The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or remote code execution vectors were found. The skill is entirely instructional, focusing on marketing analysis and providing text-based recommendations.
[DATA_EXPOSURE]: The skill instructions read a local file (product-marketing-context.md) to gain business context for its analysis. This is a standard practice for context-aware agents and does not involve accessing sensitive system credentials, SSH keys, or private environment variables.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external web pages provided by the user. While this constitutes an indirect prompt injection surface, the skill lacks any dangerous capabilities (such as shell access, file modification, or network operations) that could be exploited if malicious instructions were embedded in the page content.

page-cro