Short Clip Candidate Scorer

Scenario

Use after transcript validation to rank short windows before spending render time or publishing effort.

Input and output

Input contains normalized segments and buyer thresholds for minimum and maximum duration, hook terms, and context markers. Output contains candidate windows, scores, rejection reasons, and candidate_hash.

Execution and recovery

Run python scripts/score.py examples/input.json. Invalid thresholds return visible errors. Adjust the buyer policy and rerun; the script does not claim virality or predict performance.

Privacy boundary and acceptance

Use redacted text and metadata only. The sample must return at least one candidate and explain rejected windows when applicable.