Endpoint Lock: lock down self-hosted LLMs before attack agents run up your bill

An outside check plus an auth proxy for Ollama, vLLM and llama.cpp servers. Paid for by local-LLM users and small GPU teams.