AgentSkillsCN

article-to-script

将书面文章转化为结构化的视频脚本，配备时间标注、旁白与场景分解。

SKILL.md
--- frontmatter
name: article-to-script
description: Convert written articles into structured video scripts with timing, narration, and scene breakdowns
tags: [video, content, script, documentation]
version: 3.0.0

article-to-script

Convert written articles into structured video scripts ready for video production. This skill analyzes articles, designs scene structures, generates narrations, and calculates precise timing for video creation.

⚠️ CRITICAL REQUIREMENT: English Language Only

ALL NARRATIONS MUST BE WRITTEN IN ENGLISH

This is a non-negotiable requirement. Regardless of the input article's language (Chinese, Japanese, Spanish, etc.), all output narrations in script.json MUST be in English.

Why English?

  • •✅ International audience reach
  • •✅ TTS compatibility across all platforms
  • •✅ Professional standard for global content
  • •✅ Easier translation to other languages if needed

If you generate narrations in any language other than English, the script is INVALID and must be regenerated.

When to Use

Use this skill when you need to:

  • •Convert blog posts or articles into video scripts
  • •Design structured narratives for social media videos (108-144 seconds)
  • •Plan scene-by-scene breakdowns with visual cues
  • •Generate narrations optimized for text-to-speech playback
  • •Calculate precise timing for video production workflows

Do NOT use for:

  • •Direct video rendering (use script-to-remotion for that)
  • •Transcript generation from existing videos
  • •Interactive video content

How to Use

Input Requirements

  • •Article file (markdown format)
    • •Clear structure with headings
    • •500-2000 words recommended
    • •Contains core message, key points, and call-to-action

Output

  • •script.json: Unified JSON file containing:
    • •Article analysis (core message, audience, key points)
    • •Complete scene structure (5-7 scenes)
    • •Narrations for each scene (MUST be in English)
    • •Precise timing calculations (seconds and frames)
    • •Visual cues and animation suggestions

CRITICAL: All narrations MUST be written in English, regardless of the input article language. This is required for TTS generation and international audience reach.

Workflow Overview

The skill follows a 5-stage workflow:

  1. •Article Analysis → Extract core message, audience, key points, hook, CTA
  2. •Scene Structure Design → Plan 5-7 scenes with types and flow
  3. •Script Generation → Write narrations (2-3 words/second baseline)
  4. •Narration Expansion → Expand to 1.25 words/second for 1.2x playback
  5. •Timing Calculations → Calculate durations, frames, and timestamps

Each stage is documented in detail in the rules/ directory.

Example Usage

bash
# 1. Read the article
Read /path/to/article.md

# 2. Follow the workflow (read rule files in order)
Read .claude/skills/article-to-script/rules/workflow-overview.md
Read .claude/skills/article-to-script/rules/article-analysis.md
Read .claude/skills/article-to-script/rules/scene-structure.md
Read .claude/skills/article-to-script/rules/script-generation.md
Read .claude/skills/article-to-script/rules/narration-expansion.md
Read .claude/skills/article-to-script/rules/timing-calculations.md

# 3. Generate the script
Write /path/to/output/script.json

# 4. Validate against schema
Read .claude/skills/article-to-script/schemas/script-schema.json
# Verify structure matches

# 5. Quality check
Read .claude/skills/article-to-script/rules/quality-checklist.md

Key Features

Intelligent Scene Types

The skill supports 6 scene types optimized for different content:

  • •Hook (5-8s): Grab attention with bold statement
  • •Text Display (15-20s): Present core message or key point
  • •Number Comparison (12-18s): Show before/after or statistics
  • •Three Columns (15-20s): Compare 3 items or show progression
  • •Case Study (20-30s): Tell a story or show transformation
  • •CTA (8-12s): Clear call-to-action with contact info

Timing Optimization

  • •Baseline: 2-3 words/second for natural TTS speech
  • •Expansion: 1.25 words/second for 1.2x playback speed
  • •Frame precision: Convert seconds to frames (30 fps default)
  • •Scene pacing: Automated duration distribution

Quality Assurance

Built-in validation checks:

  • •Total duration: 108-144 seconds
  • •Scene count: 5-7 scenes
  • •Narration pacing: Within optimal ranges
  • •Visual cue completeness
  • •JSON schema compliance

Rule Files

All workflow rules are documented in rules/:

  • •workflow-overview.md: Complete end-to-end process
  • •article-analysis.md: Extract core message and structure
  • •scene-structure.md: Design scene flow and types
  • •script-generation.md: Write effective narrations
  • •narration-expansion.md: Optimize for TTS playback
  • •timing-calculations.md: Duration and frame formulas
  • •quality-checklist.md: Validation criteria

Schema Reference

  • •schemas/script-schema.json: Complete JSON Schema for output validation

Examples

  • •examples/sample-article.md: Reference article input
  • •examples/sample-script.json: Complete script output with all fields

Next Steps

After generating script.json, use the script-to-remotion skill to:

  1. •Generate React components for each scene
  2. •Assemble a complete Remotion project
  3. •Optionally generate TTS audio
  4. •Render the final video

Version History

  • •v3.0.0 (2025-01): Pure documentation skill, unified schema
  • •v2.1.0 (2024-12): CLI-based Phase 1-3 implementation
  • •v1.0.0 (2024-11): Initial monolithic skill

Support

For issues or questions:

  • •Check rules/quality-checklist.md for common problems
  • •Review examples/ for reference implementations
  • •Consult script-to-remotion skill for downstream workflow