What problem does it solve?
This Skill turns raw footage into a coherent edited video by forcing a deterministic, evidence-driven pipeline for cutting, assembling, and optional programmatic polishing instead of relying on subjective “looks right” edits.
Core Features & Use Cases
- Automated, EDL-driven cutting: Converts a transcript into an edit decision list (EDL) and executes FFmpeg cuts in EDL order to prevent concat/order mistakes.
- Deterministic assembly with hard gates: Captures, verifies intermediate artifacts (transcripts, EDL, segments, rough cut), and blocks progress when required outputs are missing.
- Optional Remotion composition layer: Wraps the assembled rough cut into a Remotion TypeScript/TSX workflow for timed overlays (titles, lower-thirds, captions) when needed.
- Optional gap-filling assets generation: Generates only missing assets (e.g., voiceover via ElevenLabs; b-roll/music via fal.ai) when the source coverage is insufficient.
Use Case Example: Create a 60–90 second product demo vlog by transcribing a screen recording, selecting narrative cut points, rendering FFmpeg segments, concatenating them into a rough cut, and then using Remotion to add timed titles and lower-thirds.
Quick Start
Use the video-editing skill to convert your screen recording into an edited vlog by generating cuts.txt from a transcript, producing assembled/rough-cut.mp4 with FFmpeg, and writing handoff-notes.txt for final human taste and export.
