n8n-binary-and-data-official

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs injecting file storage keys into the LLM system prompt and requires the LLM to "Use EXACTLY the fileName field when calling tools," forcing the model to emit those keys verbatim in tool calls (exposing secret file keys/URLs).

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

• Third-party content exposure detected (medium risk: 0.30). n8n Chat Trigger (user-supplied chat files) feeds files[] that are pre-staged and key/filenames are injected so the agent can call tools that later download from storage, meaning outsider-authored free text (the user’s message + associated file metadata) enters the runtime flow that drives tool selection.