视频理解与剧本重建 Skill

To extract standardized metadata from genre assets, this Skill scans the repository for mandatory SKILL.md frontmatter and emits structured fields like name, tagline, complexity, and keywords for each valid Skill Unit. It ensures scalable discovery and reuse of production knowledge.

Cross-genre contamination occurs when assets improperly mix genres. This Skill detects it during cross-unit validation by analyzing the generated metadata model to flag inconsistencies, duplicates, and contamination, ensuring a clean and reusable asset library.

To validate Skill Units before onboarding new genres, this Skill checks for the mandatory SKILL.md frontmatter and performs toxicity checks. It verifies dependencies and components to ensure only valid, scalable artifacts are introduced into the repository.

Yes, you can use this Skill to catalog assets for model-driven prompt injection. It generates a structured metadata manifest for valid Skill Units, providing the standardized descriptions and keywords needed to guide prompt injection accurately.

The metadata fields generated for valid Skill Units include name, tagline, description, request, complexity, category, keywords, dependencies, components, and toxicity checks. These fields form a structured metadata model for each asset.