Engineer a hook-driven short-form edit plan with beat-synced retention beats

You are a senior short-form video editor who studies retention curves, not just trends.

Build me a hook-driven edit plan for one short-form vertical video (9:16) engineered to hold attention beat by beat.

Source material:
- Raw footage summary: [WHAT YOU HAVE — e.g. 'a 2-min talking-head take plus product B-roll']
- Hook angle: [THE PAYOFF PROMISE OR TENSION — e.g. 'nobody talks about this mistake']
- Platform and length: [TIKTOK / REELS / YOUTUBE SHORTS — ~30s / ~45s / ~60s]
- Target audience: [WHO]
- Music: [GENRE + TEMPO/BPM, OR 'CAPCUT TRENDING']
- Voiceover: [YES, I'LL RECORD / NO, ON-SCREEN TEXT ONLY]

Produce an edit plan with:
1. A second-by-second beat map. For each beat: timecode range, what is on screen, and what the cut must accomplish (hook, payoff, pattern interrupt, CTA). The first 3 seconds must earn the next 3.
2. A clip-cut list: for each clip note the in/out point, the single camera move allowed (push-in / pull-out / pan / tilt / static / whip), and the single on-screen action — never stack two moves or two actions in one clip. If a clip has no clear focal action, flag it for replacement.
3. Retention beats: name 2-3 moments where the edit breaks pattern (a jump cut, a zoom punch, a text pop, a sound sting) to reset attention before the typical drop-off points (~3s, ~15s).
4. Caption and text-overlay spec: where on-screen text appears, what it says, and how long it holds. Respect platform safe zones (no text below the bottom UI rail or in the right-side rail).
5. Sound-design cues: where music ducking, risers, or SFX punctuate a cut. Beat-sync every major cut to the music downbeat; note the BPM if known.
6. A CapCut execution note: which built-in tools achieve each beat (keyframe animation for push-in, Beat Sync / Match Cut, Auto Captions, Text/Sticker overlays) so the plan maps to real buttons.

Constraints:
- Every clip obeys one-camera-move + one-action. If two ideas fight in one clip, split it into two.
- No fake-energy tricks (zooms every second, overused glitch transitions). Each effect must serve the story beat.
- If the raw footage cannot support the hook (weak payoff, no coverage), say so and propose a sharper hook before writing the plan.

Output: beat map, clip-cut list, retention beats, caption spec, sound cues, CapCut execution note.

Success signal: the output is good only if every cut maps to a named beat, every clip has exactly one camera move and one action, and the first 3 seconds are engineered to earn the next 3.