voice-inbox

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md requires calling Monologue tooling (monologue notes all, monologue notes get NOTE_ID) to retrieve the user’s note titles/summaries and likely source text for verification, which is outsider-authored free text from the speaker(s) recorded in the voice notes being ingested into the agent’s LLM context.