The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting external data to inform its simulated marketing advice. • Ingestion points: The skill reads local project files (e.g., .agents/product-marketing.md) and processes external data from web searches or video analysis tools. • Capability inventory: The skill can read and write files within the project directory and invoke research tools. • Sanitization: There are no explicit mechanisms defined to sanitize the content of the marketing context files or external research data. • Boundary markers: The skill instructions use specific output formatting but do not include strong delimiters or 'ignore' instructions for the external content it processes.
[COMMAND_EXECUTION]: The skill is configured to interact with research tools (such as deep-research and watch-video) to gather information about advisor viewpoints and recent industry changes. These commands are triggered by the 'Live Research Pass' feature to ensure advice remains current and grounded in published sources.

marketing-council