jtbd-interview-analyzer

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

• Third-party content exposure detected (high risk: 0.75). The required workflow ingests “raw JTBD interview transcripts” provided by the operating user, but the skill also explicitly extracts quotes from those transcripts at runtime; if those transcripts include outsider-authored text (e.g., other people’s interview responses), that free-form prose is fed into the agent’s LLM context via the transcript input.