script-check-workflow: Industrial Inspection and Standardization Agent Skill for AI Video Scripts

script-check-workflow is an Agent Skill for AI-executable scripts. Through a 7-stage linear inspection process, it converts creative scripts into standardized execution documents suitable for directors, storyboarding, AI generation, and industrial acceptance. Its core value lies in the "executability" orientation, filling the key gap between creativity and execution. It does not judge the quality of the story, but focuses on outputting standardized documents that guide production.

AI video generation technology is developing rapidly, but traditional literary scripts are not suitable for direct use as AI input—AI requires structured, standardized "execution scripts" rather than literary creative texts. This project is designed to address this pain point, serving as an industrial script converter to connect the creative and execution phases.

It adopts a 7-stage linear inspection process: Principle Layer (basic principles for AI executability), Scene Layer (scene completeness), Shot Layer (technical specifications), Action Layer (executability), AI Generation Adaptation Layer (model-specific requirements), Dialogue Layer (separated from performance), and Industrial Acceptance Layer (delivery standards). Key principles include inter-layer isolation (each stage runs independently) and the "Three-Zero Principle" for the visual layer (0 psychological words, 0 subjective intentions, 0 pronouns).

By default, it outputs two documents: 1. standardized-script (a standard script without annotations, easy for downstream reference); 2. diagnostics-record (a high-granularity diagnostic report including issues at each stage, modification strategies, etc.). It supports a comprehensive final review with 12 questions, as well as flexible output control (full/designated stage/single shot inspection, re-inspection, explanation mode).

It is a cross-platform Agent Skill that supports mainstream AI CLI tools such as Claude Code, Codex, Gemini CLI, OpenCode, and OpenClaw, and provides corresponding installation commands. The trigger method is a unified command, and the inspection scope can be specified (e.g., only Stage5 AI adaptation, re-inspection of specific shots, etc.).

Applicable scenarios: AI video scripts (Sora/Runway/Pika, etc.), animation execution drafts, storyboard scripts, Shot list-style documents, re-inspection drafts. Inapplicable scenarios: literary script review (does not judge story quality), plot evaluation, director style discussion, general creative divergence (does not produce executable documents).

As an important part of the AI video creation toolchain, it provides infrastructure support for industrialization and establishes production standards for the AI era. Future directions: multi-language support, model adaptation optimization, collaboration functions, and quality tracking.

script-check-workflow is a practical-focused Agent Skill that helps ideas be better executed. With the development of AI video technology, such tools will promote the efficient development of industry standards and are worth the attention and trial of AI video creation and animation production teams.