atai-task-verification-agent

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

• Third-party content exposure detected (medium risk: 0.30). SKILL runtime path uploads user-authored SOP free text via the required {endpoint}/v0.5/files step (multipart file with text/plain), then runs TVA with that text through source.text -> prepare.in at {endpoint}/agents/bundles/{bundle_id}/run, so an outsider can supply arbitrary SOP text that the LLM ingests.