Loose ends

Bookmarkable index of every diagnostic / fix-it tab. Each card shows live counts of work remaining + a link to the dedicated page. AI-built tools accumulate loose ends fast — this is where to come check what still needs human attention.

🎵 Songs needing identification

506

Audio fingerprints we couldn't auto-ID. Listen + type the title.

⚠ Song conflicts

Multiple ID services disagreed on a song title. Will populate when AudD lands.

📁 Duplicate clusters

Likely-duplicate videos detected via Chromaprint+duration.

📅 Videos missing date

Truly-undateable leftovers after the IG/YT-shortcode + ffprobe backfill. Manual entry.

🎯 Missing classifications

Videos with NULL format_focus or content_type. Auto-classifier already ran.

👤 Missing teacher attribution

408

Videos with no parsed_teacher_id AND no source_creator_id.

🖼 Suspicious thumbnails

60-random eyeball pass; click Regenerate per row to re-pick frame.

🎤 Transcripts to review

115

LLM said 'mixed' (instruction + lyrics). Pick a final kind.

📊 Coverage gaps

Concepts with only 1-3 videos. Prioritize what to download more of.

🕒 Recent ingest changes

106

Videos with file_mtime in last 14 days.

🏷 Out-of-vocab event types

110

Bulk-rename oddities like 'priv' → 'Priv' to clean facet space.

🗑 Orphan concepts + creators

104

100 concepts + 4 creators with 0 active videos.

⬇ Pending download URLs

Queued IG/YT downloads waiting for scripts/download_pending_urls.py to run.

🔒 URLs blocked by IG login

Need IG cookies (instaloader --cookies-from-browser) before re-running.

📥 Staging videos to review

216

In `1 to compress`, `2 done compressing`, `zzz` folders.

🩺 Data-quality audit (full)

→

10 DB-level integrity checks. Read-only overview.

📦 Compression compare

→

Side-by-side before/after Handbrake compression in `1 to compress`.

See also: /audit for the live data-quality report, docs/AUDIT_2026-05-03.md for the UI-interaction audit findings, and TODO_BUGS_AND_IMPROVEMENTS.md for parking-lot ideas.