Shuohao-skills is a collection of AI agent skills for short-drama production that runs in Claude Code and Codex environments.
The tool addresses the workflow of adapting novels into production-ready materials for AI video generation. It breaks down the adaptation process into five sequential skills: extracting character bibles from outlines, creating adaptation outlines that converge narrative structure, generating art direction for scenes and props, writing screenplays with synchronized dialogue and action beats, and producing storyboards with frame-by-frame timing. Each skill consumes structured JSON output from previous stages, enabling iterative refinement where the outline anchors decisions, characters and scripts develop in parallel, and storyboards serve as output-only without introducing new creative choices. The pipeline includes quality gates at each stage—ranging from 10 to 17 validation checks per skill—and supports bilingual output in Chinese and English.
Teams building AI short-form video should adopt this tool if they need a structured, end-to-end production pipeline that maintains consistency across character design, visual direction, and screenplay timing. The skills are designed to feed directly into AI generation pipelines, with specific affordances like character voice prompts for text-to-speech integration, art direction anchors for visual consistency, and storyboard frame timing calculated deterministically from speech rate. The tool suits projects where narrative adaptation and asset generation must stay synchronized. The skills function independently—any single skill can be used in isolation or copied elsewhere—and a separate HTML assembler can combine outputs from multiple skills into a single-page dashboard without modifying the underlying skill code.
Development activity shows consistent refinement of the production pipeline. The project maintains comprehensive validation logic embedded throughout each skill rather than as a separate layer. Documentation includes detailed specifications for each skill's input and output formats, enabling integration with external systems. The tool demonstrates attention to practical production concerns like frame duration constraints, reference image handling for visual generation, and dialogue aggregation by character with associated voice parameters.
