The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and review external marketing materials provided via the <material-path> argument. These materials represent untrusted data that could contain hidden instructions intended to bias the compliance review or manipulate the output artifacts.
Ingestion points: Materials located at the user-provided <material-path> (e.g., ad copy, contest rules, privacy policies).
Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when the agents process the material content.
Capability inventory: The orchestration involves multiple agents reading files and writing detailed reports, remediation guides, and checklists to the .aiwg/marketing/compliance/ directory.
Sanitization: There is no mention of sanitizing, escaping, or validating the content of the marketing materials before they are analyzed by the LLM agents.
[PROMPT_INJECTION]: The --guidance parameter explicitly allows users to provide natural language instructions that are parsed to 'tailor priorities', 'adjust agent emphasis', and 'influence scope'. While this is a functional feature, it creates a direct interface for prompt injection where a user could attempt to override the skill's compliance frameworks or safety guidelines.

legal-compliance