The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local project files such as .agents/product-marketing.md and user-defined advisor dossiers in .agents/advisors/. This creates a potential surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The agent is instructed to read .agents/product-marketing.md, .claude/product-marketing.md, and dossiers from references/advisors/ or .agents/advisors/ at runtime.
Boundary markers: A simulation disclaimer is mandated at the top of the output to clarify the nature of the responses and establish a boundary between the system instructions and simulated content.
Capability inventory: The skill uses web search and optionally deep research tools to gather information based on the user's specific query.
Sanitization: The instructions include strict grounding rules against fabricating quotes and invented endorsements to maintain persona integrity and limit the influence of potentially malicious inputs.

marketing-council