iohann.s.titov@gmail.com Ivan S. Titov Hollama is a minimal LLM chat UI that runs entirely in the browser — all conversation state is stored in the browser's localStorage, no server-side database. It talks to Ollama natively and to any OpenAI-compatible HTTP endpoint, so it pairs cleanly with this overlay's local-LLM backends (sci-misc/llama-cpp, sci-misc/llama-swap, sci-ml/lemonade, dev-python/vllm). Packaged as a SvelteKit Node bundle (`PUBLIC_ADAPTER=docker-node`) served by /usr/bin/hollama; systemd unit and openrc init.d scripts shipped under /etc/. Defaults to loopback-only. fmaclen/hollama