onboard

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

• Third-party content exposure detected (high risk: 0.85). Yes—Phase 2 “Deep Discovery Scan” instructs the agent to read the user’s emails/docs/messages (e.g., Gmail/Drive/Notion/Slack) and extract “real excerpts” that are then embedded into Phase 3/4 context files (e.g., context/voice-dna.md “Writing Samples”), which at runtime becomes LLM-readable prose in the agent’s context; these sources are outsider-authored relative to the operating user.