foundation-model-agent-loop

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

• Potentially malicious external URL detected (high risk: 0.90). The skill makes runtime POST requests to the Databricks serving endpoint at {host}/serving-endpoints/{endpoint}/invocations (e.g., https:///serving-endpoints/databricks-meta-llama-3-3-70b-instruct/invocations), and the model responses include tool_calls that directly control which local tools/code the agent executes, which the skill depends on to operate.