gemini-files-api

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

• Third-party content exposure detected (high risk: 0.75). At runtime, the required workflow in scripts/gemini-mm.mjs ingests outsider-authored free text from CLI args (--prompt / --system) into the LLM request via createUserContent(parts) and systemInstruction, and also reads user-supplied JSON config/schema files as text/JSON that can contain prompt-like instructions.