phone-call

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill contains instructions and configuration that enable large-scale automated outbound calling while advising evasion tactics (e.g., waiting to avoid spam detection) and disabling Do-Not-Call protections, which is suspicious though not explicitly destructive.

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow analyzes call transcripts returned by the fluents.ai Calls API (including “transcript” text) via scripts/analyze_call.py → get_call_details() → analyze_transcript(transcript), so outsider-authored content from the called party is ingested at runtime.