databricks-mlflow-evaluation

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill loads a system prompt at runtime from the MLflow prompt registry (prompts:/{PROMPT_NAME}@production) via mlflow.genai.load_prompt, so externally-stored prompt content is fetched during execution and directly controls the agent's prompts.