n8n-binary-and-data

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt explicitly instructs injecting storage keys/URLs into the agent's system prompt and telling the model to "use EXACTLY this key," which requires the LLM to reproduce secret keys/URLs verbatim in its generated tool arguments/outputs, creating an exfiltration risk.