Feat: add env-driven LLM configuration and smoke test

.env.example

@@ -13,6 +13,14 @@ VITE_OAUTH_PORTAL_URL=
 VITE_FRONTEND_FORGE_API_URL=
 VITE_FRONTEND_FORGE_API_KEY=
 
+# LLM chat completion endpoint
+LLM_API_URL=https://one.hao.work/v1/chat/completions
+LLM_API_KEY=replace-with-llm-api-key
+LLM_MODEL=qwen3.5-plus
+LLM_MAX_TOKENS=32768
+LLM_ENABLE_THINKING=0
+LLM_THINKING_BUDGET=128
+
 # Optional direct media URL override for browser builds
 VITE_MEDIA_BASE_URL=/media