card-news-generator
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input to generate text content and images, presenting a surface for indirect prompt injection.
- Ingestion points: The skill takes a user-provided 'Topic' through instructions in
SKILL.mdto generate content for multiple cards. - Boundary markers: No explicit delimiters or 'ignore embedded instructions' markers are used to separate user-provided topics from the agent's content generation instructions.
- Capability inventory: The skill possesses file-writing capabilities through the
PILlibrary'simg.savemethod ingenerate_card.pyand directory creation viaos.makedirsinauto_generator.py. - Sanitization: No specific sanitization or filtering of the user-provided text was found before it is passed to the generation scripts or rendered into images.
- [EXTERNAL_DOWNLOADS]: The documentation suggests installing the
PillowPython library and standard system fonts (fonts-noto-cjk).Pillowis a widely-used and trusted image processing library. - [COMMAND_EXECUTION]: The skill executes local Python scripts (
auto_generator.py,generate_card.py,interactive_generator.py) to handle the logic of parsing content, resizing images, and rendering text onto canvases.
Audit Metadata