magnific

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

• Third-party content exposure detected (high risk: 0.85). Skill runtime ingests outsider-authored free text via user-supplied prompt/text/description fields (e.g., POST /v1/ai/music-generation, /v1/ai/sound-effects, /v1/ai/audio-isolation, and many generation endpoints), which the operating user did not pre-author into the skill’s own context.