The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read context from local files (e.g., .agents/product-marketing.md) and incorporate 'voice-of-customer' data from external sources like reviews and interviews. This represents a theoretical surface for indirect prompt injection, as the agent may process untrusted text embedded in those files. However, the skill does not possess capabilities like network exfiltration or shell execution that would allow for exploitation.
Ingestion points: Processes data from .agents/product-marketing.md, .claude/product-marketing.md, and product-marketing-context.md.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the ingested context files.
Capability inventory: No subprocess calls, network operations, or file-writing tools are utilized by this skill.
Sanitization: No sanitization or validation of the ingested marketing context is performed.
[SAFE]: The skill consists entirely of instructional text and reference documentation. It does not include scripts, executables, or external dependencies, and its behavior is consistent with its stated purpose as a copywriting assistant.

copywriting