vapi-voice-call

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

• Third-party content exposure detected (high risk: 0.80). The skill polls Vapi's call API (notably "Polls GET /calls/{id}" and the "transcript" output described in SKILL.md) and ingests user-generated call transcripts from third-party callers which the assistant/assistant-model reads and can act on (e.g., continue dialog or affect call behavior), exposing it to untrusted indirect prompt injection.