geo-analyzer

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

• Third-party content exposure detected (medium risk: 0.30). The workflow reads search results via geo_search.py from external AI platforms (Doubao, Kimi, DeepSeek) where outsider-authored web content is ingested and analyzed.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill utilizes API_BASE = "https://redfox.hk" and fetches APIs from redfox.hk during runtime, which represents a third-party commercial service endpoint with unknown/unverifiable provenance, triggering a high tier.