content-production

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

• Third-party content exposure detected (high risk: 0.80). The skill's Research & Brief workflow explicitly requires identifying the "top 5-10 ranking pieces" and to "collect 3-5 credible, citable sources" (SKILL.md, Mode 1 Step 1 & Step 2) and even calls out forum results like Reddit/Quora in the intent signals, so the agent is expected to ingest and act on untrusted public web/user-generated content as part of its workflow.