gpu-document-processing

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

• Third-party content exposure detected (high risk: 0.70). The skill’s runtime workflow ingests user-provided document content (“Receive document reference… send to GPU sandbox for processing”), and that document body text is then read by the GPU sandbox and returned to the agent as structured text—if the documents are authored by outsiders, this is an outsider free-text injection path.