Blog Brand

It generates two files at your project root: BRAND.md captures audience, positioning, editorial rules, competitor differentiators, and topic boundaries. VOICE.md mirrors your persona JSON as readable prose so every downstream skill loads the same context. Run the init interview once, and every blog command after that auto-loads both files. It's the editorial equivalent of a persistent PRODUCT.md. Pairs with blog-persona for the structured JSON, but this is the human-readable layer that actually travels with your prompts.