writing-shape

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

• Third-party content exposure detected (high risk: 0.70). The required workflow explicitly instructs the agent to “Read the input file in full” and “Pull material from the pile,” meaning any outsider-authored markdown file content provided by the user (e.g., forwarded/downloaded text) would be ingested as free-form prose into the LLM context.