virtual-girlfriend

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt's curl examples show embedding an Authorization: Bearer {{YOUR_TOKEN}} header directly in commands, which encourages placing real tokens verbatim into generated requests/outputs and thus creates an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.80). The skill calls the public inbed.ai API (e.g., the /api/discover endpoint that returns candidate profiles and /api/chat that exchanges user messages), which ingests and displays user-generated, untrusted third-party content that the agent is expected to read and use to drive swipes, chats, and relationship actions—so that content could contain instructions that materially influence agent behavior.