drleadflow/short-form-content-factory ? reverse-engineered prompt

Build me a command line tool that takes raw talking head clips and turns them into finished vertical shorts with captions, titles, and clean audio.

I want to drop a bunch of clips into a folder, run a simple set of scripts, and get polished exports at the end. It should handle duplicate detection, audio prep, transcription with word level timestamps, speech cleanup with either Adobe Podcast Enhance or Auphonic, blending the cleaned audio back in, animated captions, bold hooked titles, compositing, and a final QA sheet so I can quickly spot anything weird before posting.

Please make each stage runnable on its own, since I may want to skip steps for some batches. Use environment variables for any keys, and include clear setup instructions plus a sample config for titles. If you need to look up current docs online for any service integration, do that. The main thing is that this feels reliable for real batches, not like a toy demo.