The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests data from external project files which could contain malicious instructions designed to manipulate the agent's behavior during persona simulation.\n- Ingestion points: The agent reads marketing context from .agents/product-marketing.md and custom advisor dossiers from the .agents/advisors/ directory.\n- Boundary markers: The instructions do not specify the use of clear delimiters or explicit instructions for the agent to ignore commands embedded within these external files.\n- Capability inventory: The skill has the capability to hand off tasks to other functional skills, such as deep-research, ads, and copywriting.\n- Sanitization: There are no sanitization or content validation steps mentioned for the ingested data.\n- [SAFE]: The skill implements robust 'Grounding Rules' that require simulations to be clearly labeled and prohibit the agent from fabricating quotes or endorsements, which effectively mitigates risks associated with persona misuse.\n- [SAFE]: The skill prevents skill-folder contamination by correctly instructing the agent to store user-defined custom advisor data in the project's .agents/ directory rather than within the skill's own internal directory structure.

marketing-council