extract-creator-voicelisted

# Extract Creator Voice ## Purpose Encode the creator's brand voice architecture, on-camera persona, parasocial-trust contract, and unique positioning into Compartment 1 (Creator Identity Matrix). This compartment is what every script-writing skill, every hook-writing skill, every title-strategist invocation, and every sales-copy skill reads to produce voice-faithful output. Voice fidelity is the difference between "AI-generated content" and "creator-written content" in the Blind Output Test. ## Decision Logic Voice is not "tone." Voice is the encoded combination of (a) communication style, (b) language patterns / signature cadence, (c) phrases-to-use / phrases-to-avoid, (d) persuasion style, (e) authority positioning, (f) on-camera persona, (g) parasocial-trust contract. The compartment captures all seven layers. The output is a usable spec — readable by every skill, machine-checkable against output drafts. Voice-fidelity verification (Layer 2 of Triple-Layer Verification) reads from this compartment. Without it, every output is judged against a generic LLM voice. The parasocial-trust contract is INV-16's source-of-truth. The creator names what they promise the audience long-term + what they explicitly will not do. The OS refuses to generate content that violates this contract. ## Tacit Principles 1. **Voice is captured from existing artifacts, not described in the abstract.** Read the creator's actual writing / videos / DMs / sales calls. Pull verbatim. Don't ask "desc