ok Merge branch 'main' of github.com:modelscope/ms-agent into release/1.1

Author: wangxingjun778

README.md

@@ -95,7 +95,11 @@ MS-Agent is a lightweight framework designed to empower agents with autonomous e
 ### Install from PyPI
 
 ```shell
+# For the basic functionalities
 pip install ms-agent
+
+# For the deep research functionalities
+pip install 'ms-agent[research]'
 ```
 
 

ms_agent/app/doc_research.py

@@ -2393,9 +2393,9 @@ def periodic_status_update(request: gr.Request):
                     'https://arxiv.org/abs/2505.09388\nhttps://arxiv.org/abs/2412.15115'
                 ],
                 [
-                    'Analyze and summarize the following documents in English',
+                    'Analyze and summarize the following documents. You must use English to answer.',
                     None,
-                    'https://arxiv.org/abs/2505.09388'
+                    'https://arxiv.org/abs/1706.03762'
                 ]
             ],
             inputs=[user_prompt, uploaded_files, urls_text],

ms_agent/tools/mcp_client.py

@@ -66,12 +66,12 @@ async def call_tool(self, server_name: str, tool_name: str,
         if response.isError:
             sep = '\n\n'
             if all(isinstance(item, str) for item in response.content):
-                return f'execute error: {sep.join(response.content)}'
+                return f'execute tool call error: [{server_name}]{tool_name}, {sep.join(response.content)}'
             else:
                 item_list = []
                 for item in response.content:
                     item_list.append(item.text)
-                return f'execute error: {sep.join(item_list)}'
+                return f'execute tool call error: [{server_name}]{tool_name}, {sep.join(item_list)}'
         for content in response.content:
             if content.type == 'text':
                 texts.append(content.text)
@@ -205,16 +205,26 @@ async def connect(self):
         assert self.mcp_config, 'MCP config is required'
         envs = Env.load_env()
         mcp_config = self.mcp_config['mcpServers']
+        error = dict()
         for name, server in mcp_config.items():
-            env_dict = server.pop('env', {})
-            env_dict = {
-                key: value if value else envs.get(key, '')
-                for key, value in env_dict.items()
-            }
-            if 'exclude' in server:
-                self._exclude_functions[name] = server.pop('exclude')
-            await self.connect_to_server(
-                server_name=name, env=env_dict, **server)
+            try:
+                env_dict = server.pop('env', {})
+                env_dict = {
+                    key: value if value else envs.get(key, '')
+                    for key, value in env_dict.items()
+                }
+                if 'exclude' in server:
+                    self._exclude_functions[name] = server.pop('exclude')
+                await self.connect_to_server(
+                    server_name=name, env=env_dict, **server)
+            except BaseException as exc:
+                error[name] = str(exc)
+        if error:
+            error_messages = '; '.join(f'`{srv}`: {msg}'
+                                       for srv, msg in error.items())
+            raise ConnectionError(
+                f'MCP connections failed for: {error_messages}. Please check mcp configurations and retry.'
+            )
 
     async def cleanup(self):
         """Clean up resources"""

projects/deep_research/README.md

@@ -47,11 +47,14 @@ So, how are external electric fields typically simulated in theoretical calculat
 
 To set up the Agentic Insight framework, follow these steps:
 
-* Install from source code
+* Installation
 ```bash
+# From source code
 git clone https://github.com/modelscope/ms-agent.git
-
 pip install -r requirements/research.txt
+
+# From PyPI (>=v1.1.0)
+pip install 'ms-agent[research]'
 ```
 
 ### 🚀 Quickstart

projects/doc_research/README.md

@@ -25,7 +25,7 @@
   - 📝 **Multiple Input Types** - Support multi-file uploads and URL inputs
   - 📊 **Multimodal Reports** - Support text and image reports in Markdown format
   - 🚀 **High Efficiency** - Leverage powerful LLMs for fast and accurate research, leveraging key information extraction techniques to further optimize token usage
-  - ⚙️ **Flexible Deployment** - Support local run and [ModelScope Studio](https://modelscope.cn/studios)
+  - ⚙️ **Flexible Deployment** - Support local run and [ModelScope Studio](https://modelscope.cn/studios) on both CPU and GPU environments.
   - 💰 **Free Model Inference** - Free LLM API inference calls for ModelScope users, refer to [ModelScope API-Inference](https://modelscope.cn/docs/model-service/API-Inference/intro)
 
 
@@ -56,7 +56,7 @@ conda create -n doc_research python=3.11
 conda activate doc_research
 
 # Version requirement: ms-agent>=1.1.0
-pip install ms-agent[research]
+pip install 'ms-agent[research]'
 ```
 
 ### 2. Configure Environment Variables
@@ -99,7 +99,7 @@ ms-agent app --doc_research \
 > `share`: (store_true action), whether to share the app publicly. <br>
 
 * Notes
-> When running locally, the default address is http://0.0.0.0:7860/. You need to disable your VPN to access it.
+> When running locally, the default address is http://0.0.0.0:7860/. If the page can't be accessed, try disabling proxy.
 
 
 <br>

projects/doc_research/README_zh.md

@@ -25,7 +25,7 @@
 - 📝 **多种输入类型** - 支持多文件上传和URLs输入
 - 📊 **多模态报告** - 支持Markdown格式的图文报告输出
 - 🚀 **精准高效** - 利用强大的LLM进行快速准确的研究，采用关键信息抽取技术进一步优化了token使用
-- ⚙️ **灵活部署** - 支持本地运行和魔搭创空间运行模式
+- ⚙️ **灵活部署** - 支持本地运行和魔搭创空间运行模式，兼容CPU和GPU环境
 - 💰 **免费模型推理** - 魔搭ModelScope用户可免费调用LLM API推理，参考 [ModelScope API-Inference](https://modelscope.cn/docs/model-service/API-Inference/intro)
 
 
@@ -56,7 +56,7 @@ conda create -n doc_research python=3.11
 conda activate doc_research
 
 # 版本要求：ms-agent>=1.1.0
-pip install ms-agent[research]
+pip install 'ms-agent[research]'
 ```
 
 ### 2. 配置环境变量
@@ -101,7 +101,7 @@ ms-agent app --doc_research \
 > `share`: (store_true action), 是否对外分享，默认关闭.  <br>
 
 * 备注：
-  > 本地运行时，默认访问地址为 `http://0.0.0.0:7860/` ，需关闭VPN才能访问  <br>
+  > 本地运行时，默认访问地址为 `http://0.0.0.0:7860/` ，如无法访问，可尝试关闭VPN  <br>
 
 
 <br>