creative-brief

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill researches company information by fetching content from external websites and public sources, creating a surface for indirect prompt injection if the source content contains malicious instructions.
  • Ingestion points: External URLs and web pages fetched during the research phase specified in SKILL.md.
  • Boundary markers: The skill does not define clear delimiters or 'ignore' instructions for the content retrieved from external sources.
  • Capability inventory: The agent can fetch network resources and write data to .agents/product-context.md.
  • Sanitization: There are no instructions to sanitize or validate the content retrieved from the web before processing it.
  • [PROMPT_INJECTION]: The skill instructions include a directive to 'not return a draft with a warning on it' and instead stop and ask for missing inputs. While this is intended to prevent users from accidentally using incomplete or 'caveated' drafts, it technically implements a form of output concealment that overrides default agent reporting behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:22 PM