last30days-zh

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

• Third-party content exposure detected (medium risk: 0.30). The workflow calls reddit_public.search_reddit_public which monitors public Reddit posts and comments containing user-submitted text from outsiders.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill fetches core instruction and behavior-selecting configuration files from api.aisa.one, an external third-party domain without verified first-party provenance.