The Agent Skills Directory

[SAFE]: The skill instructions and accompanying documentation consist entirely of plain-text marketing and conversion rate optimization (CRO) expertise. No executable code or shell commands are present.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill refers to reading a local context file, specifically .agents/product-marketing.md. This is a legitimate pattern for the agent to acquire project-specific knowledge. No access to sensitive system directories (e.g., SSH keys, AWS credentials) or environment variables is requested.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided product marketing data. While it lacks explicit boundary markers for this untrusted input, the risk is negligible because the skill does not possess capabilities for file modification, network communication, or command execution.
Ingestion points: .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md.
Boundary markers: None identified.
Capability inventory: Text generation only. No network tools, file writers, or subprocess access enabled in the configuration.
Sanitization: None identified.

onboarding