The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied marketing copy for editing, which represents a natural ingestion point for external data. However, as the skill lacks high-privilege capabilities such as network access, file system modifications, or shell command execution, this surface does not pose a significant risk. Mandatory Evidence: 1) Ingestion points: User-provided marketing text and .agents/product-marketing-context.md. 2) Boundary markers: None. 3) Capability inventory: None. 4) Sanitization: None.- [DATA_EXFILTRATION]: The skill attempts to read from .agents/product-marketing-context.md or .claude/product-marketing-context.md to align with brand voice. These are standard local project files for agent context and do not contain system-level credentials or sensitive private keys. No network operations are present to facilitate data exfiltration.- [SAFE]: All instructional content and frameworks, such as the Seven Sweeps Framework, represent standard marketing editing practices. No patterns of obfuscation, privilege escalation, or persistence were detected.

copy-editing