hdsp-jupyter-extension 2.0.13__py3-none-any.whl → 2.0.16__py3-none-any.whl

agent_server/core/llm_service.py

@@ -71,9 +71,10 @@ class LLMService:
     def _get_vllm_config(self) -> tuple[str, str, Dict[str, str]]:
         """Get vLLM config: (model, url, headers)."""
         cfg = self.config.get("vllm", {})
-        endpoint = cfg.get("endpoint", "http://localhost:8000")
+        # User provides full base URL (e.g., https://openrouter.ai/api/v1)
+        endpoint = cfg.get("endpoint", "http://localhost:8000/v1")
         model = cfg.get("model", "default")
-        url = f"{endpoint}/v1/chat/completions"
+        url = f"{endpoint}/chat/completions"
         headers = {"Content-Type": "application/json"}
         if cfg.get("apiKey"):
             headers["Authorization"] = f"Bearer {cfg['apiKey']}"

agent_server/langchain/llm_factory.py

@@ -93,18 +93,21 @@ def _create_vllm_llm(llm_config: Dict[str, Any], callbacks):
     from langchain_openai import ChatOpenAI
 
     vllm_config = llm_config.get("vllm", {})
-    endpoint = vllm_config.get("endpoint", "http://localhost:8000")
+    # User provides full base URL (e.g., https://openrouter.ai/api/v1)
+    endpoint = vllm_config.get("endpoint", "http://localhost:8000/v1")
     model = vllm_config.get("model", "default")
     api_key = vllm_config.get("apiKey", "dummy")
+    use_responses_api = vllm_config.get("useResponsesApi", False)
 
-    logger.info(f"Creating vLLM LLM with model: {model}, endpoint: {endpoint}")
+    logger.info(f"Creating vLLM LLM with model: {model}, endpoint: {endpoint}, use_responses_api: {use_responses_api}")
 
     return ChatOpenAI(
         model=model,
         api_key=api_key,
-        base_url=f"{endpoint}/v1",
+        base_url=endpoint,  # Use endpoint as-is (no /v1 suffix added)
+        use_responses_api=use_responses_api,
         temperature=0.0,
-        max_tokens=4096,
+        max_tokens=32768,
         callbacks=callbacks,
     )
 
@@ -149,14 +152,17 @@ def create_summarization_llm(llm_config: Dict[str, Any]):
             from langchain_openai import ChatOpenAI
 
             vllm_config = llm_config.get("vllm", {})
-            endpoint = vllm_config.get("endpoint", "http://localhost:8000")
+            # User provides full base URL (e.g., https://openrouter.ai/api/v1)
+            endpoint = vllm_config.get("endpoint", "http://localhost:8000/v1")
             model = vllm_config.get("model", "default")
             api_key = vllm_config.get("apiKey", "dummy")
+            use_responses_api = vllm_config.get("useResponsesApi", False)
 
             return ChatOpenAI(
                 model=model,
                 api_key=api_key,
-                base_url=f"{endpoint}/v1",
+                base_url=endpoint,  # Use endpoint as-is
+                use_responses_api=use_responses_api,
                 temperature=0.0,
             )
     except Exception as e:

hdsp_agent_core/llm/service.py

@@ -72,9 +72,10 @@ class LLMService:
     def _get_vllm_config(self) -> tuple[str, str, Dict[str, str]]:
         """Get vLLM config: (model, url, headers)."""
         cfg = self.config.get("vllm", {})
-        endpoint = cfg.get("endpoint", "http://localhost:8000")
+        # User provides full base URL (e.g., https://openrouter.ai/api/v1)
+        endpoint = cfg.get("endpoint", "http://localhost:8000/v1")
         model = cfg.get("model", "default")
-        url = f"{endpoint}/v1/chat/completions"
+        url = f"{endpoint}/chat/completions"
         headers = {"Content-Type": "application/json"}
         if cfg.get("apiKey"):
             headers["Authorization"] = f"Bearer {cfg['apiKey']}"

{hdsp_jupyter_extension-2.0.13.data → hdsp_jupyter_extension-2.0.16.data}/data/share/jupyter/labextensions/hdsp-agent/build_log.json

@@ -722,7 +722,7 @@
             "@mui/material": {},
             "react-markdown": {},
             "hdsp-agent": {
-              "version": "2.0.13",
+              "version": "2.0.16",
               "singleton": true,
               "import": "/Users/a421721/Desktop/hdsp/hdsp_agent/extensions/jupyter/lib/index.js"
             }

{hdsp_jupyter_extension-2.0.13.data → hdsp_jupyter_extension-2.0.16.data}/data/share/jupyter/labextensions/hdsp-agent/package.json

@@ -1,6 +1,6 @@
 {
   "name": "hdsp-agent",
-  "version": "2.0.13",
+  "version": "2.0.16",
   "description": "HDSP Agent JupyterLab Extension - Thin client for Agent Server",
   "keywords": [
     "jupyter",
@@ -132,7 +132,7 @@
       }
     },
     "_build": {
-      "load": "static/remoteEntry.a8e0b064eb9b1c1ff463.js",
+      "load": "static/remoteEntry.44355425a0862dac7bd1.js",
       "extension": "./extension",
       "style": "./style"
     }

jupyter_ext/labextension/static/lib_index_js.5449ba3c7e25177d2987.js → hdsp_jupyter_extension-2.0.16.data/data/share/jupyter/labextensions/hdsp-agent/static/lib_index_js.90a86cec4c50b0798fb2.js

@@ -716,7 +716,7 @@ const ChatPanel = (0,react__WEBPACK_IMPORTED_MODULE_0__.forwardRef)(({ apiServic
         }
         else if (config.provider === 'vllm') {
             console.log('vLLM Model:', config.vllm?.model || 'default');
-            console.log('vLLM Endpoint:', config.vllm?.endpoint || 'http://localhost:8000');
+            console.log('vLLM Endpoint:', config.vllm?.endpoint || 'http://localhost:8000/v1');
             console.log('vLLM API Key:', config.vllm?.apiKey ? '✓ Configured' : '✗ Not configured');
         }
         else if (config.provider === 'openai') {
@@ -4219,9 +4219,10 @@ const SettingsPanel = ({ onClose, onSave, currentConfig }) => {
     };
     const [geminiModel, setGeminiModel] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(validateGeminiModel(initConfig.gemini?.model));
     // vLLM settings
-    const [vllmEndpoint, setVllmEndpoint] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.vllm?.endpoint || 'http://localhost:8000');
+    const [vllmEndpoint, setVllmEndpoint] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.vllm?.endpoint || 'http://localhost:8000/v1');
     const [vllmApiKey, setVllmApiKey] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.vllm?.apiKey || '');
     const [vllmModel, setVllmModel] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.vllm?.model || 'meta-llama/Llama-2-7b-chat-hf');
+    const [vllmUseResponsesApi, setVllmUseResponsesApi] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(Boolean(initConfig.vllm?.useResponsesApi));
     // OpenAI settings
     const [openaiApiKey, setOpenaiApiKey] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.openai?.apiKey || '');
     const [openaiModel, setOpenaiModel] = (0,react__WEBPACK_IMPORTED_MODULE_0__.useState)(initConfig.openai?.model || 'gpt-4');
@@ -4261,9 +4262,10 @@ const SettingsPanel = ({ onClose, onSave, currentConfig }) => {
                 setGeminiApiKeys(['']);
             }
             setGeminiModel(validateGeminiModel(currentConfig.gemini?.model));
-            setVllmEndpoint(currentConfig.vllm?.endpoint || 'http://localhost:8000');
+            setVllmEndpoint(currentConfig.vllm?.endpoint || 'http://localhost:8000/v1');
             setVllmApiKey(currentConfig.vllm?.apiKey || '');
             setVllmModel(currentConfig.vllm?.model || 'meta-llama/Llama-2-7b-chat-hf');
+            setVllmUseResponsesApi(Boolean(currentConfig.vllm?.useResponsesApi));
             setOpenaiApiKey(currentConfig.openai?.apiKey || '');
             setOpenaiModel(currentConfig.openai?.model || 'gpt-4');
             setSystemPrompt(currentConfig.systemPrompt || (0,_services_ApiKeyManager__WEBPACK_IMPORTED_MODULE_10__.getDefaultLLMConfig)().systemPrompt || '');
@@ -4292,7 +4294,8 @@ const SettingsPanel = ({ onClose, onSave, currentConfig }) => {
         vllm: {
             endpoint: vllmEndpoint,
             apiKey: vllmApiKey,
-            model: vllmModel
+            model: vllmModel,
+            useResponsesApi: vllmUseResponsesApi
         },
         openai: {
             apiKey: openaiApiKey,
@@ -4445,16 +4448,21 @@ const SettingsPanel = ({ onClose, onSave, currentConfig }) => {
                             react__WEBPACK_IMPORTED_MODULE_0___default().createElement("option", { value: "gemini-2.5-flash" }, "Gemini 2.5 Flash"),
                             react__WEBPACK_IMPORTED_MODULE_0___default().createElement("option", { value: "gemini-2.5-pro" }, "Gemini 2.5 Pro"))))),
                 provider === 'vllm' && (react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-provider" },
-                    react__WEBPACK_IMPORTED_MODULE_0___default().createElement("h3", null, "vLLM \uC124\uC815"),
+                    react__WEBPACK_IMPORTED_MODULE_0___default().createElement("h3", null, "vLLM / OpenAI Compatible \uC124\uC815"),
                     react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-group" },
-                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("label", { className: "jp-agent-settings-label" }, "\uC11C\uBC84 \uC8FC\uC18C"),
-                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "text", className: "jp-agent-settings-input", value: vllmEndpoint, onChange: (e) => setVllmEndpoint(e.target.value), placeholder: "http://localhost:8000" })),
+                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("label", { className: "jp-agent-settings-label" }, "API Base URL (\uC804\uCCB4 \uACBD\uB85C \uC785\uB825)"),
+                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "text", className: "jp-agent-settings-input", value: vllmEndpoint, onChange: (e) => setVllmEndpoint(e.target.value), placeholder: "https://openrouter.ai/api/v1" })),
                     react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-group" },
                         react__WEBPACK_IMPORTED_MODULE_0___default().createElement("label", { className: "jp-agent-settings-label" }, "API \uD0A4 (\uC120\uD0DD\uC0AC\uD56D)"),
                         react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "password", className: "jp-agent-settings-input", value: vllmApiKey, onChange: (e) => setVllmApiKey(e.target.value), placeholder: "API \uD0A4\uAC00 \uD544\uC694\uD55C \uACBD\uC6B0 \uC785\uB825" })),
                     react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-group" },
                         react__WEBPACK_IMPORTED_MODULE_0___default().createElement("label", { className: "jp-agent-settings-label" }, "\uBAA8\uB378 \uC774\uB984"),
-                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "text", className: "jp-agent-settings-input", value: vllmModel, onChange: (e) => setVllmModel(e.target.value), placeholder: "meta-llama/Llama-2-7b-chat-hf" })))),
+                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "text", className: "jp-agent-settings-input", value: vllmModel, onChange: (e) => setVllmModel(e.target.value), placeholder: "meta-llama/Llama-2-7b-chat-hf" })),
+                    react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-group" },
+                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("label", { className: "jp-agent-settings-checkbox" },
+                            react__WEBPACK_IMPORTED_MODULE_0___default().createElement("input", { type: "checkbox", checked: vllmUseResponsesApi, onChange: (e) => setVllmUseResponsesApi(e.target.checked) }),
+                            react__WEBPACK_IMPORTED_MODULE_0___default().createElement("span", null, "Use Responses API (OpenAI \uC804\uC6A9, /chat/completions \uB300\uC2E0 /responses \uC0AC\uC6A9)")),
+                        react__WEBPACK_IMPORTED_MODULE_0___default().createElement("small", { style: { color: '#666', marginTop: '4px', display: 'block', marginLeft: '24px' } }, "OpenAI \uACF5\uC2DD API\uC5D0\uC11C\uB9CC \uC9C0\uC6D0\uB429\uB2C8\uB2E4. OpenRouter \uB4F1 \uC11C\uB4DC\uD30C\uD2F0\uC5D0\uC11C\uB294 \uC791\uB3D9\uD558\uC9C0 \uC54A\uC744 \uC218 \uC788\uC2B5\uB2C8\uB2E4.")))),
                 provider === 'openai' && (react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-provider" },
                     react__WEBPACK_IMPORTED_MODULE_0___default().createElement("h3", null, "OpenAI \uC124\uC815"),
                     react__WEBPACK_IMPORTED_MODULE_0___default().createElement("div", { className: "jp-agent-settings-group" },
@@ -8609,7 +8617,7 @@ function getDefaultLLMConfig() {
             model: 'gpt-4'
         },
         vllm: {
-            endpoint: 'http://localhost:8000',
+            endpoint: 'http://localhost:8000/v1',
             model: 'default'
         },
         systemPrompt: prompts?.single || DEFAULT_LANGCHAIN_SYSTEM_PROMPT,
@@ -11795,4 +11803,4 @@ __webpack_require__.r(__webpack_exports__);
 /***/ }
 
 }]);
-//# sourceMappingURL=lib_index_js.5449ba3c7e25177d2987.js.map
+//# sourceMappingURL=lib_index_js.90a86cec4c50b0798fb2.js.map