Kling 3.0声元素契约

Use this skill before a Kling VIDEO 3.0 Omni reference workflow. It validates a redacted element contract so visual identity, voice binding, reference video, and shot roles are explicit before an authorized operator uploads anything.

Prerequisites

• - Python 3.9 or newer.
• A UTF-8 JSON file with elements, shots, and rights_review fields.
• The buyer\'s own reference images, videos, and voice decisions. This skill checks metadata only and never reads media.

Inputs and outputs

Run python scripts/check_element.py examples/pass_contract.json --out contract.json --report contract.md.

Each element declares a unique ID, an element type, a visual reference ID, a voice reference ID, a reference duration, and a rights status. Shots declare which element is bound and what the voice is allowed to do. The checker enforces non-empty IDs, supported types, 3-to-8 second video reference metadata, explicit voice binding, known shot elements, and confirmed rights. It writes JSON and Markdown reports with upload_authorized: false.

Execution and acceptance

1. 1. Replace redacted IDs and factual fields only.
2. Run the checker. Exit code 0 means the metadata is internally consistent; exit code 1 means hold the request.
3. Confirm the current Kling product mode, voice-binding behavior, region access, reference permissions, and cost manually before any upload or generation.

Passing the checker does not prove that a person, voice, video, or appearance will remain consistent. It also does not prove access or quality.

Privacy and recovery

Keep faces, voices, names, private URLs, tokens, and media bytes out of examples. Get explicit permission for likeness and voice use. If an element lacks a rights decision, voice binding, or valid reference duration, correct it instead of retrying a paid request.

Standalone use

The contract can be used independently for a character-asset handoff or shot plan; the companion auditor is optional.