What this is: a browser-only retrieval demo. It ranks real commands from this Space's 70,006-command base for your English intent (BM25 over command, title and description). Every result is a command that exists in the base. It does not run the LoRA model (that runs locally with infer_rag.py from the model repo). Values such as VLAN ids or IPs are examples from the base, so adapt them. Review before running on a device.