The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text and overlays it onto images, creating a surface for indirect prompt injection aimed at users or subsequent processing steps.
Ingestion points: The --text command-line argument in scripts/image_processor.py accepts arbitrary strings.
Boundary markers: The skill does not use delimiters or instructions to isolate the caption text from other instructions.
Capability inventory: The skill performs file reading and writing using the Pillow library and can be executed via the Bash tool.
Sanitization: There is no sanitization, validation, or escaping of the input text before it is processed and rendered.

image-processing