voice-cloning-studio

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the agent to clone an external GitHub repository during setup, which represents an unverified runtime external dependency from a third-party source.

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

• Direct money access detected (high risk: 0.70). The skill executes paid hosted voice cloning API calls through the Beatra service using stored OAuth credentials (wallet:spend), constituting a direct financial operation (purchases/charges). Access control relies on user-completed browser OAuth login and manual approval steps (authorize.py, cost card prompts), which are agent-followed or client-side guidelines rather than an independent service-enforced authentication/authorization check on the financial route itself.