freestylefly/xiaohongshu-skills
Overview
This skill generates Xiaohongshu-style (Xiaohongshu = Little Red Book) vertical cover images based on user topics. It produces clean, youthful 3:4 mobile-friendly covers saved to the current working directory. Use it to quickly create platform-ready cover art with readable text and no watermarks.
How this skill works
You provide a topic or short brief and the skill calls the image generation backend using an API key (CANGHE_API_KEY or provided at runtime). The generator creates a high-quality 3:4 image, applies Xiaohongshu-style layout and typography, removes watermarks/logos automatically, and saves the file as xiaohongshu-cover-{timestamp}.png in the directory where the skill is invoked. If the topic is unclear, the skill prompts for clarification before generating.
When to use it
- Creating a mobile-optimized Xiaohongshu cover for a post or note.
- You have a topic, hook line, or keywords like "Xiaohongshu cover", "小红书封面", or "封面生成".
- Rapid prototyping of social media thumbnails in the Xiaohongshu aesthetic.
- When you need a clean, readably-typed cover without watermarks for publishing.
Best practices
- Provide a concise topic plus 1–2 style hints (mood, dominant color, or short headline).
- Ensure CANGHE_API_KEY is set in the environment or pass a valid API key when running.
- Use short, clear text for cover headlines so they remain readable at mobile scale.
- Run generation from the target project directory so the output file is saved where you want it.
- Confirm the topic if the intent is ambiguous to avoid multiple iterations.
Example use cases
- Generate a cover for a beauty tutorial titled "5-minute morning makeup" with a pastel aesthetic.
- Create a travel note cover for "Best cafes in Taipei" using warm, cozy tones.
- Produce an e-commerce product highlight cover focusing on a bold single-color background.
- Make a lifestyle post cover that emphasizes a clean, minimalist layout for text overlays.
FAQ
A valid CANGHE_API_KEY from https://api.canghe.ai/ is required; set it in the environment or supply it when running the generator.
Where is the image saved?
The generated file is saved to the current working directory with the name xiaohongshu-cover-{timestamp}.png (timestamp in milliseconds).