liyue-aigc/xianxia-visual-director

This repository is an Agent Skill for turning a scene idea, existing prompt, or visual reference into a structured cinematic xianxia image-prompt package. The documented workflow locks parameters, selects scene and style routes, applies shared visual rules, and produces a director’s plan, prompt, negative constraints, and variations. The README documents the capabilities and route relationships; no source excerpts were provided, so internal call-level relationships cannot be verified.