problem-framing-canvas

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md runs an interactive flow where the agent directly ingests outsider-authored free text provided by the user (“You can paste this content directly…”, and then “User response: [selection or custom]” across Q1–Q3 and further questions), without restricting it to trusted first-party sources.