cps-rca-central-african-republic-jeanne-sulzer

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow ingests and reads third-party free text only from cpsrca.cf (and other verification sources like MINUSCA/JusticeInfo/HRW) via web_fetch during its verification/fallback ladder, which means outsider-authored content can be fetched and used.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires a runtime web_fetch to the CPS website (cpsrca.cf / https://www.cpsrca.cf/) to verify citations in the current conversation, meaning external content fetched at runtime will be injected into and directly influence the agent's outputs.