Video Caption Extraction and Script Replication

Problem

This skill addresses a specific content workflow: turning a YouTube video, or a transcript already copied by the user, into a reusable Chinese spoken script. It is not a line-by-line translation tool and it does not produce video files. The core deliverables are two: a Chinese spoken script and a replication structure.

How It Works

The skill separates two input paths: user-pasted transcripts that can be processed directly, and public caption queries that require a Beatra tool call. Queries are optional, paid, and must be confirmed before execution. The standard flow is:

• Collect the source text and mark each sentence as caption-derived or user-provided.
• Write the Chinese spoken script while preserving hooks, pacing, and speaker changes.
• Extract the retention structure and distinguish supported beats from inference.
• If a paid query was run, include task_id, final state, and billing.net_charged_credits.

Boundaries

It is suitable for replicating existing spoken content, not for inventing dialogue when no transcript is available. Captions exist only when the catalog returns them. If query access is unavailable and the user does not provide a transcript, the skill states the missing input instead of fabricating lines.