The Agent Skills Directory

[SAFE]: A comprehensive analysis of all files, including the skill instructions and individual advisor dossiers, revealed no malicious patterns or security vulnerabilities.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads project-level context from files like .agents/product-marketing.md and writes user-defined advisor profiles to the .agents/advisors/ directory. These operations are limited to the local environment and are standard for maintaining project context and persistence.
[INDIRECT_PROMPT_INJECTION]: The skill features a 'Live Research Pass' that incorporates external data from web searches or research tools. While this creates a potential surface for indirect prompt injection, the skill includes robust mitigation measures, such as mandatory simulation labeling, strict grounding rules that prevent the fabrication of quotes, and a clear separation between strategic advice and execution tasks.

marketing-council