creative-director

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFENO_CODEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The campaign library contains frontmatter with source_url fields pointing to established external platforms such as YouTube, Vimeo, D&AD, and Business Insider. These URLs are provided for human reference and do not represent automated download or execution patterns by the agent.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied briefs and a large local library of campaign cards. While this ingestion of external and local data creates a surface for potential indirect instructions, the skill does not possess high-risk capabilities such as file system writing or network operations that could be exploited. No boundary markers or sanitization procedures for user input were observed in the instructions.
  • [NO_CODE]: The skill files consist of static markdown documentation and YAML metadata. Although the documentation mentions Python scripts used for library maintenance, these scripts are not included in the agent's operational logic and are described as maintainer tools to be run manually.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:31 PM