The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted content from Markdown files, web links, or user text and uses it to construct prompts for image generation tools.
Ingestion points: Processes content from local paths (e.g., 1-Wechat/ing/*.md) and external URLs provided by the user.
Boundary markers: The skill does not define clear delimiters or use instructions to ignore embedded commands when processing external data.
Capability inventory: Can execute image generation tools (imagegen), image viewers (view_image), and an image upload utility (1-upload-images-to-picgo).
Sanitization: No input validation or sanitization is performed on the ingested text before it is interpolated into the image generation templates.
[DATA_EXFILTRATION]: The skill includes a procedure to migrate external images to a hosting service using the 1-upload-images-to-picgo tool. This involves scanning user-provided Markdown files for external URLs and performing network-based upload operations.
[COMMAND_EXECUTION]: The skill provides instructions to execute specific local scripts and platform-integrated tools to view, generate, and upload images. This includes logic to save generated PNG files to local directories like assets/ or output/.

z-xkcd-panda-comic