-You — rolled up from track scores to
album/artist granularity. Built in the daily 3am BuildSystemPlaylists
pass, atomic-replaced alongside the system playlists, and read back by
/api/home (+ /api/home/index).
Cold-start gate: skips generation entirely below 20 distinct unskipped
tracks AND 5 distinct artists, so a thin profile ships empty rows rather
than near-random tiles.
- migration 0034: you_might_like_albums / you_might_like_artists (id+rank,
CASCADE, per-user rank index).
- playlists/you_might_like.go: cold-start gate + similarity roll-up
(sum-of-top-3 aggregation, per-artist album cap, daily-rotating via the
same userIDHash jitter as For-You) + atomic-replace persist in the tx.
- recommendation/home.go: two new HomePayload sections with read-time
cross-section dedup vs Most Played / Rediscover / Last Played, trimmed
to 10 each.
- api: you_might_like_albums / you_might_like_artists on /api/home and
/api/home/index, reusing albumRefFrom / artistRefFromCovered.
- tests: pure roll-up/aggregation/cap unit tests + DB-backed gate,
sufficiency, and atomic-replace tests (all green vs real Postgres).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Audit of every Android↔server connection point (2026-06-11) cleared the
silent-contract class that caused the events `type` bug, but surfaced a
cluster of offline/playback-robustness defects. Fixes:
1. Playback double-count on background. PlayEventsReporter no longer
enqueues a partial play_offline + leaves the live row open on every
screen-lock. closeCurrent() now routes by whether the server has an
open row: close-by-id (durable PLAY_ENDED on failure) when it does,
offline only when no row exists, and a no-op while a play_started is
in flight (the server auto-closes that orphan). onStop only durably
closes a *paused* play — a still-playing one is left to the live path
under the foreground service. Adds the PLAY_ENDED mutation kind.
2. Replayer poison rows. MutationReplayer now classifies each replay as
SENT / DROP / RETRY: permanent 4xx (and corrupt payloads) are dropped
instead of retried forever; 408/429/5xx/transport still retry.
3. Offline-play / close-by-id idempotency (server). RecordOfflinePlay
dedups on (user, track, started_at); RecordPlayEnded skips a second
skip_events insert when re-closing an already-ended row. Makes the
at-least-once replay safe against lost-response duplicates.
4. Like-toggle collapse. Replayer drops like-toggles superseded by a
later toggle for the same entity, so partial-failure + differential
retry can't invert the final like state.
5. Connectivity-return trigger. MutationReplayer + SyncController now
also drain/sync when NetworkStatusController recovers to Healthy, so
an offline→online transition mid-session doesn't wait for a cold
start. SyncController.syncSafe gains a single-in-flight mutex.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
autoClosePriorOpen hardcoded was_skipped=true for every orphaned
play_event (a play_started whose play_ended never arrived, e.g. the
client backgrounded mid-track). That hid fully-listened tracks from
History — a play that sat open past its own length was capped to the
track duration (ratio ~1) yet still flagged skipped. Observed live:
History showed 3 plays for a day of listening because most rows were
auto-closed orphans marked skipped.
Now the auto-close applies the same skip rule as RecordPlayEnded to the
duration-capped elapsed estimate: ratio >= threshold OR elapsed >= the
duration floor -> a real play that lands in History; a genuine
quick-abandon still classifies as a skip. Still writes no skip_events
row, so the ambiguous auto-close never feeds the skip-ratio /
recommendation signal.
This is the server half. The client-side root cause (backgrounded
track transitions never closed, orphaning the rows in the first place)
is tracked separately.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
GET /api/artists/{id}/similar — in-library artists ranked by similarity
score (deduped across sources), ArtistRef list with cover + album count.
GET /api/artists/{id}/top-tracks — current user's most-played tracks for
the artist (skips excluded, quarantine filtered).
Wire the fsnotify watcher and a fixed 12h safety-net delta walk in main;
remove the configurable scan scheduler (scheduler.go, scan_schedule table via
migration 0033, GET/PATCH /api/admin/scan/schedule, and the server/api
plumbing). Manual scan + scan status are unchanged.
Add Scanner.ScanFiles (watcher-driven targeted scan returning changed album
IDs) and a recursive fsnotify Watcher that debounces filesystem events and
enriches just the affected albums inline. Pure classifyEvent/drainPending
seams unit-tested; ScanFiles covered in the scanner integration test.
Closes Scribe #614, #615, server half of #616 surfaced by the 2026-06-04 divergent-provider audit.
- streamURL helper now used everywhere /api/tracks/{id}/stream is built (was inline concat in playlists.go and cast_token.go); add streamURLWithExt for the .ext cast variant.
- audioContentType in media.go is the canonical file_format -> MIME lookup; mimeForFormat in cast_token.go is now a thin wrapper that overrides the unknown-format fallback to audio/mpeg (Sonos rejects octet-stream). Adds mpeg/vorbis/wave aliases. Subsonic's contentTypeForFormat stays frozen per docs.
- coverart.ResolveAlbumPath extracted; api and subsonic both delegate to it.
bvandeusen