channel-formula

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime “Extraction Workflow” ingests “3–5 top videos” and “1–3 flops” from a channel’s own data, which is outsider-authored user-provided content whenever the user supplies those video texts (titles/retention/thumbnails/metrics) into the workflow without selecting trusted first-party documentation items.