No UI field to set custom Ollama host/port — app hardcodes 0.0.0.0 which fails as a client address on Windows

User Sir_SabreWolf on Windows 11 runs an existing Ollama server (alongside OpenClaw, Open WebUI, LM Studio) on port 11434. OpenHuman cannot detect the existing server at 0.0.0.0:11434 and instead tries to bootstrap a fresh Ollama binary — which then fails because the port is already occupied. Issue 1 — no custom server UI field (feature gap, confirmed by team) There is no UI input to point OpenHuman at an existing Ollama instance. The only flow is "install a fresh binary." Team confirmed this is a real gap and committed to adding a UI field. Workaround provided: manually edit %USERPROFILE%\.openhuman\config.toml and set base_url = "http://127.0.0.1:11434/" under [local_ai], then restart. Issue 2 — hardcoded 0.0.0.0 used as client address (bug) After the new build, OpenHuman attempts to reach Ollama at http://0.0.0.0:11434/ — a server-side bind address that is invalid as a client-side connection target on Windows. Ollama is confirmed running and responding at localhost:11434. Error in debug: "ollama request failed... error sending request for url http://0.0.0.0:11434/api/generate". Fix: change endpoint to http://localhost:11434/ or http://127.0.0.1:11434/ in config (config.yaml, .env, or env var OLLAMA_HOST). Resolution unconfirmed — follow-up sent 22/05/2026 and again 28/05/2026 with no response yet.