robolab-taskgen

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

• Third-party content exposure detected (medium risk: 0.30). The skill consumes outsider-provided natural-language “Task instruction” text directly and uses it to generate task metadata/termination parameters (and to build "default"/instruction variants) in the post-input generation workflow, so attacker-authored free text is ingested at runtime without first selecting a specific trusted item.