The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided copy and reads from a local context file (.claude/product-marketing-context.md). This represents an ingestion point for untrusted data. However, since the skill's capabilities are limited to text generation and it lacks network or shell access, the risk is negligible.
[DATA_EXPOSURE]: The skill checks for the existence of .claude/product-marketing-context.md to gather background information. This is a standard practice for project-specific customization and does not target sensitive system files or credentials.

copywriting