XHS Cover Generator

Generate Xiaohongshu (小红书) cover images with Chinese text overlays. Use when asked to create social media cover images, Xiaohongshu post images, or RED post...

Security Scan

The skill's code and instructions are consistent with its stated purpose (generating Xiaohongshu cover images); it downloads a font and optionally calls Pollinations.ai but does not request credentials or perform unexpected file or network access.

Purpose & Capability

The skill's name/description match the included code and SKILL.md: it generates cover images, renders Chinese text with a Noto Sans CJK font, and can fetch base images from Pollinations.ai. One minor inconsistency: the skill metadata lists no required binaries, but SKILL.md and the script rely on curl and Python with Pillow (PIL). This is an omission in metadata rather than a functional mismatch.

Instruction Scope

Runtime instructions and the script stay within the stated purpose: creating images, downloading a font into /tmp, and optionally fetching an AI-generated base image from Pollinations.ai. The instructions do not ask the agent to read unrelated files, environment variables, or credentials.

Install Mechanism

There is no install spec (low-risk). The script performs runtime downloads: a font from a GitHub raw URL and images from image.pollinations.ai. Both are public endpoints; the font download is from GoogleFonts on GitHub (raw), which is an expected source. Downloads occur at runtime and are written to /tmp.

Credentials

The skill requests no environment variables or credentials. It does require network access to download the font and optionally to call Pollinations.ai, which is reasonable for the described functionality. The script does not access or exfiltrate secrets.

Persistence & Privilege

The skill does not request permanent/system-wide installation or 'always' privilege, and it does not modify other skills. It caches the font in the system temp directory (/tmp), which is normal for this use case.

Assessment

This skill appears to do what it says: generate Xiaohongshu cover images. Before installing or running it, consider: (1) ensure the agent environment has Python3 and Pillow and that curl is available (SKILL.md requires these, but the registry metadata omitted them); (2) the script downloads a Noto Sans CJK font from a GitHub raw URL and optionally sends your text prompt to Pollinations.ai — if your prompts contain sensitive information, they will be transmitted to an external service; (3) the script does not verify the font SHA256 (FONT_SHA256 is None), so if you are security-sensitive you may want to supply a local font file instead or add a hash check; (4) the script caches files in /tmp and writes the resultant JPEG to the requested output path — verify file paths before running. If any of these network calls or automatic downloads are unacceptable, review/modify the script to use a preinstalled font or a different image source.

Generate publication-ready Xiaohongshu (小红书) cover images (1080×1440, 3:4 ratio) with Chinese text overlays.

How It Works

1. Base image: Either AI-generated via Pollinations.ai (--base-prompt) or solid color background
2. Text overlay: Chinese text rendered with Noto Sans CJK font via PIL (auto-downloaded on first run)
3. Output: JPEG at 95% quality, ready to post

Styles

Style	Use Case	Example
split	A vs B comparisons	PUA式AI vs 暖心AI
gradient	Single-topic posts	AI工具测评
card	Clean/minimal posts	效率技巧分享

Usage

Run the generate script:

python3 scripts/generate.py \
  --title "主标题文字" \
  --subtitle "副标题" \
  --style split \
  --left-label "左侧标签" \
  --right-label "右侧标签" \
  --base-prompt "english prompt for AI base image" \
  --output cover.jpg

Parameters

• --title (required): Main title text (Chinese OK)
• --subtitle: Secondary text line
• --tagline: Bottom tagline (split style only)
• --style: split | gradient | card (default: gradient)
• --left-label / --right-label: Labels for split style
• --left-color / --right-color: R,G,B colors (default: red/green)
• --base-prompt: English prompt for Pollinations.ai base image; if omitted, uses solid color
• --width / --height: Dimensions (default: 1080×1440)
• --output: Output path (default: cover.jpg)

Examples

Split comparison cover:

python3 scripts/generate.py \
  --title "同一个问题，两种AI的回答" \
  --subtitle "差距有多大?" \
  --tagline "一招设置反PUA人设" \
  --style split \
  --left-label "PUA式AI" --right-label "暖心AI" \
  --base-prompt "cute flat illustration, split screen, left pink sad robot, right green happy robot, kawaii pastel" \
  --output pua-vs-warm.jpg

Gradient tool review:

python3 scripts/generate.py \
  --title "2026最强AI编程助手" \
  --subtitle "Cursor vs Copilot vs Windsurf" \
  --style gradient \
  --base-prompt "futuristic code editor interface, purple blue gradient, clean minimal" \
  --output coding-tools.jpg

Card style tip:

python3 scripts/generate.py \
  --title "用AI写周报只需30秒" \
  --subtitle "打工人效率神器" \
  --style card \
  --left-color "70,130,255" \
  --output weekly-report.jpg

Requirements

• Python 3 with Pillow (PIL)
• curl for font download and Pollinations.ai
• Internet access (first run downloads ~17MB CJK font, cached in /tmp)

Typical Workflow (for agents)

1. Determine the post topic and style
2. Craft an English prompt for the base image (descriptive, aesthetic keywords)
3. Run generate.py with Chinese title/subtitle
4. Send the resulting image to the user