Video Director

Produce multi-scene AI videos with automated QA. This skill handles the production pipeline — generation, speech verification, visual QA, assembly, and voice swap. It works with any format by converting the format's recording brief or prompt template into a vidlang spec.

The process is always the same:

1. Pick a format (talking-head, ugc-b2c, ugc-yapper, etc.)
2. Write a vidlang spec using that format's structure
3. Run the production pipeline below

Step 0: Choose a Format

The format determines the scene structure, setting, and tone. The production pipeline stays the same regardless of format.