feat: update embedding and LLM provider configurations

.env.example

@@ -53,68 +53,71 @@ STORAGE_PATH=.contextseek/store
 # OB_TABLE_NAME=contextseek_items
 
 # =============================================================================
-# 2. Embedding (LangChain)
+# 2. Embedding
 # =============================================================================
-# Provider: "none" or "langchain"
+# Provider: "none" | "openai" | "dashscope" | "ollama" | "huggingface" | "langchain"
 EMBEDDING_PROVIDER=none
-# Fully qualified Embeddings class (required when provider is langchain)
-# EMBEDDING_CLASS_PATH=langchain_openai.OpenAIEmbeddings
 # EMBEDDING_MODEL=text-embedding-3-small
+# Optional: inferred from known providers when omitted
 # EMBEDDING_DIMS=1536
 # Optional base URL for compatible providers (OpenAI-compatible, Ollama, etc.)
 # EMBEDDING_BASE_URL=
 
 ## Typical provider snippets (uncomment one block as needed)
 # OpenAI:
-#   EMBEDDING_CLASS_PATH=langchain_openai.OpenAIEmbeddings
+#   EMBEDDING_PROVIDER=openai
 #   EMBEDDING_MODEL=text-embedding-3-small
-#   EMBEDDING_DIMS=1536
 #   OPENAI_API_KEY=sk-...
 #   EMBEDDING_BASE_URL=https://api.openai.com/v1
 #
 # DashScope:
-#   EMBEDDING_CLASS_PATH=langchain_community.embeddings.DashScopeEmbeddings
+#   EMBEDDING_PROVIDER=dashscope
 #   EMBEDDING_MODEL=text-embedding-v3
-#   EMBEDDING_DIMS=1024
 #   DASHSCOPE_API_KEY=sk-...
 #
 # Ollama (local):
-#   EMBEDDING_CLASS_PATH=langchain_ollama.OllamaEmbeddings
+#   EMBEDDING_PROVIDER=ollama
 #   EMBEDDING_MODEL=nomic-embed-text
-#   EMBEDDING_DIMS=768
 #   EMBEDDING_BASE_URL=http://127.0.0.1:11434
 #
 # HuggingFace (local):
-#   EMBEDDING_CLASS_PATH=langchain_huggingface.HuggingFaceEmbeddings
+#   EMBEDDING_PROVIDER=huggingface
 #   EMBEDDING_MODEL=BAAI/bge-small-zh-v1.5
-#   EMBEDDING_DIMS=512
+#
+# Advanced custom LangChain class:
+#   EMBEDDING_PROVIDER=langchain
+#   EMBEDDING_CLASS_PATH=your_package.YourEmbeddings
+#   EMBEDDING_DIMS=1536
 
 # =============================================================================
-# 3. LLM (LangChain — rerank / extract / summarize)
+# 3. LLM (rerank / extract / summarize)
 # =============================================================================
-# Provider: "none" or "langchain"
+# Provider: "none" | "openai" | "dashscope" | "ollama" | "langchain"
 LLM_PROVIDER=none
-# LLM_CLASS_PATH=langchain_openai.ChatOpenAI
 # LLM_MODEL=gpt-4o-mini
 # Optional base URL for compatible providers (OpenAI-compatible, Ollama, etc.)
 # LLM_BASE_URL=
 
 ## Typical provider snippets
 # OpenAI:
-#   LLM_CLASS_PATH=langchain_openai.ChatOpenAI
+#   LLM_PROVIDER=openai
 #   LLM_MODEL=gpt-4o-mini
 #   OPENAI_API_KEY=sk-...
 #   LLM_BASE_URL=https://api.openai.com/v1
 #
 # DashScope:
-#   LLM_CLASS_PATH=langchain_community.chat_models.ChatTongyi
+#   LLM_PROVIDER=dashscope
 #   LLM_MODEL=qwen-plus
 #   DASHSCOPE_API_KEY=sk-...
 #
 # Ollama:
-#   LLM_CLASS_PATH=langchain_ollama.ChatOllama
+#   LLM_PROVIDER=ollama
 #   LLM_MODEL=qwen2.5:7b
 #   LLM_BASE_URL=http://127.0.0.1:11434
+#
+# Advanced custom LangChain class:
+#   LLM_PROVIDER=langchain
+#   LLM_CLASS_PATH=your_package.YourChatModel
 
 # =============================================================================
 # 4. Summarizer (drives L2 abstract / L1 overview generation in ContextSeek.add)

dashboard/src/lib/ctxClient.ts

@@ -8,6 +8,8 @@ import type {
   CompactRequest,
   CompactResponse,
   Config,
+  ConfigTestRequest,
+  ConfigTestResponse,
   ConfigUpdateRequest,
   DeleteRequest,
   DreamRequest,
@@ -125,6 +127,7 @@ export const ctx = {
   config: () => get<Config>("/config"),
   updateConfig: (req: ConfigUpdateRequest) =>
     put<{ status: string; restart_required: boolean }>("/config", req),
+  testConfig: (req: ConfigTestRequest) => post<ConfigTestResponse>("/config/test", req),
   restart: async () => {
     const invoke = getTauriInvoke();
     if (invoke) {

dashboard/src/lib/i18n.tsx

@@ -234,6 +234,9 @@ const zh: Dict = {
   "settings.embedder": "Embedder",
   "settings.embedder.desc": "向量嵌入模型配置。",
   "settings.embedder.model": "模型",
+  "settings.embedder.dims": "维度",
+  "settings.provider.label": "Provider",
+  "settings.connectionTest": "连接测试",
   "settings.provider.enabled": "状态",
   "settings.provider.on": "启用",
   "settings.provider.off": "关闭",
@@ -467,6 +470,9 @@ const en: Dict = {
   "settings.embedder": "Embedder",
   "settings.embedder.desc": "Vector embedding model configuration.",
   "settings.embedder.model": "Model",
+  "settings.embedder.dims": "Dimensions",
+  "settings.provider.label": "Provider",
+  "settings.connectionTest": "Connection test",
   "settings.provider.enabled": "Status",
   "settings.provider.on": "Enabled",
   "settings.provider.off": "Off",
@@ -700,6 +706,9 @@ const ja: Dict = {
   "settings.embedder": "Embedder",
   "settings.embedder.desc": "ベクトル埋め込みモデル設定。",
   "settings.embedder.model": "モデル",
+  "settings.embedder.dims": "次元",
+  "settings.provider.label": "Provider",
+  "settings.connectionTest": "接続テスト",
   "settings.provider.enabled": "状態",
   "settings.provider.on": "有効",
   "settings.provider.off": "無効",

dashboard/src/lib/types.ts

@@ -200,6 +200,7 @@ export interface Config {
   llm_api_key?: string;
   embedding_provider?: string;
   embedding_model: string;
+  embedding_dims?: string;
   embedding_base_url?: string;
   embedding_api_key?: string;
   default_scope: string;
@@ -232,6 +233,7 @@ export interface ConfigUpdateRequest {
   llm_api_key?: string;
   embedding_provider?: string;
   embedding_model?: string;
+  embedding_dims?: string;
   embedding_base_url?: string;
   embedding_api_key?: string;
   ob_host?: string;
@@ -246,6 +248,23 @@ export interface ConfigUpdateRequest {
   storage_path?: string;
 }
 
+export interface ConfigTestRequest {
+  target: "llm" | "embedding";
+  provider: string;
+  model?: string;
+  base_url?: string;
+  api_key?: string;
+  dims?: string;
+}
+
+export interface ConfigTestResponse {
+  ok: boolean;
+  message: string;
+  detail?: string;
+  dimension?: number;
+  configured_dimension?: number;
+}
+
 // ---- request payloads ----
 
 export interface AddRequest {