cmo-advisor
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions mandate reading an external, potentially untrusted file (
company-context.md) to inform its strategic recommendations, creating a vector for indirect prompt injection. \n - Ingestion points: The
SKILL.mdfile explicitly instructs the agent to "Always readcompany-context.mdbefore responding" to integrate company-specific information. \n - Boundary markers: The skill lacks instructions to treat the content of
company-context.mdas data rather than instructions, and does not provide delimiters to separate this external content from the system prompt. \n - Capability inventory: The skill includes and executes local Python scripts (
marketing_budget_modeler.py,growth_model_simulator.py) which represent a local code execution surface. \n - Sanitization: There is no mechanism described for sanitizing, validating, or filtering the content of the external context file before it is processed by the agent.
Audit Metadata