curviate

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

• Third-party content exposure detected (medium risk: 0.30). The skill acts as a dispatcher for a LinkedIn CLI tool, enabling the agent to read messages, inbox items, comments, and posts authored by outsiders.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 0.70). The skill documentation mentions that job publish and recruiter job publish spend real money ("Two commands take a --mode that is a different flag wearing the same name, and both spend real money"), which constitutes a direct financial operation (purchasing/promoting job postings). However, the skill references CURVIATE_API_KEY for credentials and authentication against the Curviate API, but the documented command surface does not independently enforce a separate qualifying payment authorization control (such as secure checkout-time user prompt or escrow) at the payment layer beyond standard API key usage.