multilingual-score

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly accepts original and translated content as URLs ("Input Required" — Original content: ... or a URL; Translated content: ... or a URL) and the Process steps (technical scoring, eval-runner, brand-voice checks) require reading and interpreting that content to compute scores and drive publish/review/re-translate decisions, so arbitrary public or user-provided webpages could be fetched and thereby inject instructions that influence agent actions.