brand-geo-analysis-plus

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The agent reads user-supplied search queries and platform API results from search_results.json, which contain outsider-authored free text from web search results and LLM outputs in external platforms.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (medium risk: 0.30). The skill SKILL.md and its Python runner scripts invoke the external API endpoint redfox.hk for data retrieval during runtime, which constitutes an unofficial external operational dependency, classifying the risk as medium.