The Agent Skills Directory

[SAFE]: The skill functions as a set of instructions and a reference library for behavioral science frameworks. It does not include scripts, binaries, or commands that could compromise the host environment.
[INDIRECT_PROMPT_INJECTION]: The skill references an external context file .agents/marketing-context.md and processes user-provided behavior descriptions. While this establishes a surface for indirect prompt injection, the skill lacks any capabilities (such as network access or file system writes) that could be exploited via this vector, rendering the risk negligible.
[DATA_EXPOSURE]: The skill accesses a shared project file for marketing context, which is a standard pattern for agentic workflows and does not involve the exposure of credentials or system-level sensitive data.

behavior-design