The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill fetches documentation indices and content from https://cloudinary.com/documentation/llms.txt. As this targets the official domain of a well-known service (Cloudinary) and matches the author's identified vendor infrastructure, it is documented as a safe and intended functional behavior.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external content from the Cloudinary documentation site. This represents a potential attack surface for indirect prompt injection. * Ingestion points: Documentation URLs listed in llms.txt on cloudinary.com (SKILL.md). * Boundary markers: None provided in the instructions for the fetched content. * Capability inventory: No high-risk capabilities detected; the skill is limited to reading and processing text for response generation. * Sanitization: No explicit sanitization or filtering of the remote content is performed. Given the high trust level of the source and the lack of dangerous capabilities, this is considered a safe surface.

cloudinary-docs