The Agent Skills Directory

[External Data Processing]: The skill orchestrates a research subagent and subsequently reads findings from local files, creating a potential surface for indirect prompt injection from external research sources. * Ingestion points: Research findings are read from the research/ directory (SKILL.md). * Boundary markers: No explicit delimiters are defined to isolate research data from the agent's instructions. * Capability inventory: The skill performs file writing and uses an image generation tool. * Sanitization: There are no explicit instructions for the agent to sanitize the research data prior to content generation.

social-media