continuous-improvement-loop

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to "Read inputs" including "customer feedback aggregations" and "ORM — review sites, social mentions" (Source 2) and to ingest "competitor monitor outputs" (Production Process / Examples), so it clearly consumes untrusted, user-generated third-party content that directly influences recommendations and next actions.