huashu-douyin-script_skill

Overview

This skill automates a complete Douyin/TikTok-style short video script workflow from competitor teardown to final script and AI-aware proofreading. It turns guesswork into a repeatable process: download target videos, run Gemini-based video analysis, distill proven hit formulas, generate script + shot-list, and apply compliance and anti-AI-flavor checks. Designed for both organic "seed" videos and paid 千川 ad assets.

How this skill works

I collect product details and 1–5 competitor Douyin links, download the videos, and run Gemini video analysis to extract hooks, shot structure, and rhythm. I aggregate findings into a category-level hit formula, generate a time-marked script and storyboard tailored to either seed or 千川 formats, then run five focused AI-proof and compliance checks and deliver the revised script and a change log.

When to use it

• You want a high-conversion Douyin seed (种草) video script
• Preparing 15s千川 ad creative with strict information density
• You have competitor videos and need structural teardown and copyable formulas
• You need compliance-safe wording for ad platforms
• You want a fast end-to-end pipeline from video download to audited script

Best practices

• Provide 1–5 representative competitor links plus clear product core selling points
• Specify target runtime and campaign goal (e.g., click, live-room traffic, purchase) up front
• Use Chrome logged into Douyin for reliable downloads; fallback to manual upload if needed
• Confirm the distilled formula before script generation to ensure brand fit
• Review the AI-proofing change list and approve edits before final export

Example use cases

• Create a 30s seed-style script to drive organic product recommendations
• Generate a 15s 千川 ad optimized for immediate CTA and platform review
• Break down top-performing competitor videos to extract repeatable hook templates
• Turn product brief and images into a shot-by-shot storyboard for shoot day
• Audit existing script for compliance, colloquial tone, and removal of AI-sounding phrasing