identify locally installed LLM models, runtimes, quantizations, and inference artifacts · runs locally
heuristic screener · parses artifacts locally · not definitive proof