The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests article content or pasted text to determine visual themes and metaphors for image generation prompts. This represents a standard indirect prompt injection surface where malicious content in the article could potentially influence the resulting image prompt.
Ingestion points: Processes local files (e.g., article.md) or direct user text input to extract topic and visual metaphors.
Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the input content in the analysis step.
Capability inventory: The skill performs file system writes (saving prompts to prompts/cover.md and images to cover.png) and invokes external image generation skills.
Sanitization: The instructions do not specify any sanitization or validation of the input text before extraction and interpolation into the generation prompt.

cover-image