The Agent Skills Directory

[SAFE]: The skill consists of natural language instructions and templates for sales professionals. It does not execute code, perform network requests, or access sensitive system files.
[DATA_EXPOSURE]: The skill instructions include reading a local context file (.agents/sales-context.md). This is a standard practice for project-specific context and does not involve sensitive system or user data.
[INDIRECT_PROMPT_INJECTION]: There is a potential surface for indirect prompt injection via the .agents/sales-context.md file if it were to contain malicious instructions. However, the skill lacks high-risk capabilities like command execution or external data exfiltration, making the risk negligible. • Ingestion points: .agents/sales-context.md (read operation) • Boundary markers: Not explicitly defined in the prompt template • Capability inventory: Text generation for sales scripts; no subprocess or network capabilities found • Sanitization: Not explicitly implemented

cold-call