coffee-brewing

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

• Third-party content exposure detected (medium risk: 0.30). The skill ingests free text and images provided by the end user in the required runtime workflow (“Trigger when the user shares coffee photos…or reports a shot tasting sour, bitter, weak, or ‘off’”), which is outsider-authored content that the agent directly reads in Step 1/Step 4.