LLION WITH A MICPLATFORM PLAN

i. Current State of the Pool

ii. Tier Strategy

Three-tier funnel to get from 740 raw videos to ~150 platform-ready.

TIER 1 · Score ⩾ 9 · DONE — auto-flagged platform=true

All 36 score-⩾9 videos + all 18 score-⩾9 audio (the canonical mix of each of the 18 top songs) are flagged platform=true. These are the unambiguous marquee pieces: 6 Music Videos, 7 long-form Reels, 11 Music Reel Takes (one per song), 12 Topical Monologues (the manifesto pieces). Nothing about these requires human judgment — they're all hand-scored 9 or 10 already.

TIER 2 · Score 8 · ~96 to pick from 540 → atom-eye review tomorrow

540 videos sit at score 8 — solid but not flagship. To get to 150 platform total, we need ~96 more. Proposed selection logic:

• Cap per project at 8. 41 song-projects × up to 8 video assets each = max 328. Already over budget so further cuts needed.
• Prioritize Music Reels (Polished Export) over Reel (Take). Polished = finished cuts, Take = raw studio. Polished gets in first.
• Topical Monologue (long-form) before Monologue (Topical · Reel). Long-form goes deeper.
• Distribute across projects. Don't let any one song eat the platform.
• One BTS / process piece per song max. Keep behind-the-scenes lean.

TIER 3 · Score 7 + Score 6 · NOT for the platform

414 score-7 + 103 score-6 videos stay in the library but don't ship to the official site. Searchable through /web/index.html for atom-eye's own use, not user-facing.

iii. Proposed Playlists / Categories for lionwithamic.com/theatre

Six playlists organized by what the user is trying to do when they land. Music Video / Reel / Monologue are not great categories from the listener's POV — the listener wants "play me the political ones," "show me the new song," etc. Reorganize accordingly.

iv. The Tier 1 Auto-Flagged Set · Confirm These First

These 36 videos + 18 audio are platform-flagged right now. atom-eye should click through each at /web/index.html (★ PLATFORM filter chip in the toolbar) and approve or unflag any that don't belong. Default tomorrow: if no changes, these are shipping as-is.

Score 10 Videos (14 · the marquee)

Score 9 Videos (22)

v. media.json — What atom-eye's Official Site Will Read

The site at lionwithamic.com/theatre pulls a media.json feed. Schema proposal (1 endpoint per playlist):

{
  "generated_at": 1781110000,
  "playlists": {
    "theatre":     { "title": "Theatre / Marquee", "slugs": ["legacy-reel-full","disgusted","tjlu-vertical",...] },
    "manifesto":   { "title": "Manifesto Pieces", "slugs": ["modern-colonialism-youtube-updated",...] },
    "wild":        { "title": "Songs in the Wild", "slugs": ["legacy-reel-7","savage-5","reels-06",...] },
    "daily":       { "title": "Daily Reels", "slugs": [...] },
    "bts":         { "title": "Behind the Curtain", "slugs": [...] },
    "anthems":     { "title": "The Lion Anthems", "slugs": ["radheya-acapella-wav","wia-09-master-acapella-mp3",...] }
  },
  "items": {
    "legacy-reel-full": {
      "title": "Legacy / Reel Full Length",
      "hls_url": "https://dropbox-fresh-url-here.m3u8",
      "duration_sec": 142.67,
      "thumb": "/library/thumbnails/legacy-reel-full.jpg",
      "transcript": "/library/transcripts/legacy-reel-full.txt",
      "song_project": "Legacy",
      "songs_page_anchor": "/web/songs.html#legacy"
    },
    ...
  }
}

I can ship a generator script build_media_json.py that pulls every platform=true card, bucketed into the 6 playlists by content_type + heuristics, with fresh dropbox HLS URLs auto-refreshed daily. Output written to /var/www/lwm-agi/data/media.json + served at /media.json. Drop-in for atom-eye's site code.

vi. Open Decisions for atom-eye (Tomorrow)

Three things I need you to decide before I ship the official-site feed:

• Playlist names + ordering. Are the 6 above the right buckets? Want a different cut (e.g., "Anti-War" / "Identity" / "Hope" as theme-pivot vs format-pivot)?
• Hard cap or soft cap on the 150? Strict 150 or "up to 200 if the quality is there"?
• Daily auto-refresh of HLS URLs. Want me to set a daily cron via /schedule that re-runs refresh_urls.py + build_media_json.py so the official site never breaks on expired Dropbox links?

vii. Same Plan for Master Zi — Tomorrow

Zi pipeline: 536/624 video transcripts done (88 still in flight via 4 background whisper workers). When Zi transcripts hit 100%, same plan: tier-1 auto-flag (already have score-10 / score-9 entries from the hand-curation), build /mz/web/platform.html, ship media.json for masterzi.com/theatre. Playlist proposal for Zi will lean toward: Marquee Lessons (Century Project) · Daily Dharma (EveryDayTao + EveryDayVedas) · The Long Talks (Dispassionate Observer) · Practice (Qi Gong + Tai Chi forms) · DPO Series · Origin Stories.