The Agent Skills Directory

[SAFE]: The skill's primary function is image processing using the Python Imaging Library (Pillow). It uses local scripts (generate_card.py, auto_generator.py, interactive_generator.py) to handle text wrapping, image resizing, and font rendering for Korean characters.
[EXTERNAL_DOWNLOADS]: The skill recommends installing the Pillow package from PyPI, which is a standard, well-known library for image processing. It also suggests system-level font installations (e.g., fonts-noto-cjk) on Linux, which are official distribution packages.
[COMMAND_EXECUTION]: The skill uses argparse to handle CLI inputs and sys.stdin.read() to ingest content generated by the AI agent. There are no patterns of executing arbitrary system commands or spawning shells with unvalidated user input.
[DATA_EXFILTRATION]: All file operations are restricted to local paths provided by the user or the standard /mnt/user-data/outputs/ directory. No network operations or credential-accessing patterns were detected.
[PROMPT_INJECTION]: The SKILL.md instructions focus on workflow management and content formatting (character limits for titles and body). It does not contain instructions that attempt to bypass safety filters or override agent constraints.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user-provided topics and generated content to create images, it lacks high-privilege capabilities that could be exploited via indirect injection. The risk surface is minimal as the output is a static image file.

card-news-generator