get_notebook_mcp_server 1.0.3 → 1.1.0

package/README.md

@@ -1,15 +1,21 @@
-# Get Notes MCP Server
+# Get Notes MCP Server / Get 笔记 MCP 服务器
+
+[English](#english) | [中文](#chinese)
+
+<a name="english"></a>
+
+## 🇬🇧 English
 
 A Model Context Protocol (MCP) server for integrating with Get Notes API. This server provides tools to search and recall knowledge from your Get Notes knowledge base.
 
-## Features
+### Features
 
 - **Knowledge Search**: AI-processed search that returns synthesized answers and references.
 - **Knowledge Recall**: Raw recall of relevant notes and files.
 - **Rate Limiting**: Built-in protection with QPS < 2 and Total Requests < 5000 limits.
 - **Retry Mechanism**: Automatic retries for transient network errors (5xx).
 
-## Installation
+### Installation
 
 1. Clone the repository
 2. Install dependencies:
@@ -22,90 +28,448 @@ A Model Context Protocol (MCP) server for integrating with Get Notes API. This s
    ```
 4. Configure your API key in `.env`:
    ```
-   GET_API_KEY=your_api_key_here
+   GET_API_KEY=T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=
+   GET_NOTE_TOPIC_ID=RYk1kmRJ
    ```
 
-## Usage
+### Usage
 
-### Running the Server
+#### Running with npx
+
+You can run the server directly without installing it using `npx`:
 
 ```bash
-node index.js
+npx get_notebook_mcp_server
 ```
 
-### Testing with MCP Inspector
+#### MCP Configuration (Claude Desktop)
 
-You can test the MCP server interactively using the MCP Inspector:
+**Option 1: Single Knowledge Base (Simple Setup)**
 
-```bash
-npx @modelcontextprotocol/inspector node index.js
+For a single knowledge base, add this to your `claude_desktop_config.json`:
+
+```json
+{
+  "mcpServers": {
+    "get-notes": {
+      "command": "npx",
+      "args": ["-y", "get_notebook_mcp_server"],
+      "env": {
+        "GET_API_KEY": "T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=",
+        "GET_NOTE_TOPIC_ID": "RYk1kmRJ"
+      }
+    }
+  }
+}
 ```
 
-### Running with npx
+**Option 2: Multiple Knowledge Bases (Grouped Configuration - Recommended)**
 
-You can run the server directly without installing it using `npx`:
+For multiple knowledge bases sharing the same API key, use grouped configuration. The system will automatically expand them into individual knowledge bases:
 
-```bash
-npx get_notebook_mcp_server
+```json
+{
+  "mcpServers": {
+    "get-notes": {
+      "command": "npx",
+      "args": ["-y", "get_notebook_mcp_server"],
+      "env": {
+        "GET_KNOWLEDGE_BASES": "[{\"id\":\"medical_group\",\"name\":\"Medical Knowledge Bases\",\"config\":{\"api_key\":\"T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=\",\"topics\":[{\"id\":\"kb_pancreatic_trials\",\"name\":\"Tumor Clinical Trials KB\",\"topic_id\":\"RYk1kmRJ\"},{\"id\":\"kb_patient_experience\",\"name\":\"Patient Experience KB\",\"topic_id\":\"mnyV9OjY\"}]}}]"
+      }
+    }
+  }
+}
 ```
 
-### MCP Configuration (Claude Desktop)
+This configuration will be automatically expanded into two independent knowledge bases: `kb_pancreatic_trials` and `kb_patient_experience`.
+
+**Option 3: Using Local Configuration File (Recommended for Development)**
+
+Create a `knowledge_bases.json` file in your project root:
+
+```json
+[
+  {
+    "id": "demo_group",
+    "name": "Sam's Knowledge Bases",
+    "description": "Knowledge bases sharing the same API Key",
+    "config": {
+      "api_key": "T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=",
+      "api_endpoint": "https://open-api.biji.com/getnote/openapi",
+      "topics": [
+        {
+          "id": "kb_pancreatic_trials",
+          "name": "Tumor Clinical Trials Knowledge Base",
+          "description": "Knowledge base about tumor clinical trials",
+          "topic_id": "RYk1kmRJ"
+        },
+        {
+          "id": "kb_patient_experience",
+          "name": "Patient Experience Information",
+          "description": "Patient experience info including guidelines and treatment experiences",
+          "topic_id": "mnyV9OjY"
+        }
+      ]
+    }
+  }
+]
+```
 
-Add the following configuration to your `claude_desktop_config.json`:
+Then configure Claude Desktop to use the local file:
 
 ```json
 {
   "mcpServers": {
     "get-notes": {
       "command": "npx",
-      "args": [
-        "-y",
-        "get_notebook_mcp_server"
-      ],
-      "env": {
-        "GET_API_KEY": "your_api_key_here",
-        "GET_NOTE_TOPIC_ID": "your_topic_id_here"
-      }
+      "args": ["-y", "get_notebook_mcp_server"]
     }
   }
 }
 ```
 
+#### How Claude Uses Multiple Knowledge Bases
+
+**Claude automatically selects the appropriate knowledge base - no manual selection needed!**
+
+**Workflow:**
+1. **User asks a question**: "How long after pancreatic cancer surgery should chemotherapy start?"
+2. **Claude automatically calls `list_knowledge_bases`**: Views all available knowledge bases
+3. **Claude intelligently selects**: Based on the question content, automatically chooses the most relevant KB (e.g., `kb_pancreatic_trials`)
+4. **Claude calls `search_knowledge`**: Passes the selected `kb_id` and the question
+5. **Returns results**: Claude synthesizes the search results into an answer
+
+**Users can also specify explicitly:**
+- "Search in the Tumor Clinical Trials KB: pancreatic cancer chemotherapy timing"
+- "Find in Patient Experience: chemotherapy side effect management"
+
+This way Claude knows exactly which knowledge base to use.
+
+#### Running Locally
+
+```bash
+node index.js
+```
+
+#### Testing with MCP Inspector
+
+You can test the MCP server interactively using the MCP Inspector:
+
+```bash
+npx @modelcontextprotocol/inspector node index.js
+```
+
 ### Tools
 
+#### `list_knowledge_bases`
+List all available knowledge bases with their IDs, names, and descriptions.
+
+**Parameters:** None
+
+**Returns:** Array of knowledge base information (without sensitive API keys)
+
 #### `search_knowledge`
 Search the knowledge base with AI processing.
 
 **Parameters:**
+- `kb_id` (string, optional): Knowledge base ID. If not provided, uses the first configured KB.
 - `question` (string, required): The question to ask.
-- `topic_ids` (array<string>, optional): List of knowledge base IDs. If not provided, uses `GET_NOTE_TOPIC_ID` from environment.
-- `deep_seek` (boolean): Enable deep thinking mode (default: true).
+- `deep_seek` (boolean): Enable deep thinking mode (default: false).
 - `history` (array): Chat history for context.
 
 #### `recall_knowledge`
 Raw recall from knowledge base without AI synthesis.
 
 **Parameters:**
+- `kb_id` (string, optional): Knowledge base ID. If not provided, uses the first configured KB.
 - `question` (string, required): The question or query.
-- `topic_id` (string, optional): Knowledge base ID. If not provided, uses `GET_NOTE_TOPIC_ID` from environment.
 - `top_k` (number): Number of results to return (default: 10).
 - `intent_rewrite` (boolean): Enable intent rewrite (default: false).
 
-## Development
+### Development
 
-### Running Tests
+#### Running Tests
 
 ```bash
 npm test
 ```
 
-### Project Structure
+#### Project Structure
 
 - `src/api`: API client implementation
 - `src/utils`: Utility classes (RateLimiter, etc.)
 - `tests`: Unit and integration tests
 - `index.js`: Main MCP server entry point
 
-## License
+### License
+
+ISC
+
+### ❤️ Acknowledgments
+
+This project is contributed by the **Xiao-X-Bao Community** - an AI open-source public welfare community focused on improving medical information access for cancer/rare disease patients and their families throughout their life cycle.
+
+#### About Xiao-X-Bao Community
+
+We are an AI open-source public welfare community dedicated to using AI technology to build a high-quality, high-density patient knowledge ecosystem to improve medical information barriers. Our community brings together a diverse group of open-source contributors, AI technology experts, cancer patients, family members, medical professionals, and public welfare volunteers, working together to improve the information access challenges faced by patients.
+
+**Community Website**: https://info.xiao-x-bao.com.cn
+
+**Our Mission**: To use AI technology to break down medical information barriers and build a comprehensive knowledge ecosystem for patients.
+
+**Origin of Community Name**: All our AI assistants are named "Xiao-X-Bao" (Little X Treasure), such as "Xiao Yi Bao" (Little Pancreas Treasure), "Xiao Fen Bao" (Little Pink Treasure), etc., hence our community is named "Xiao-X-Bao Community".
+
+**Community Development**: In 2024, the project was donated to the "Tiangong Kaiwu Foundation", which provides funding and guidance to promote public welfare initiatives.
+
+**Progress Updates**: Published via the WeChat Official Account "Xiao Yi Bao Assistant".
+
+#### Community Contact
+
+**Detailed Community Introduction**: [View Introduction](https://uei55ql5ok.feishu.cn/wiki/ULuKwHuv1iEM8gkrFTecP8DqnSd)
+
+**Contact Us**:
+- WeChat: qinxiaoqiang2002
+- WeChat: hhxdeweixinxin
+- WeChat: zhuangbiaowei
+
+We welcome all contributors, developers, medical professionals, and volunteers to join our community and use technology to improve medical information accessibility! 🌟
+
+---
+
+<a name="chinese"></a>
+
+## 🇨🇳 中文
+
+这是一个用于集成 Get 笔记 API 的 Model Context Protocol (MCP) 服务器。该服务器提供了从您的 Get 笔记知识库中搜索和召回知识的工具。
+
+### 功能特性
+
+- **知识库搜索 (Knowledge Search)**：经过 AI 处理的搜索，返回综合的答案和引用。
+- **知识库召回 (Knowledge Recall)**：相关笔记和文件的原始召回。
+- **速率限制 (Rate Limiting)**：内置保护，限制 QPS < 2 和总请求数 < 5000。
+- **重试机制 (Retry Mechanism)**：针对瞬时网络错误 (5xx) 的自动重试。
+
+### 安装
+
+1. 克隆仓库
+2. 安装依赖：
+   ```bash
+   npm install
+   ```
+3. 从示例创建 `.env` 文件：
+   ```bash
+   cp .env.example .env
+   ```
+4. 在 `.env` 中配置您的 API 密钥：
+   ```
+   GET_API_KEY=T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=
+   GET_NOTE_TOPIC_ID=RYk1kmRJ
+   ```
+
+### 使用方法
+
+#### 使用 npx 运行
+
+您可以直接使用 `npx` 运行服务器，无需安装：
+
+```bash
+npx get_notebook_mcp_server
+```
+
+#### MCP 配置 (Claude Desktop)
+
+将以下配置添加到您的 `claude_desktop_config.json`：
+
+**方式1：单个知识库（简单配置）**
+
+```json
+{
+  "mcpServers": {
+    "get-notes": {
+      "command": "npx",
+      "args": [
+        "-y",
+        "get_notebook_mcp_server"
+      ],
+      "env": {
+        "GET_API_KEY": "T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=",
+        "GET_NOTE_TOPIC_ID": "RYk1kmRJ"
+      }
+    }
+  }
+}
+```
+
+**方式2：多知识库（分组配置，推荐）**
+
+如果多个知识库共享同一个 API Key，可以使用分组配置。系统会自动将其展开为独立的知识库：
+
+```json
+{
+  "mcpServers": {
+    "get-notes-multi": {
+      "command": "npx",
+      "args": [
+        "-y",
+        "get_notebook_mcp_server"
+      ],
+      "env": {
+        "GET_KNOWLEDGE_BASES": "[{\"id\":\"medical_group\",\"name\":\"医学知识库组\",\"config\":{\"api_key\":\"T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=\",\"topics\":[{\"id\":\"kb_pancreatic_trials\",\"name\":\"肿瘤临床试验知识库\",\"topic_id\":\"RYk1kmRJ\"},{\"id\":\"kb_patient_experience\",\"name\":\"小胰宝病友经验信息\",\"topic_id\":\"mnyV9OjY\"}]}}]"
+      }
+    }
+  }
+}
+```
+
+上述配置会自动展开为两个独立的知识库：`kb_pancreatic_trials` 和 `kb_patient_experience`。
+
+**方式3：使用本地配置文件（推荐用于开发）**
+
+在项目根目录创建 `knowledge_bases.json` 文件：
+
+```json
+[
+  {
+    "id": "demo_group",
+    "name": "sam的知识库",
+    "description": "使用同一个 API Key 的知识库",
+    "config": {
+      "api_key": "T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=",
+      "api_endpoint": "https://open-api.biji.com/getnote/openapi",
+      "topics": [
+        {
+          "id": "kb_pancreatic_trials",
+          "name": "肿瘤临床试验知识库",
+          "description": "有关肿瘤临床试验的知识库",
+          "topic_id": "RYk1kmRJ"
+        },
+        {
+          "id": "kb_patient_experience",
+          "name": "小胰宝病友经验信息",
+          "description": "小胰宝病友经验信息，包括各种指南和治疗经验",
+          "topic_id": "mnyV9OjY"
+        }
+      ]
+    }
+  }
+]
+```
+
+然后在 Claude Desktop 配置中使用本地文件：
+
+```json
+{
+  "mcpServers": {
+    "get-notes": {
+      "command": "npx",
+      "args": ["-y", "get_notebook_mcp_server"]
+    }
+  }
+}
+```
+
+#### Claude 如何使用多知识库
+
+**Claude 会自动选择合适的知识库，无需手动选择！**
+
+**工作流程：**
+1. **用户提问**："胰腺癌术后多久开始化疗？"
+2. **Claude 自动调用 `list_knowledge_bases`**：查看所有可用的知识库
+3. **Claude 智能选择**：根据问题内容，自动选择最相关的知识库（如 `kb_pancreatic_trials`）
+4. **Claude 调用 `search_knowledge`**：传入选定的 `kb_id` 和问题
+5. **返回结果**：Claude 综合搜索结果给出答案
+
+**用户也可以明确指定：**
+- "在肿瘤临床试验知识库中搜索：胰腺癌化疗方案"
+- "在小胰宝病友经验中查找：化疗副作用处理方法"
+
+这样 Claude 就会明确使用指定的知识库。
+
+#### 本地运行
+
+```bash
+node index.js
+```
+
+#### 使用 MCP Inspector 测试
+
+您可以使用 MCP Inspector 交互式测试 MCP 服务器：
+
+```bash
+npx @modelcontextprotocol/inspector node index.js
+```
+
+### 工具说明
+
+#### `list_knowledge_bases`
+列出所有可用的知识库及其 ID、名称和描述。
+
+**参数：** 无
+
+**返回：** 知识库信息数组（不包含敏感的 API 密钥）
+
+#### `search_knowledge`
+使用 AI 处理搜索知识库。
+
+**参数：**
+- `kb_id` (string, 可选): 知识库 ID。如果未提供，使用第一个配置的知识库。
+- `question` (string, 必需): 要提问的问题。
+- `deep_seek` (boolean): 启用深度思考模式（默认值：false）。
+- `history` (array): 用于上下文的聊天历史。
+
+#### `recall_knowledge`
+从知识库原始召回，不进行 AI 综合。
+
+**参数：**
+- `kb_id` (string, 可选): 知识库 ID。如果未提供，使用第一个配置的知识库。
+- `question` (string, 必需): 问题或查询。
+- `top_k` (number): 返回结果数量（默认值：10）。
+- `intent_rewrite` (boolean): 启用意图重写（默认值：false）。
+
+### 开发
+
+#### 运行测试
+
+```bash
+npm test
+```
+
+#### 项目结构
+
+- `src/api`：API 客户端实现
+- `src/utils`：工具类（速率限制器等）
+- `tests`：单元和集成测试
+- `index.js`：MCP 服务器主入口点
+
+### 许可证
 
 ISC
+
+### ❤️ 致谢
+
+本项目由**小x宝社区**贡献 - 一个专注于改善癌症/罕见病患者及其家庭整个生命周期医疗信息获取的AI开源公益社区。
+
+#### 关于小x宝社区
+
+我们是一个AI开源公益社区，致力于使用AI技术构建高质量、高密度的患者知识生态系统，以改善医疗信息壁垒。我们的社区汇聚了开源贡献者、AI技术专家、癌症患者、家庭成员、医疗专业人员和公益志愿者等多元化群体，共同努力改善患者面临的信息获取挑战。
+
+**社区网站**：https://info.xiao-x-bao.com.cn
+
+**我们的使命**：使用AI技术打破医疗信息壁垒，为患者构建全面的知识生态系统。
+
+**社区名称由来**：我们所有的AI助手都以"小x宝"命名，如"小胰宝"、"小粉宝"等，因此我们的社区命名为"小x宝社区"。
+
+**社区发展**：2024年，项目捐赠给"天工开物基金会"，该基金会提供资金和指导以促进公益倡议。
+
+**进展更新**：通过微信公众号"小胰宝助手"发布。
+
+#### 社区联系方式
+
+**详细社区介绍**： [查看介绍](https://uei55ql5ok.feishu.cn/wiki/ULuKwHuv1iEM8gkrFTecP8DqnSd)
+
+**社区联系方式**：
+- 微信号: qinxiaoqiang2002
+- 微信号: hhxdeweixinxin  
+- 微信号: zhuangbiaowei
+
+我们欢迎所有贡献者、开发人员、医疗专业人员和志愿者加入我们的社区，使用技术改善医疗信息可及性！🌟

package/index.js

@@ -9,19 +9,10 @@ import { z } from 'zod';
 import dotenv from 'dotenv';
 import { ApiClient } from './src/api/client.js';
 import logger from './src/logger.js';
+import { loadKnowledgeBases, getKnowledgeBases, getKnowledgeBaseById } from './src/config/knowledgeBases.js';
 
 dotenv.config();
 
-const API_KEY = process.env.GET_API_KEY;
-const DEFAULT_TOPIC_ID = process.env.GET_NOTE_TOPIC_ID;
-
-if (!API_KEY) {
-    logger.error('GET_API_KEY environment variable is required');
-    process.exit(1);
-}
-
-const apiClient = new ApiClient({ apiKey: API_KEY });
-
 const server = new Server(
     {
         name: 'get-notes-mcp',
@@ -37,25 +28,33 @@ const server = new Server(
 /**
  * Tool Definitions
  */
+const LIST_KNOWLEDGE_BASES_TOOL = {
+    name: 'list_knowledge_bases',
+    description: 'List available knowledge bases with their IDs and descriptions. You MUST use this tool first to find the correct kb_id for search_knowledge or recall_knowledge.',
+    inputSchema: {
+        type: 'object',
+        properties: {},
+    }
+};
+
 const SEARCH_KNOWLEDGE_TOOL = {
     name: 'search_knowledge',
     description: 'Search knowledge base with AI processing. Returns synthesized answers and references.',
     inputSchema: {
         type: 'object',
         properties: {
+            kb_id: {
+                type: 'string',
+                description: 'The ID of the knowledge base to search. Optional if only one KB is configured or to use the default.'
+            },
             question: {
                 type: 'string',
                 description: 'The question to ask'
             },
-            topic_ids: {
-                type: 'array',
-                items: { type: 'string' },
-                description: 'List of knowledge base IDs. If not provided, uses the default configured topic ID.'
-            },
             deep_seek: {
                 type: 'boolean',
                 description: 'Enable deep thinking mode',
-                default: true
+                default: false
             },
             history: {
                 type: 'array',
@@ -79,13 +78,13 @@ const RECALL_KNOWLEDGE_TOOL = {
     inputSchema: {
         type: 'object',
         properties: {
-            question: {
+            kb_id: {
                 type: 'string',
-                description: 'The question or query'
+                description: 'The ID of the knowledge base to search. Optional if only one KB is configured or to use the default.'
             },
-            topic_id: {
+            question: {
                 type: 'string',
-                description: 'Knowledge base ID. If not provided, uses the default configured topic ID.'
+                description: 'The question or query'
             },
             top_k: {
                 type: 'number',
@@ -104,7 +103,7 @@ const RECALL_KNOWLEDGE_TOOL = {
 
 server.setRequestHandler(ListToolsRequestSchema, async () => {
     return {
-        tools: [SEARCH_KNOWLEDGE_TOOL, RECALL_KNOWLEDGE_TOOL],
+        tools: [LIST_KNOWLEDGE_BASES_TOOL, SEARCH_KNOWLEDGE_TOOL, RECALL_KNOWLEDGE_TOOL],
     };
 });
 
@@ -113,24 +112,55 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
         const { name, arguments: args } = request.params;
 
         switch (name) {
+            case 'list_knowledge_bases': {
+                const kbs = getKnowledgeBases();
+                // Return simplified list (hide sensitive config)
+                const publicKbs = kbs.map(kb => ({
+                    id: kb.id,
+                    name: kb.name,
+                    description: kb.description
+                }));
+
+                return {
+                    content: [
+                        {
+                            type: 'text',
+                            text: JSON.stringify(publicKbs, null, 2),
+                        },
+                    ],
+                };
+            }
+
             case 'search_knowledge': {
-                const params = { ...args };
-
-                // Handle default topic_ids
-                if (!params.topic_ids || params.topic_ids.length === 0) {
-                    if (DEFAULT_TOPIC_ID) {
-                        params.topic_ids = [DEFAULT_TOPIC_ID];
-                    } else {
-                        throw new Error('topic_ids is required (no default configured)');
+                const { kb_id, ...params } = args;
+                let kb;
+
+                if (kb_id) {
+                    kb = getKnowledgeBaseById(kb_id);
+                    if (!kb) {
+                        throw new Error(`Knowledge base not found: ${kb_id}`);
+                    }
+                } else {
+                    // Default to the first one
+                    const kbs = getKnowledgeBases();
+                    if (kbs.length === 0) {
+                        throw new Error('No knowledge bases configured');
                     }
+                    kb = kbs[0];
                 }
 
-                // Ensure topic_ids is array
-                if (!Array.isArray(params.topic_ids)) {
-                    throw new Error('topic_ids must be an array');
+                // Initialize client for this specific KB
+                const client = new ApiClient({
+                    apiKey: kb.config.api_key,
+                    baseUrl: kb.config.api_endpoint
+                });
+
+                // Inject topic_id from config (API requires array format but only supports 1)
+                if (kb.config.topic_id) {
+                    params.topic_ids = [kb.config.topic_id];
                 }
 
-                const response = await apiClient.searchKnowledge(params);
+                const response = await client.searchKnowledge(params);
 
                 return {
                     content: [
@@ -143,18 +173,34 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
             }
 
             case 'recall_knowledge': {
-                const params = { ...args };
-
-                // Handle default topic_id
-                if (!params.topic_id) {
-                    if (DEFAULT_TOPIC_ID) {
-                        params.topic_id = DEFAULT_TOPIC_ID;
-                    } else {
-                        throw new Error('topic_id is required (no default configured)');
+                const { kb_id, ...params } = args;
+                let kb;
+
+                if (kb_id) {
+                    kb = getKnowledgeBaseById(kb_id);
+                    if (!kb) {
+                        throw new Error(`Knowledge base not found: ${kb_id}`);
                     }
+                } else {
+                    // Default to the first one
+                    const kbs = getKnowledgeBases();
+                    if (kbs.length === 0) {
+                        throw new Error('No knowledge bases configured');
+                    }
+                    kb = kbs[0];
+                }
+
+                const client = new ApiClient({
+                    apiKey: kb.config.api_key,
+                    baseUrl: kb.config.api_endpoint
+                });
+
+                // Inject topic_id from config
+                if (kb.config.topic_id) {
+                    params.topic_id = kb.config.topic_id;
                 }
 
-                const response = await apiClient.recallKnowledge(params);
+                const response = await client.recallKnowledge(params);
                 return {
                     content: [
                         {
@@ -183,6 +229,7 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
 });
 
 async function runServer() {
+    await loadKnowledgeBases();
     const transport = new StdioServerTransport();
     await server.connect(transport);
     logger.info('Get Notes MCP Server running on stdio');

package/knowledge_bases.json

@@ -0,0 +1,25 @@
+[
+    {
+        "id": "demo_group",
+        "name": "sam的知识库",
+        "description": "使用同一个 API Key 的示例知识库（方案3：分组配置）",
+        "config": {
+            "api_key": "T0TR1HWqw+tL00gVc5PPoXDnGIUIuPw7QCBeb1dMC6afCojR9WpHyuZicEFJjkMg3oZYS0HB/S3vqOj7+0e5FGoVtYuv938zkR8=",
+            "api_endpoint": "https://open-api.biji.com/getnote/openapi",
+            "topics": [
+                {
+                    "id": "kb_pancreatic_trials",
+                    "name": "肿瘤临床试验知识库",
+                    "description": "有关肿瘤临床试验的知识库",
+                    "topic_id": "RYk1kmRJ"
+                },
+                {
+                    "id": "kb_patient_experience",
+                    "name": "小胰宝病友经验信息",
+                    "description": "小胰宝病友经验信息，包括各种指南和治疗经验",
+                    "topic_id": "mnyV9OjY"
+                }
+            ]
+        }
+    }
+]

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "get_notebook_mcp_server",
-  "version": "1.0.3",
-  "description": "MCP server for Get Notes API integration",
+  "version": "1.1.0",
+  "description": "MCP server for Get Notes API integration with multi-knowledge base support",
   "type": "module",
   "main": "index.js",
   "bin": {
@@ -9,7 +9,8 @@
   },
   "files": [
     "index.js",
-    "src"
+    "src",
+    "knowledge_bases.json"
   ],
   "scripts": {
     "test": "node --experimental-vm-modules node_modules/jest/bin/jest.js"

package/src/config/knowledgeBases.js

@@ -0,0 +1,130 @@
+import fs from 'fs/promises';
+import path from 'path';
+import { fileURLToPath } from 'url';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+const CONFIG_PATH = path.resolve(__dirname, '../../knowledge_bases.json');
+
+let knowledgeBases = [];
+
+export async function loadKnowledgeBases() {
+    try {
+        const data = await fs.readFile(CONFIG_PATH, 'utf-8');
+        knowledgeBases = JSON.parse(data);
+    } catch (error) {
+        console.warn('Warning: Could not load knowledge_bases.json, using empty list or environment variables.', error.message);
+        knowledgeBases = [];
+    }
+
+    // Backward compatibility: Inject KB from environment variables if they exist
+    const envApiKey = process.env.GET_API_KEY;
+    const envTopicId = process.env.GET_NOTE_TOPIC_ID;
+
+    if (envApiKey) {
+        // Check if already exists to avoid duplicates
+        const defaultId = 'default_env_kb';
+        const exists = knowledgeBases.find(kb => kb.id === defaultId);
+
+        if (!exists) {
+            knowledgeBases.unshift({
+                id: defaultId,
+                name: "Default Knowledge Base (Env)",
+                description: "Configured via environment variables",
+                config: {
+                    api_endpoint: process.env.GET_API_ENDPOINT || 'https://open-api.biji.com/getnote/openapi',
+                    api_key: envApiKey,
+                    topic_id: envTopicId || ''
+                }
+            });
+        }
+    }
+
+    // Support for multiple KBs via environment variable (JSON string)
+    const envKbsJson = process.env.GET_KNOWLEDGE_BASES;
+    if (envKbsJson) {
+        try {
+            const envKbs = JSON.parse(envKbsJson);
+            if (Array.isArray(envKbs)) {
+                const DEFAULT_ENDPOINT = 'https://open-api.biji.com/getnote/openapi';
+                // Merge strategies: Append? Prepend? Overwrite?
+                // Let's append, but filter out duplicates by ID
+                for (const kb of envKbs) {
+                    if (!knowledgeBases.find(existing => existing.id === kb.id)) {
+                        // Apply defaults
+                        if (kb.config && !kb.config.api_endpoint) {
+                            kb.config.api_endpoint = DEFAULT_ENDPOINT;
+                        }
+                        knowledgeBases.push(kb);
+                    }
+                }
+            }
+        } catch (e) {
+            console.warn('Failed to parse GET_KNOWLEDGE_BASES environment variable:', e.message);
+        }
+    }
+
+    // Expand grouped configurations
+    knowledgeBases = expandKnowledgeBases(knowledgeBases);
+
+    // Normalize configs: ensure api_endpoint and topic_id are set
+    const DEFAULT_ENDPOINT = 'https://open-api.biji.com/getnote/openapi';
+    knowledgeBases.forEach(kb => {
+        if (kb.config) {
+            if (!kb.config.api_endpoint) {
+                kb.config.api_endpoint = DEFAULT_ENDPOINT;
+            }
+            // Ensure topic_id is set (no longer support topic_ids array)
+            if (!kb.config.topic_id) {
+                kb.config.topic_id = '';
+            }
+        }
+    });
+
+    return knowledgeBases;
+}
+
+/**
+ * Expand grouped knowledge base configurations into individual entries.
+ * A grouped config has a 'topics' array in its config, where each topic
+ * becomes an independent knowledge base entry.
+ * 
+ * @param {Array} rawConfig - Raw configuration array
+ * @returns {Array} Expanded configuration array
+ */
+function expandKnowledgeBases(rawConfig) {
+    const expanded = [];
+
+    for (const group of rawConfig) {
+        if (group.config && group.config.topics && Array.isArray(group.config.topics)) {
+            // Grouped configuration: expand into multiple independent KBs
+            for (const topic of group.config.topics) {
+                expanded.push({
+                    id: topic.id,
+                    name: topic.name,
+                    description: topic.description || '',
+                    config: {
+                        api_key: group.config.api_key,
+                        api_endpoint: group.config.api_endpoint || 'https://open-api.biji.com/getnote/openapi',
+                        topic_id: topic.topic_id
+                    }
+                });
+            }
+        } else {
+            // Regular configuration: keep as-is
+            expanded.push(group);
+        }
+    }
+
+    return expanded;
+}
+
+export function getKnowledgeBases() {
+    return knowledgeBases;
+}
+
+export function getKnowledgeBaseById(id) {
+    return knowledgeBases.find(kb => kb.id === id);
+}