The Agent Skills Directory

[SAFE]: The skill instructions and its reference files (copy-frameworks.md and natural-transitions.md) focus exclusively on copywriting best practices and templates. No malicious instructions, unauthorized file access, or suspicious network operations were detected.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external context and user input, which creates a potential surface for indirect prompt injection, although the risk is low due to the limited capability scope of the skill.\n
Ingestion points: Accesses context from .claude/product-marketing-context.md and ingests user-supplied details regarding audience and product offers.\n
Boundary markers: None identified to separate ingested context from agent instructions.\n
Capability inventory: The skill is limited to text generation and interactive drafting; it lacks capabilities for file-system writing, network exfiltration, or command execution.\n
Sanitization: No specific sanitization or filtering logic is present for the context data ingested at runtime.

copywriting