multitask

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s workflow ingests “each ticket” (and “touched code”) into per-lane workers via task-to-pr; if those tickets are authored by outsiders (e.g., issue/PR/discussion text from non-operating users), their free-form body text becomes LLM-readable context at runtime through the ticket-reading step.