cortex-classify-tutorial

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to web_fetch the Snowflake CLASSIFY_TEXT docs at runtime and "use the fetched docs as the source of truth" (https://docs.snowflake.com/en/sql-reference/functions/classify_text-snowflake-cortex), meaning remote content fetched at runtime will directly control prompts/instructions.