The Agent Skills Directory

[SAFE]: No malicious patterns or high-risk command executions were detected. The included Python script (roas_calculator.py) performs local mathematical calculations using standard libraries and does not engage in network activity or unauthorized file access.
[PROMPT_INJECTION]: The skill features a surface for indirect prompt injection by design, as it is instructed to read .claude/product-marketing-context.md to personalize its output. While this is functional for marketing tasks, instructions within that context file could influence agent behavior.
Ingestion points: .claude/product-marketing-context.md and campaign data files via roas_calculator.py.
Boundary markers: Absent; the instructions do not specify the use of delimiters when reading external marketing context.
Capability inventory: Campaign architecture generation, ad copy creation, and budget optimization recommendations.
Sanitization: The skill does not explicitly validate the content of the marketing context file for hidden instructions.

paid-ads