The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external untrusted data through figure descriptions and image files during its quality-control audit phase.
Ingestion points: The skill accepts user-provided context regarding research areas and methodology, and it explicitly instructs the agent to use a Read tool to load image files from a user-supplied path for vision-based inspection in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between the skill's core instructions and potentially malicious instructions embedded within the processed image metadata or the input text.
Capability inventory: The skill utilizes the Read tool for local filesystem access and provides recommendations for generating Python scripts and using external design tools.
Sanitization: The skill does not perform or specify any sanitization, validation, or filtering of the user-provided text or image content before it is processed by the agent's vision and reasoning capabilities.

figure-designer