descript-export-flowlisted

# Descript Export Flow ## When to Use Load this skill when: - You just exported a vertical short from Descript and need to publish-ready it - The first frame is black (Descript Chapter Title default) and needs to be replaced with a real cover - The audio sounds quiet (Descript exports often land around -21 LUFS, well below the -14 LUFS platform target) - The script reads slightly slowly and you want a 1.2x speed bump to land in the short-video pacing sweet spot Do NOT use for: - Long-form YouTube — those have different loudness targets (-14 LUFS works but the speed-up step usually doesn't) - Files you haven't seen / measured — always probe first, don't blindly apply ## Composes With - `cover-design` (provides the `cover.png` consumed in step 1) - `script-voice` and `pronunciation-drill` are upstream (script → audio); this skill is downstream (mp4 → publish) ## Constraints ### 1. One step per ffmpeg invocation Don't try to do cover + loudnorm + speed in a single complex filter chain. Three sequential files is way easier to debug and verify. Disk is cheap, your time isn't. ### 2. Preserve original resolution and frame rate Source is 1080×1920 @ 30fps. Output stays 1080×1920 @ 30fps. Never downscale "to save bandwidth" — platforms re-encode, you only lose quality. ### 3. Match source bitrate when re-encoding If source is ~28 Mbps, output target is 28 Mbps. Don't let `libx264 -crf 18` decide for you — it'll go 4-6 Mbps which is fine for re-upload but **not** archiva