phone-agent

Warn

Audited by Snyk on Mar 4, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 1.00). The skill ingests untrusted, user-generated content from incoming phone calls via the Twilio WebSocket (scripts/server.py websocket_endpoint -> deepgram_receiver / process_transcript) and feeds those transcripts into the LLM to drive TTS responses and outbound call logic, which could allow indirect prompt injection.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 4, 2026, 04:26 PM