marketing-council
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes information from external web searches and user-defined advisor files to simulate persona responses. This provides a potential surface for indirect prompt injection if those external sources contain malicious instructions.
- Ingestion points: Ingests data from .agents/product-marketing.md, external search results, and custom files in .agents/advisors/.
- Boundary markers: The skill employs a mandatory simulation disclaimer at the beginning of its output.
- Capability inventory: Capabilities include reading local markdown files and invoking external research tools like web search or video analysis.
- Sanitization: No explicit sanitization of external search results is described; the skill relies on grounding rules and primary source preference.
- [COMMAND_EXECUTION]: The skill instructs the agent to create and save new markdown files in the .agents/advisors/ directory to support custom advisor profiles. This involves direct interaction with the file system for configuration storage.
Audit Metadata