The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill defines a workflow that processes untrusted external content, specifically 'artifacts' to be styled and user-provided inputs for custom theme creation. This creates an attack surface where malicious instructions could be embedded in the processed data to attempt to override agent behavior. \n
Ingestion points: SKILL.md instructions for applying themes to 'any artifact' and using 'provided inputs' for custom generation. \n
Boundary markers: The skill lacks explicit instructions for the agent to use delimiters or ignore embedded natural language commands within the processed artifacts. \n
Capability inventory: The skill involves reading local files and generating or modifying styled output (internal influence/generation). \n
Sanitization: No validation or sanitization of input descriptions or artifact content is specified. \n- [Metadata Poisoning] (LOW): The skill metadata in SKILL.md references external files (LICENSE.txt and theme-showcase.pdf) that are not present in the skill package, which is a best-practice violation and prevents full verification of the skill's assets.

theme-factory