The Agent Skills Directory

[SAFE]: No security issues were detected. The skill follows best practices by instructing the agent to respect user authorization, protect personal data in analytics, and verify all recommendations and commands before implementation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external marketing pages and form data, which is an inherent ingestion surface for untrusted content. 1. Ingestion points: External URLs and local context files like product-marketing.md are used for analysis. 2. Boundary markers: The skill contains explicit 'Limitations' and 'Evidence' sections that warn to verify all commands and generated code, and to maintain legal/privacy boundaries. 3. Capability inventory: The skill focus is on providing copy alternatives and strategic recommendations; it does not request specific tools for network access or file writes in its frontmatter. 4. Sanitization: The instructions emphasize that session recordings must be masked and personal data should not be exposed.

cro