manualize

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

• Third-party content exposure detected (low risk: 0.10). In the required runtime workflow, the agent ingests outsider-authored free text only via user-supplied local files passed as inputs (e.g., request_file=/absolute/request.md → --input manual=/absolute/request.md), and then validates that text offline as Markdown without selecting any specific item from a broader feed.