Wan 3.0 Native Audio Cue Contract

Use this skill before an authorized Wan 3.0 generation call when dialogue, music, ambience, or sound effects must be intentional. It turns a redacted audio brief into a prompt-ready contract without reading media or calling a provider.

Input and output

Provide JSON with dialogue, music, ambience, sound_effects, silence_policy, duration, and human_confirmed. The executor emits JSON and Markdown with status, an ordered audio_prompt, exact dialogue lines, blockers, and api_call_authorized=false.

Rules

• - Every dialogue item needs a speaker, exact line, emotion, pace, timbre, and accent. Exact lines are quoted in the generated block.
• Every sound effect needs a source material, action, and acoustic space. Music needs a score description and genre.
• INLINECODE10 may explicitly disable dialogue, music, or both; it cannot silently remove a requested channel.
• Duration must be an integer from 2 through 30 seconds. A timed cue must fit inside that window.
• Human confirmation is required before any upload or paid call.

Boundaries

This is a static prompt contract. It does not create audio, inspect a video, call Wan, verify pronunciation, or prove synchronization. The operator must review the rendered audio and rights before delivery.