The skill is vulnerable to path traversal, command
Claims to do
Xiaohongshu Image Card Series Generator: Break down complex content into eye-catching Xiaohongshu image card series with multiple style options.
Actually does
This skill orchestrates the creation of Xiaohongshu image card series. It analyzes user-provided content, loads user preferences from `EXTEND.md`, generates content outlines and image prompts based on selected styles and layouts, and manages the workflow for an external image generation skill, including ensuring visual consistency via reference images and applying watermarks. It creates a structured directory for all generated artifacts.
The skill constructs file paths and names using dynamically generated slugs from user content. Insufficient sanitization of these inputs could lead to command injection or path traversal vulnerabilities when performing file system operations or executing shell commands.
Slug Generation: Extract main topic from content (2-4 words, kebab-case)
xhs-images/{topic-slug}/
test -f .baoyu-skills/baoyu-xhs-images/EXTEND.md
Save source content
Save prompt to prompts/NN-{type}-[slug].mdThe skill constructs prompts for an external image generation skill, incorporating dynamic and potentially user-controlled elements like watermarks. This creates a vector for an attacker to inject malicious instructions into the downstream image generation process.
Add to each image generation prompt: Include a subtle watermark "[content]" positioned at [position]. Image Generation Skill Selection
The skill uses `EXTEND.md` to store user preferences and custom configurations, checking for its existence in various user and project directories. If an attacker can write to this file, they could persistently alter the skill's behavior or inject malicious configurations that are loaded on subsequent runs.
EXTEND.md Supports: Watermark | Preferred style/layout | Custom style definitions | Language preference Save EXTEND.md test -f .baoyu-skills/baoyu-xhs-images/EXTEND.md
The skill performs extensive file system operations, including creating directories, saving source content, analysis, outlines, prompts, and generated images. While necessary for its function, this broad access, especially with dynamic naming, increases the attack surface for unauthorized file manipulation or data exfiltration if combined with other vulnerabilities.
xhs-images/{topic-slug}/
source-{slug}.{ext}
analysis.md
outline.md
prompts/
Remove filesThe `--yes` flag allows the skill to bypass all user confirmations and proceed with automated actions based on defaults or auto-recommendations. This removes a critical human-in-the-loop security control, increasing the risk of unintended or malicious actions if the skill is invoked with compromised inputs.
`--yes` | Non-interactive mode: skip all confirmations. `--yes` mode: Skip this entire step. No AskUserQuestion calls.
The skill's description implies it directly generates images, but the detailed workflow indicates it acts as an orchestrator, preparing prompts and managing the process for an *external* image generation skill.
The 'Image Generation Skill Selection' section states: 'Check available image generation skills', implying it relies on another skill for the actual image generation.
[](https://mondoo.com/ai-agent-security/skills/github/jimliu/baoyu-skills/baoyu-xhs-images)<a href="https://mondoo.com/ai-agent-security/skills/github/jimliu/baoyu-skills/baoyu-xhs-images"><img src="https://mondoo.com/ai-agent-security/api/badge/github/jimliu/baoyu-skills/baoyu-xhs-images.svg" alt="Mondoo Skill Check" /></a>https://mondoo.com/ai-agent-security/api/badge/github/jimliu/baoyu-skills/baoyu-xhs-images.svgSkills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.