gemini-gen

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

• Third-party content exposure detected (high risk: 0.80). The skill (SKILL.md and scripts/gemini.mjs) explicitly accepts arbitrary file URIs via --file-uri and includes those remote media URLs in requests to the Renoise/Gemini pipeline so the model will ingest and analyze untrusted public media content (images/videos), which could contain instructions that alter its behavior.