新增系统管理功能，可在Web界面修改配置，无需手动修改配置文件，无需重新构建docker镜像

2025-10-26 13:38:17 +08:00 · 2025-10-26 13:38:17 +08:00 · 165372cfd6
parent c13f266557
commit 165372cfd6
6 changed files with 879 additions and 73 deletions
--- a/README.md
+++ b/README.md
@ -37,8 +37,9 @@

 - **AI 驱动的深度分析**：超越传统静态分析，理解代码意图，发现深层逻辑问题。
 - **多维度、全方位评估**：从**安全性**、**性能**、**可维护性**到**代码风格**，提供 360 度无死角的质量评估。
- **清晰、可行的修复建议**：独创 **What-Why-How** 模式，不仅告诉您“是什么”问题，还解释“为什么”，并提供“如何修复”的具体代码示例。
+- **清晰、可行的修复建议**：独创 **What-Why-How** 模式，不仅告诉您"是什么"问题，还解释"为什么"，并提供"如何修复"的具体代码示例。
 - **多平台LLM/本地LLM支持**: 已实现 10+ 主流平台API调用功能（Gemini、OpenAI、Claude、通义千问、DeepSeek、智谱AI、Kimi、文心一言、MiniMax、豆包、Ollama本地大模型），支持用户自由配置和切换。
+- **可视化运行时配置**：无需重新构建镜像，直接在浏览器中配置所有 LLM 参数和 API Keys，支持 API 中转站，配置保存在本地浏览器，安全便捷。
 - **现代化、高颜值的用户界面**：基于 React + TypeScript 构建，提供流畅、直观的操作体验。

 ## 🎬 项目演示
@ -72,19 +73,23 @@
 git clone https://github.com/lintsinghua/XCodeReviewer.git
 cd XCodeReviewer

-# 2. 配置环境变量
-cp .env.example .env
-# 编辑 .env 文件，至少配置以下内容：
-# VITE_LLM_PROVIDER=gemini
-# VITE_LLM_API_KEY=your_api_key_here
-
-# 3. 构建并启动
+# 2. 构建并启动（无需预先配置）
 docker-compose up -d

-# 4. 访问应用
+# 3. 访问应用
 # 浏览器打开 http://localhost:5174
 ```

+**✨ 运行时配置**
+
+Docker 部署后，您可以直接在浏览器中配置所有设置，无需重新构建镜像：
+
+1. 访问 `http://localhost:5174/admin`（系统管理页面）
+2. 在"系统配置"标签页中配置 LLM API Keys 和其他参数
+3. 点击保存并刷新页面即可使用
+
+> 📖 **详细配置说明请参考**：[系统配置使用指南](#系统配置首次使用必看)
+
 ### 💻 本地开发部署

 适合需要开发或自定义修改的场景。
@ -164,7 +169,16 @@ VITE_LLM_GAP_MS=1000                         # 增加请求间隔
 <details>
 <summary><b>如何快速切换 LLM 平台？</b></summary>

-只需修改 `.env` 中的 `VITE_LLM_PROVIDER` 和对应的 API Key：
+**方式一：浏览器配置（推荐）**
+
+1. 访问 `http://localhost:5174/admin` 系统管理页面
+2. 在"系统配置"标签页选择不同的 LLM 提供商
+3. 填入对应的 API Key
+4. 保存并刷新页面
+
+**方式二：环境变量配置**
+
+修改 `.env` 中的配置：

 ```env
 # 切换到 OpenAI
@ -235,11 +249,29 @@ VITE_BAIDU_API_KEY=your_api_key:your_secret_key
 获取地址：https://console.bce.baidu.com/qianfan/
 </details>

+<details>
+<summary><b>如何使用 API 中转站？</b></summary>
+
+许多用户使用 API 中转服务来访问 LLM（更稳定、更便宜）。配置方法：
+
+1. 访问系统管理页面（`/admin`）
+2. 在"系统配置"标签页中：
+   - 选择 LLM 提供商（如 OpenAI）
+   - **API 基础 URL**: 填入中转站地址（如 `https://your-proxy.com/v1`）
+   - **API Key**: 填入中转站提供的密钥（而非官方密钥）
+3. 保存并刷新页面
+
+**注意**：
+- 中转站 URL 通常以 `/v1` 结尾（OpenAI 兼容格式）
+- 使用中转站的 API Key，不是官方的
+- 确认中转站支持你选择的 AI 模型
+</details>
+
 <details>
 <summary><b>如何备份本地数据库？</b></summary>

 本地数据存储在浏览器 IndexedDB 中：
- 在应用的"数据库管理"页面导出为 JSON 文件
+- 在应用的"系统管理"页面导出为 JSON 文件
 - 通过导入 JSON 文件恢复数据
 - 注意：清除浏览器数据会删除所有本地数据
 </details>
@ -356,18 +388,28 @@ VITE_OPENAI_API_KEY=your_openai_key
 </details>

 <details>
-<summary><b>💾 本地数据库管理</b></summary>
+<summary><b>⚙️ 系统管理</b></summary>

- **三种数据库模式**：
-  - 🏠 **本地模式**：使用浏览器 IndexedDB，数据完全本地化，隐私安全
-  - ☁️ **云端模式**：使用 Supabase，支持多设备同步
-  - 🎭 **演示模式**：无需配置，快速体验功能
- **数据管理功能**：
+访问 `/admin` 页面，提供完整的系统配置和数据管理功能：
+
+- **🔧 可视化配置管理**（运行时配置）：
+  - 🎯 **LLM 配置**：在浏览器中直接配置 API Keys、模型、超时等参数
+  - 🔑 **平台密钥**：管理 10+ LLM 平台的 API Keys，支持快速切换
+  - ⚡ **分析参数**：调整并发数、间隔时间、最大文件数等
+  - 🌐 **API 中转站支持**：轻松配置第三方 API 代理服务
+  - 💾 **配置优先级**：运行时配置 > 构建时配置，无需重新构建镜像
+  
+- **💾 数据库管理**：
+  - 🏠 **三种模式**：本地 IndexedDB / Supabase 云端 / 演示模式
  - 📤 **导出备份**：将数据导出为 JSON 文件
  - 📥 **导入恢复**：从备份文件恢复数据
  - 🗑️ **清空数据**：一键清理所有本地数据
  - 📊 **存储监控**：实时查看存储空间使用情况
- **智能统计**：项目、任务、问题的完整统计和可视化展示
+  
+- **📈 数据概览**：
+  - 项目、任务、问题的完整统计
+  - 可视化图表展示质量趋势
+  - 存储使用情况分析
 </details>

 ## 🛠️ 技术栈
@ -397,6 +439,7 @@ XCodeReviewer/
 │   ├── components/         # React 组件
 │   │   ├── layout/         # 布局组件 (Header, Footer, PageMeta)
 │   │   ├── ui/             # UI 组件库 (基于 Radix UI)
+│   │   ├── system/         # 系统配置组件
 │   │   ├── database/       # 数据库管理组件
 │   │   └── debug/          # 调试组件
 │   ├── pages/              # 页面组件
@ -404,7 +447,7 @@ XCodeReviewer/
 │   │   ├── Projects.tsx    # 项目管理
 │   │   ├── InstantAnalysis.tsx # 即时分析
 │   │   ├── AuditTasks.tsx  # 审计任务
-│   │   └── AdminDashboard.tsx # 数据库管理
+│   │   └── AdminDashboard.tsx # 系统管理
 │   ├── features/           # 功能模块
 │   │   ├── analysis/       # 分析相关服务
 │   │   │   └── services/   # AI 代码分析引擎
@ -432,6 +475,26 @@ XCodeReviewer/

 ## 🎯 使用指南

+### 系统配置（首次使用必看）
+
+访问 `/admin` 系统管理页面，在"系统配置"标签页中配置：
+
+#### 1. **配置 LLM 提供商**
+- 选择您要使用的 LLM 平台（Gemini、OpenAI、Claude 等）
+- 填入 API Key（支持通用 Key 或平台专用 Key）
+- 可选：配置模型名称、API 基础 URL（用于中转站）
+
+#### 2. **配置 API 中转站**（如果使用）
+- 在"API 基础 URL"中填入中转站地址（如 `https://your-proxy.com/v1`）
+- 填入中转站提供的 API Key
+- 保存配置
+
+#### 3. **调整分析参数**（可选）
+- 最大分析文件数、并发请求数、请求间隔
+- 输出语言（中文/英文）
+
+**配置完成后点击"保存所有更改"并刷新页面即可使用。**
+
 ### 即时代码分析
 1. 访问 `/instant-analysis` 页面
 2. 选择编程语言（支持 10+ 种语言）
@ -579,8 +642,9 @@ pnpm lint

 - **✅ 多平台LLM支持**: 已实现 10+ 主流平台API调用功能（Gemini、OpenAI、Claude、通义千问、DeepSeek、智谱AI、Kimi、文心一言、MiniMax、豆包、Ollama本地大模型），支持用户自由配置和切换
 - **✅ 本地模型支持**: 已加入对 Ollama 本地大模型的调用功能，满足数据隐私需求
- **Multi-Agent Collaboration**: 考虑引入多智能体协作架构，会实现`Agent+人工对话`反馈的功能，包括多轮对话流程展示，人工对话中断干涉等，以获得更清晰、透明、监督性的审计过程，提升审计质量
+- **✅ 可视化配置管理**: 已实现运行时配置系统，支持在浏览器中直接配置所有 LLM 参数、API Keys，支持 API 中转站，无需重新构建镜像
 - **✅ 专业报告文件生成**: 根据不同的需求生成相关格式的专业审计报告文件，支持文件报告格式定制等
+- **Multi-Agent Collaboration**: 考虑引入多智能体协作架构，会实现`Agent+人工对话`反馈的功能，包括多轮对话流程展示，人工对话中断干涉等，以获得更清晰、透明、监督性的审计过程，提升审计质量
 - **审计标准自定义**: 不同团队有自己的编码规范，不同项目有特定的安全要求，也正是我们这个项目想后续做的东西。当前的版本还属于一个“半黑盒模式”，项目通过 Prompt 工程来引导分析方向和定义审计标准，实际分析效果由强大的预训练AI 模型内置知识决定。后续将结合强化学习、监督学习微调等方法，开发以支持自定义规则配置，通过YAML或者JSON定义团队特定规则，提供常见框架的最佳实践模板等等，以获得更加符合需求和标准的审计结果
 ---

--- a/README_EN.md
+++ b/README_EN.md
@ -39,6 +39,7 @@ In the fast-paced world of software development, ensuring code quality is crucia
 - **🎯 Multi-dimensional, Comprehensive Assessment**: From **security**, **performance**, **maintainability** to **code style**, providing 360-degree quality evaluation.
 - **💡 Clear, Actionable Fix Suggestions**: Innovative **What-Why-How** approach that not only tells you "what" the problem is, but also explains "why" and provides "how to fix" with specific code examples.
 - **✅ Multi-Platform LLM/Local Model Support**: Implemented API calling functionality for 10+ mainstream platforms (Gemini, OpenAI, Claude, Qwen, DeepSeek, Zhipu AI, Kimi, ERNIE, MiniMax, Doubao, Ollama Local Models), with support for free configuration and switching
+- **⚙️ Visual Runtime Configuration**: Configure all LLM parameters and API Keys directly in the browser without rebuilding images. Supports API relay services, with configurations saved locally in the browser for security and convenience.
 - **✨ Modern, Beautiful User Interface**: Built with React + TypeScript, providing a smooth and intuitive user experience.

 ## 🎬 Project Demo
@ -72,19 +73,29 @@ One-click deployment with Docker, no Node.js environment required:
 git clone https://github.com/lintsinghua/XCodeReviewer.git
 cd XCodeReviewer

-# 2. Configure environment variables
-cp .env.example .env
-# Edit .env file, configure at least:
-# VITE_LLM_PROVIDER=gemini
-# VITE_LLM_API_KEY=your_api_key_here
-
-# 3. Build and start
+# 2. Build and start (no pre-configuration needed)
 docker-compose up -d

-# 4. Access the application
+# 3. Access the application
 # Open http://localhost:5174 in your browser
 ```

+**✨ New Feature: Runtime Configuration**
+
+After Docker deployment, you can configure all settings directly in the browser without rebuilding the image:
+
+1. Visit `http://localhost:5174/admin` (System Management page)
+2. Configure LLM API Keys and other parameters in the "System Configuration" tab
+3. Click save and refresh the page to use
+
+Benefits:
+- ✅ No need to hardcode API Keys in Docker images (more secure)
+- ✅ Modify configuration anytime without rebuilding
+- ✅ Support for API relay services
+- ✅ Quickly switch between different LLM platforms
+
+> 📖 **For detailed configuration instructions, see**: [System Configuration Guide](#system-configuration-first-time-setup)
+
 ### 💻 Local Development Deployment

 Suitable for development or custom modifications.
@ -164,7 +175,16 @@ VITE_LLM_GAP_MS=1000                         # Increase request interval
 <details>
 <summary><b>How to quickly switch LLM platforms?</b></summary>

-Simply modify `VITE_LLM_PROVIDER` and the corresponding API Key in `.env`:
+**Method 1: Browser Configuration (Recommended)**
+
+1. Visit `http://localhost:5174/admin` System Management page
+2. Select different LLM provider in the "System Configuration" tab
+3. Enter the corresponding API Key
+4. Save and refresh the page
+
+**Method 2: Environment Variable Configuration**
+
+Modify configuration in `.env`:

 ```env
 # Switch to OpenAI
@ -235,11 +255,29 @@ VITE_BAIDU_API_KEY=your_api_key:your_secret_key
 Get from: https://console.bce.baidu.com/qianfan/
 </details>

+<details>
+<summary><b>How to use API relay services?</b></summary>
+
+Many users use API relay services to access LLMs (more stable and cheaper). Configuration method:
+
+1. Visit System Management page (`/admin`)
+2. In the "System Configuration" tab:
+   - Select LLM provider (e.g., OpenAI)
+   - **API Base URL**: Enter relay service address (e.g., `https://your-proxy.com/v1`)
+   - **API Key**: Enter the key provided by the relay service (not the official key)
+3. Save and refresh the page
+
+**Notes**:
+- Relay service URLs usually end with `/v1` (OpenAI-compatible format)
+- Use the relay service's API Key, not the official one
+- Confirm that the relay service supports your chosen AI model
+</details>
+
 <details>
 <summary><b>How to backup local database?</b></summary>

 Local data is stored in browser IndexedDB:
- Export as JSON file from "Database Management" page
+- Export as JSON file from "System Management" page
 - Import JSON file to restore data
 - Note: Clearing browser data will delete all local data
 </details>
@ -357,18 +395,28 @@ For cloud data sync:
 </details>

 <details>
-<summary><b>💾 Local Database Management</b></summary>
+<summary><b>⚙️ System Management</b></summary>

- **Three Database Modes**:
-  - 🏠 **Local Mode**: Uses browser IndexedDB, data is completely localized, privacy-secure
-  - ☁️ **Cloud Mode**: Uses Supabase, supports multi-device synchronization
-  - 🎭 **Demo Mode**: No configuration needed, quick feature preview
- **Data Management Features**:
+Visit `/admin` page for complete system configuration and data management features:
+
+- **🔧 Visual Configuration Management** (Runtime Configuration):
+  - 🎯 **LLM Configuration**: Configure API Keys, models, timeout, and other parameters directly in the browser
+  - 🔑 **Platform Keys**: Manage API Keys for 10+ LLM platforms with quick switching support
+  - ⚡ **Analysis Parameters**: Adjust concurrency, interval time, max files, etc.
+  - 🌐 **API Relay Support**: Easily configure third-party API relay services
+  - 💾 **Configuration Priority**: Runtime config > Build-time config, no need to rebuild images
+  
+- **💾 Database Management**:
+  - 🏠 **Three Modes**: Local IndexedDB / Supabase Cloud / Demo Mode
  - 📤 **Export Backup**: Export data as JSON files
  - 📥 **Import Recovery**: Restore data from backup files
  - 🗑️ **Clear Data**: One-click cleanup of all local data
  - 📊 **Storage Monitoring**: Real-time view of storage space usage
- **Smart Statistics**: Complete statistics and visualization of projects, tasks, and issues
+  
+- **📈 Data Overview**:
+  - Complete statistics for projects, tasks, and issues
+  - Visual charts showing quality trends
+  - Storage usage analysis
 </details>

 ## 🛠️ Tech Stack
@ -399,6 +447,7 @@ XCodeReviewer/
 │   ├── components/         # React components
 │   │   ├── layout/         # Layout components (Header, Footer, PageMeta)
 │   │   ├── ui/             # UI component library (based on Radix UI)
+│   │   ├── system/         # System configuration components
 │   │   ├── database/       # Database management components
 │   │   └── debug/          # Debug components
 │   ├── pages/              # Page components
@ -406,7 +455,7 @@ XCodeReviewer/
 │   │   ├── Projects.tsx    # Project management
 │   │   ├── InstantAnalysis.tsx # Instant analysis
 │   │   ├── AuditTasks.tsx  # Audit tasks
-│   │   └── AdminDashboard.tsx # Database management
+│   │   └── AdminDashboard.tsx # System management
 │   ├── features/           # Feature modules
 │   │   ├── analysis/       # Analysis related services
 │   │   │   └── services/   # AI code analysis engine
@ -434,6 +483,26 @@ XCodeReviewer/

 ## 🎯 Usage Guide

+### System Configuration (First-Time Setup)
+
+Visit `/admin` System Management page and configure in the "System Configuration" tab:
+
+#### 1. **Configure LLM Provider**
+- Select the LLM platform you want to use (Gemini, OpenAI, Claude, etc.)
+- Enter API Key (supports universal Key or platform-specific Key)
+- Optional: Configure model name, API base URL (for relay services)
+
+#### 2. **Configure API Relay Service** (if using)
+- Enter relay service address in "API Base URL" (e.g., `https://your-proxy.com/v1`)
+- Enter the API Key provided by the relay service
+- Save configuration
+
+#### 3. **Adjust Analysis Parameters** (optional)
+- Max analyze files, concurrent requests, request interval
+- Output language (Chinese/English)
+
+**After configuration, click "Save All Changes" and refresh the page to use.**
+
 ### Instant Code Analysis
 1. Visit the `/instant-analysis` page
 2. Select programming language (supports 10+ languages)
@ -580,10 +649,10 @@ Currently, XCodeReviewer is in rapid prototype validation stage. Based on projec

 - ✅ **Multi-Platform LLM Support**: Implemented API integration for 10+ mainstream platforms (Gemini, OpenAI, Claude, Qwen, DeepSeek, Zhipu AI, Kimi, ERNIE, MiniMax, Doubao, Ollama), with flexible configuration and switching
 - ✅ **Local Model Support**: Added Ollama local model integration to meet data privacy requirements
- ✅ **Local Database Support**: Implemented IndexedDB-based local database for fully localized data storage and privacy protection
- **Multi-Agent Collaboration**: Introduce multi-agent architecture with `Agent + Human Dialogue` feedback, including multi-round dialogue visualization and human intervention for clearer, transparent, and supervised audit processes
+- ✅ **Visual Configuration Management**: Implemented runtime configuration system supporting browser-based configuration of all LLM parameters and API Keys, API relay service support, no need to rebuild images
 - ✅ **Professional Report Generation**: Generate professional audit reports in various formats based on different needs, with customizable templates and format configurations
- **Custom Audit Standards**: Support custom audit rule configuration via YAML/JSON, provide best practice templates for common frameworks, and leverage reinforcement learning and supervised fi
+- **Multi-Agent Collaboration**: Introduce multi-agent architecture with `Agent + Human Dialogue` feedback, including multi-round dialogue visualization and human intervention for clearer, transparent, and supervised audit processes
+- **Custom Audit Standards**: Support custom audit rule configuration via YAML/JSON, provide best practice templates for common frameworks, and leverage reinforcement learning and supervised fine-tuning for more targeted and standards-compliant audit results

 ---

--- a/src/app/routes.tsx
+++ b/src/app/routes.tsx
@ -52,7 +52,7 @@ const routes: RouteConfig[] = [
    visible: false,
  },
  {
-    name: "数据库管理",
+    name: "系统管理",
    path: "/admin",
    element: <AdminDashboard />,
    visible: true,
--- a/src/components/system/SystemConfig.tsx
+++ b/src/components/system/SystemConfig.tsx
@ -0,0 +1,652 @@
+import { useState, useEffect } from "react";
+import { Card, CardContent, CardHeader, CardTitle, CardDescription } from "@/components/ui/card";
+import { Button } from "@/components/ui/button";
+import { Input } from "@/components/ui/input";
+import { Label } from "@/components/ui/label";
+import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from "@/components/ui/select";
+import { Tabs, TabsContent, TabsList, TabsTrigger } from "@/components/ui/tabs";
+import { Alert, AlertDescription } from "@/components/ui/alert";
+import { Badge } from "@/components/ui/badge";
+import {
+  Settings,
+  Save,
+  RotateCcw,
+  Eye,
+  EyeOff,
+  CheckCircle2,
+  AlertCircle,
+  Info,
+  Key,
+  Zap,
+  Globe,
+  Database
+} from "lucide-react";
+import { toast } from "sonner";
+
+// LLM 提供商配置
+const LLM_PROVIDERS = [
+  { value: 'gemini', label: 'Google Gemini', icon: '🔵', category: 'international' },
+  { value: 'openai', label: 'OpenAI GPT', icon: '🟢', category: 'international' },
+  { value: 'claude', label: 'Anthropic Claude', icon: '🟣', category: 'international' },
+  { value: 'deepseek', label: 'DeepSeek', icon: '🔷', category: 'international' },
+  { value: 'qwen', label: '阿里云通义千问', icon: '🟠', category: 'domestic' },
+  { value: 'zhipu', label: '智谱AI (GLM)', icon: '🔴', category: 'domestic' },
+  { value: 'moonshot', label: 'Moonshot (Kimi)', icon: '🌙', category: 'domestic' },
+  { value: 'baidu', label: '百度文心一言', icon: '🔵', category: 'domestic' },
+  { value: 'minimax', label: 'MiniMax', icon: '⚡', category: 'domestic' },
+  { value: 'doubao', label: '字节豆包', icon: '🎯', category: 'domestic' },
+  { value: 'ollama', label: 'Ollama 本地模型', icon: '🖥️', category: 'local' },
+];
+
+// 默认模型配置
+const DEFAULT_MODELS = {
+  gemini: 'gemini-2.5-flash',
+  openai: 'gpt-4o-mini',
+  claude: 'claude-3-5-sonnet-20241022',
+  qwen: 'qwen-turbo',
+  deepseek: 'deepseek-chat',
+  zhipu: 'glm-4-flash',
+  moonshot: 'moonshot-v1-8k',
+  baidu: 'ERNIE-3.5-8K',
+  minimax: 'abab6.5-chat',
+  doubao: 'doubao-pro-32k',
+  ollama: 'llama3',
+};
+
+interface SystemConfigData {
+  // LLM 配置
+  llmProvider: string;
+  llmApiKey: string;
+  llmModel: string;
+  llmBaseUrl: string;
+  llmTimeout: number;
+  llmTemperature: number;
+  llmMaxTokens: number;
+  
+  // 平台专用配置
+  geminiApiKey: string;
+  openaiApiKey: string;
+  claudeApiKey: string;
+  qwenApiKey: string;
+  deepseekApiKey: string;
+  zhipuApiKey: string;
+  moonshotApiKey: string;
+  baiduApiKey: string;
+  minimaxApiKey: string;
+  doubaoApiKey: string;
+  ollamaBaseUrl: string;
+  
+  // GitHub 配置
+  githubToken: string;
+  
+  // 分析配置
+  maxAnalyzeFiles: number;
+  llmConcurrency: number;
+  llmGapMs: number;
+  outputLanguage: string;
+}
+
+const STORAGE_KEY = 'xcodereviewer_runtime_config';
+
+export function SystemConfig() {
+  const [config, setConfig] = useState<SystemConfigData>({
+    llmProvider: 'gemini',
+    llmApiKey: '',
+    llmModel: '',
+    llmBaseUrl: '',
+    llmTimeout: 150000,
+    llmTemperature: 0.2,
+    llmMaxTokens: 4096,
+    geminiApiKey: '',
+    openaiApiKey: '',
+    claudeApiKey: '',
+    qwenApiKey: '',
+    deepseekApiKey: '',
+    zhipuApiKey: '',
+    moonshotApiKey: '',
+    baiduApiKey: '',
+    minimaxApiKey: '',
+    doubaoApiKey: '',
+    ollamaBaseUrl: 'http://localhost:11434/v1',
+    githubToken: '',
+    maxAnalyzeFiles: 40,
+    llmConcurrency: 2,
+    llmGapMs: 500,
+    outputLanguage: 'zh-CN',
+  });
+
+  const [showApiKeys, setShowApiKeys] = useState<Record<string, boolean>>({});
+  const [hasChanges, setHasChanges] = useState(false);
+  const [configSource, setConfigSource] = useState<'runtime' | 'build'>('build');
+
+  // 加载配置
+  useEffect(() => {
+    loadConfig();
+  }, []);
+
+  const loadConfig = () => {
+    try {
+      // 尝试从 localStorage 加载运行时配置
+      const savedConfig = localStorage.getItem(STORAGE_KEY);
+      
+      if (savedConfig) {
+        const parsedConfig = JSON.parse(savedConfig);
+        setConfig(parsedConfig);
+        setConfigSource('runtime');
+        toast.success("已加载运行时配置");
+      } else {
+        // 使用构建时配置
+        loadFromEnv();
+        setConfigSource('build');
+      }
+    } catch (error) {
+      console.error('Failed to load config:', error);
+      loadFromEnv();
+    }
+  };
+
+  const loadFromEnv = () => {
+    // 从环境变量加载（构建时配置）
+    const envConfig: SystemConfigData = {
+      llmProvider: import.meta.env.VITE_LLM_PROVIDER || 'gemini',
+      llmApiKey: import.meta.env.VITE_LLM_API_KEY || '',
+      llmModel: import.meta.env.VITE_LLM_MODEL || '',
+      llmBaseUrl: import.meta.env.VITE_LLM_BASE_URL || '',
+      llmTimeout: Number(import.meta.env.VITE_LLM_TIMEOUT) || 150000,
+      llmTemperature: Number(import.meta.env.VITE_LLM_TEMPERATURE) || 0.2,
+      llmMaxTokens: Number(import.meta.env.VITE_LLM_MAX_TOKENS) || 4096,
+      geminiApiKey: import.meta.env.VITE_GEMINI_API_KEY || '',
+      openaiApiKey: import.meta.env.VITE_OPENAI_API_KEY || '',
+      claudeApiKey: import.meta.env.VITE_CLAUDE_API_KEY || '',
+      qwenApiKey: import.meta.env.VITE_QWEN_API_KEY || '',
+      deepseekApiKey: import.meta.env.VITE_DEEPSEEK_API_KEY || '',
+      zhipuApiKey: import.meta.env.VITE_ZHIPU_API_KEY || '',
+      moonshotApiKey: import.meta.env.VITE_MOONSHOT_API_KEY || '',
+      baiduApiKey: import.meta.env.VITE_BAIDU_API_KEY || '',
+      minimaxApiKey: import.meta.env.VITE_MINIMAX_API_KEY || '',
+      doubaoApiKey: import.meta.env.VITE_DOUBAO_API_KEY || '',
+      ollamaBaseUrl: import.meta.env.VITE_OLLAMA_BASE_URL || 'http://localhost:11434/v1',
+      githubToken: import.meta.env.VITE_GITHUB_TOKEN || '',
+      maxAnalyzeFiles: Number(import.meta.env.VITE_MAX_ANALYZE_FILES) || 40,
+      llmConcurrency: Number(import.meta.env.VITE_LLM_CONCURRENCY) || 2,
+      llmGapMs: Number(import.meta.env.VITE_LLM_GAP_MS) || 500,
+      outputLanguage: import.meta.env.VITE_OUTPUT_LANGUAGE || 'zh-CN',
+    };
+    setConfig(envConfig);
+  };
+
+  const saveConfig = () => {
+    try {
+      localStorage.setItem(STORAGE_KEY, JSON.stringify(config));
+      setHasChanges(false);
+      setConfigSource('runtime');
+      toast.success("配置已保存！刷新页面后生效");
+      
+      // 提示用户刷新页面
+      setTimeout(() => {
+        if (window.confirm("配置已保存。是否立即刷新页面使配置生效？")) {
+          window.location.reload();
+        }
+      }, 1000);
+    } catch (error) {
+      console.error('Failed to save config:', error);
+      toast.error("保存配置失败");
+    }
+  };
+
+  const resetConfig = () => {
+    if (window.confirm("确定要重置为构建时配置吗？这将清除所有运行时配置。")) {
+      localStorage.removeItem(STORAGE_KEY);
+      loadFromEnv();
+      setHasChanges(false);
+      setConfigSource('build');
+      toast.success("已重置为构建时配置");
+    }
+  };
+
+  const updateConfig = (key: keyof SystemConfigData, value: any) => {
+    setConfig(prev => ({ ...prev, [key]: value }));
+    setHasChanges(true);
+  };
+
+  const toggleShowApiKey = (field: string) => {
+    setShowApiKeys(prev => ({ ...prev, [field]: !prev[field] }));
+  };
+
+  const getCurrentApiKey = () => {
+    const provider = config.llmProvider.toLowerCase();
+    const keyMap: Record<string, string> = {
+      gemini: config.geminiApiKey,
+      openai: config.openaiApiKey,
+      claude: config.claudeApiKey,
+      qwen: config.qwenApiKey,
+      deepseek: config.deepseekApiKey,
+      zhipu: config.zhipuApiKey,
+      moonshot: config.moonshotApiKey,
+      baidu: config.baiduApiKey,
+      minimax: config.minimaxApiKey,
+      doubao: config.doubaoApiKey,
+      ollama: 'ollama',
+    };
+    
+    return config.llmApiKey || keyMap[provider] || '';
+  };
+
+  const isConfigured = getCurrentApiKey() !== '';
+
+  return (
+    <div className="space-y-6">
+      {/* 配置状态提示 */}
+      <Alert>
+        <Info className="h-4 w-4" />
+        <AlertDescription className="flex items-center justify-between">
+          <div>
+            <strong>当前配置来源：</strong>
+            {configSource === 'runtime' ? (
+              <Badge variant="default" className="ml-2">运行时配置</Badge>
+            ) : (
+              <Badge variant="outline" className="ml-2">构建时配置</Badge>
+            )}
+            <span className="ml-4 text-sm">
+              {isConfigured ? (
+                <span className="text-green-600 flex items-center gap-1">
+                  <CheckCircle2 className="h-3 w-3" /> LLM 已配置
+                </span>
+              ) : (
+                <span className="text-orange-600 flex items-center gap-1">
+                  <AlertCircle className="h-3 w-3" /> 未配置 LLM
+                </span>
+              )}
+            </span>
+          </div>
+          <div className="flex gap-2">
+            {hasChanges && (
+              <Button onClick={saveConfig} size="sm">
+                <Save className="w-4 h-4 mr-2" />
+                保存配置
+              </Button>
+            )}
+            {configSource === 'runtime' && (
+              <Button onClick={resetConfig} variant="outline" size="sm">
+                <RotateCcw className="w-4 h-4 mr-2" />
+                重置
+              </Button>
+            )}
+          </div>
+        </AlertDescription>
+      </Alert>
+
+      <Tabs defaultValue="llm" className="w-full">
+        <TabsList className="grid w-full grid-cols-4">
+          <TabsTrigger value="llm">
+            <Zap className="w-4 h-4 mr-2" />
+            LLM 配置
+          </TabsTrigger>
+          <TabsTrigger value="platforms">
+            <Key className="w-4 h-4 mr-2" />
+            平台密钥
+          </TabsTrigger>
+          <TabsTrigger value="analysis">
+            <Settings className="w-4 h-4 mr-2" />
+            分析参数
+          </TabsTrigger>
+          <TabsTrigger value="other">
+            <Globe className="w-4 h-4 mr-2" />
+            其他配置
+          </TabsTrigger>
+        </TabsList>
+
+        {/* LLM 基础配置 */}
+        <TabsContent value="llm" className="space-y-6">
+          <Card>
+            <CardHeader>
+              <CardTitle>LLM 提供商配置</CardTitle>
+              <CardDescription>选择和配置大语言模型服务</CardDescription>
+            </CardHeader>
+            <CardContent className="space-y-4">
+              <div className="space-y-2">
+                <Label>当前使用的 LLM 提供商</Label>
+                <Select
+                  value={config.llmProvider}
+                  onValueChange={(value) => updateConfig('llmProvider', value)}
+                >
+                  <SelectTrigger>
+                    <SelectValue />
+                  </SelectTrigger>
+                  <SelectContent>
+                    <div className="px-2 py-1.5 text-sm font-semibold text-muted-foreground">国际平台</div>
+                    {LLM_PROVIDERS.filter(p => p.category === 'international').map(provider => (
+                      <SelectItem key={provider.value} value={provider.value}>
+                        {provider.icon} {provider.label}
+                      </SelectItem>
+                    ))}
+                    <div className="px-2 py-1.5 text-sm font-semibold text-muted-foreground mt-2">国内平台</div>
+                    {LLM_PROVIDERS.filter(p => p.category === 'domestic').map(provider => (
+                      <SelectItem key={provider.value} value={provider.value}>
+                        {provider.icon} {provider.label}
+                      </SelectItem>
+                    ))}
+                    <div className="px-2 py-1.5 text-sm font-semibold text-muted-foreground mt-2">本地部署</div>
+                    {LLM_PROVIDERS.filter(p => p.category === 'local').map(provider => (
+                      <SelectItem key={provider.value} value={provider.value}>
+                        {provider.icon} {provider.label}
+                      </SelectItem>
+                    ))}
+                  </SelectContent>
+                </Select>
+              </div>
+
+              <div className="space-y-2">
+                <Label>通用 API Key（可选）</Label>
+                <div className="flex gap-2">
+                  <Input
+                    type={showApiKeys['llm'] ? 'text' : 'password'}
+                    value={config.llmApiKey}
+                    onChange={(e) => updateConfig('llmApiKey', e.target.value)}
+                    placeholder="留空则使用平台专用 API Key"
+                  />
+                  <Button
+                    variant="outline"
+                    size="icon"
+                    onClick={() => toggleShowApiKey('llm')}
+                  >
+                    {showApiKeys['llm'] ? <EyeOff className="h-4 w-4" /> : <Eye className="h-4 w-4" />}
+                  </Button>
+                </div>
+                <p className="text-xs text-muted-foreground">
+                  如果设置，将优先使用此 API Key；否则使用下方对应平台的专用 API Key
+                </p>
+              </div>
+
+              <div className="space-y-2">
+                <Label>模型名称（可选）</Label>
+                <Input
+                  value={config.llmModel}
+                  onChange={(e) => updateConfig('llmModel', e.target.value)}
+                  placeholder={`默认：${DEFAULT_MODELS[config.llmProvider as keyof typeof DEFAULT_MODELS] || '自动'}`}
+                />
+                <p className="text-xs text-muted-foreground">
+                  留空使用默认模型
+                </p>
+              </div>
+
+              <div className="space-y-2">
+                <Label>API 基础 URL（推荐配置）</Label>
+                <Input
+                  value={config.llmBaseUrl}
+                  onChange={(e) => updateConfig('llmBaseUrl', e.target.value)}
+                  placeholder="例如：https://api.example.com/v1"
+                />
+                <div className="text-xs text-muted-foreground space-y-1">
+                  <p>💡 <strong>使用 API 中转站？</strong>在这里填入中转站地址</p>
+                  <details className="cursor-pointer">
+                    <summary className="text-primary hover:underline">查看常见 API 中转示例</summary>
+                    <div className="mt-2 p-3 bg-muted rounded space-y-1 text-xs">
+                      <p><strong>OpenAI 兼容格式：</strong></p>
+                      <p>• https://your-proxy.com/v1</p>
+                      <p>• https://api.openai-proxy.org/v1</p>
+                      <p className="pt-2"><strong>其他中转格式：</strong></p>
+                      <p>• https://your-api-gateway.com/openai</p>
+                      <p>• https://custom-endpoint.com/api</p>
+                      <p className="pt-2 text-orange-600">⚠️ 确保中转站支持你选择的 LLM 平台</p>
+                    </div>
+                  </details>
+                </div>
+              </div>
+
+              <div className="grid grid-cols-1 md:grid-cols-3 gap-4">
+                <div className="space-y-2">
+                  <Label>超时时间（毫秒）</Label>
+                  <Input
+                    type="number"
+                    value={config.llmTimeout}
+                    onChange={(e) => updateConfig('llmTimeout', Number(e.target.value))}
+                  />
+                </div>
+                <div className="space-y-2">
+                  <Label>温度参数（0-2）</Label>
+                  <Input
+                    type="number"
+                    step="0.1"
+                    min="0"
+                    max="2"
+                    value={config.llmTemperature}
+                    onChange={(e) => updateConfig('llmTemperature', Number(e.target.value))}
+                  />
+                </div>
+                <div className="space-y-2">
+                  <Label>最大 Tokens</Label>
+                  <Input
+                    type="number"
+                    value={config.llmMaxTokens}
+                    onChange={(e) => updateConfig('llmMaxTokens', Number(e.target.value))}
+                  />
+                </div>
+              </div>
+            </CardContent>
+          </Card>
+
+        </TabsContent>
+
+        {/* 平台专用密钥 */}
+        <TabsContent value="platforms" className="space-y-6">
+          <Alert>
+            <Key className="h-4 w-4" />
+            <AlertDescription>
+              <div className="space-y-1">
+                <p>配置各平台的 API Key，方便快速切换。如果设置了通用 API Key，将优先使用通用配置。</p>
+                <p className="text-xs text-muted-foreground pt-1">
+                  💡 <strong>使用 API 中转站的用户注意：</strong>这里填入的应该是<strong>中转站提供的 API Key</strong>，而不是官方 Key。
+                  中转站地址请在「LLM 配置」标签页的「API 基础 URL」中填写。
+                </p>
+              </div>
+            </AlertDescription>
+          </Alert>
+
+          {[
+            { key: 'geminiApiKey', label: 'Google Gemini API Key', icon: '🔵', hint: '官方：https://makersuite.google.com/app/apikey | 或使用中转站 Key' },
+            { key: 'openaiApiKey', label: 'OpenAI API Key', icon: '🟢', hint: '官方：https://platform.openai.com/api-keys | 或使用中转站 Key' },
+            { key: 'claudeApiKey', label: 'Claude API Key', icon: '🟣', hint: '官方：https://console.anthropic.com/ | 或使用中转站 Key' },
+            { key: 'qwenApiKey', label: '通义千问 API Key', icon: '🟠', hint: '官方：https://dashscope.console.aliyun.com/ | 或使用中转站 Key' },
+            { key: 'deepseekApiKey', label: 'DeepSeek API Key', icon: '🔷', hint: '官方：https://platform.deepseek.com/ | 或使用中转站 Key' },
+            { key: 'zhipuApiKey', label: '智谱AI API Key', icon: '🔴', hint: '官方：https://open.bigmodel.cn/ | 或使用中转站 Key' },
+            { key: 'moonshotApiKey', label: 'Moonshot API Key', icon: '🌙', hint: '官方：https://platform.moonshot.cn/ | 或使用中转站 Key' },
+            { key: 'baiduApiKey', label: '百度文心 API Key', icon: '🔵', hint: '官方格式：API_KEY:SECRET_KEY | 或使用中转站 Key' },
+            { key: 'minimaxApiKey', label: 'MiniMax API Key', icon: '⚡', hint: '官方：https://www.minimaxi.com/ | 或使用中转站 Key' },
+            { key: 'doubaoApiKey', label: '字节豆包 API Key', icon: '🎯', hint: '官方：https://console.volcengine.com/ark | 或使用中转站 Key' },
+          ].map(({ key, label, icon, hint }) => (
+            <Card key={key}>
+              <CardHeader>
+                <CardTitle className="text-base flex items-center gap-2">
+                  <span>{icon}</span>
+                  {label}
+                </CardTitle>
+                <CardDescription className="text-xs">{hint}</CardDescription>
+              </CardHeader>
+              <CardContent>
+                <div className="flex gap-2">
+                  <Input
+                    type={showApiKeys[key] ? 'text' : 'password'}
+                    value={config[key as keyof SystemConfigData] as string}
+                    onChange={(e) => updateConfig(key as keyof SystemConfigData, e.target.value)}
+                    placeholder={`输入 ${label}`}
+                  />
+                  <Button
+                    variant="outline"
+                    size="icon"
+                    onClick={() => toggleShowApiKey(key)}
+                  >
+                    {showApiKeys[key] ? <EyeOff className="h-4 w-4" /> : <Eye className="h-4 w-4" />}
+                  </Button>
+                </div>
+              </CardContent>
+            </Card>
+          ))}
+
+          <Card>
+            <CardHeader>
+              <CardTitle className="text-base flex items-center gap-2">
+                <span>🖥️</span>
+                Ollama 基础 URL
+              </CardTitle>
+              <CardDescription className="text-xs">本地 Ollama 服务的 API 端点</CardDescription>
+            </CardHeader>
+            <CardContent>
+              <Input
+                value={config.ollamaBaseUrl}
+                onChange={(e) => updateConfig('ollamaBaseUrl', e.target.value)}
+                placeholder="http://localhost:11434/v1"
+              />
+            </CardContent>
+          </Card>
+        </TabsContent>
+
+        {/* 分析参数配置 */}
+        <TabsContent value="analysis" className="space-y-6">
+          <Card>
+            <CardHeader>
+              <CardTitle>代码分析参数</CardTitle>
+              <CardDescription>调整代码分析的行为和性能</CardDescription>
+            </CardHeader>
+            <CardContent className="space-y-4">
+              <div className="space-y-2">
+                <Label>最大分析文件数</Label>
+                <Input
+                  type="number"
+                  value={config.maxAnalyzeFiles}
+                  onChange={(e) => updateConfig('maxAnalyzeFiles', Number(e.target.value))}
+                />
+                <p className="text-xs text-muted-foreground">
+                  单次分析任务最多处理的文件数量
+                </p>
+              </div>
+
+              <div className="space-y-2">
+                <Label>LLM 并发请求数</Label>
+                <Input
+                  type="number"
+                  value={config.llmConcurrency}
+                  onChange={(e) => updateConfig('llmConcurrency', Number(e.target.value))}
+                />
+                <p className="text-xs text-muted-foreground">
+                  同时发送给 LLM 的请求数量（降低可避免速率限制）
+                </p>
+              </div>
+
+              <div className="space-y-2">
+                <Label>请求间隔（毫秒）</Label>
+                <Input
+                  type="number"
+                  value={config.llmGapMs}
+                  onChange={(e) => updateConfig('llmGapMs', Number(e.target.value))}
+                />
+                <p className="text-xs text-muted-foreground">
+                  每个 LLM 请求之间的延迟时间
+                </p>
+              </div>
+
+              <div className="space-y-2">
+                <Label>输出语言</Label>
+                <Select
+                  value={config.outputLanguage}
+                  onValueChange={(value) => updateConfig('outputLanguage', value)}
+                >
+                  <SelectTrigger>
+                    <SelectValue />
+                  </SelectTrigger>
+                  <SelectContent>
+                    <SelectItem value="zh-CN">🇨🇳 中文</SelectItem>
+                    <SelectItem value="en-US">🇺🇸 English</SelectItem>
+                  </SelectContent>
+                </Select>
+              </div>
+            </CardContent>
+          </Card>
+        </TabsContent>
+
+        {/* 其他配置 */}
+        <TabsContent value="other" className="space-y-6">
+          <Card>
+            <CardHeader>
+              <CardTitle>GitHub 集成</CardTitle>
+              <CardDescription>配置 GitHub Personal Access Token 以访问私有仓库</CardDescription>
+            </CardHeader>
+            <CardContent>
+              <div className="space-y-2">
+                <Label>GitHub Token（可选）</Label>
+                <div className="flex gap-2">
+                  <Input
+                    type={showApiKeys['github'] ? 'text' : 'password'}
+                    value={config.githubToken}
+                    onChange={(e) => updateConfig('githubToken', e.target.value)}
+                    placeholder="ghp_xxxxxxxxxxxx"
+                  />
+                  <Button
+                    variant="outline"
+                    size="icon"
+                    onClick={() => toggleShowApiKey('github')}
+                  >
+                    {showApiKeys['github'] ? <EyeOff className="h-4 w-4" /> : <Eye className="h-4 w-4" />}
+                  </Button>
+                </div>
+                <p className="text-xs text-muted-foreground">
+                  获取：https://github.com/settings/tokens
+                </p>
+              </div>
+            </CardContent>
+          </Card>
+
+          <Card>
+            <CardHeader>
+              <CardTitle>配置说明</CardTitle>
+            </CardHeader>
+            <CardContent className="space-y-3 text-sm text-muted-foreground">
+              <div className="flex items-start gap-3 p-3 bg-muted rounded-lg">
+                <Database className="h-5 w-5 text-primary mt-0.5" />
+                <div>
+                  <p className="font-medium text-foreground">运行时配置</p>
+                  <p>
+                    配置保存在浏览器 localStorage 中，刷新页面后立即生效。
+                    可以在不重新构建 Docker 镜像的情况下修改配置。
+                  </p>
+                </div>
+              </div>
+              <div className="flex items-start gap-3 p-3 bg-muted rounded-lg">
+                <Settings className="h-5 w-5 text-green-600 mt-0.5" />
+                <div>
+                  <p className="font-medium text-foreground">配置优先级</p>
+                  <p>
+                    运行时配置 &gt; 构建时配置。如果设置了运行时配置，将覆盖构建时的环境变量。
+                  </p>
+                </div>
+              </div>
+              <div className="flex items-start gap-3 p-3 bg-muted rounded-lg">
+                <Key className="h-5 w-5 text-orange-600 mt-0.5" />
+                <div>
+                  <p className="font-medium text-foreground">安全提示</p>
+                  <p>
+                    API Keys 存储在浏览器本地，其他网站无法访问。但清除浏览器数据会删除所有配置。
+                  </p>
+                </div>
+              </div>
+            </CardContent>
+          </Card>
+        </TabsContent>
+      </Tabs>
+
+      {/* 底部操作按钮 */}
+      {hasChanges && (
+        <div className="fixed bottom-6 right-6 flex gap-3 bg-background border rounded-lg shadow-lg p-4">
+          <Button onClick={saveConfig} size="lg">
+            <Save className="w-4 h-4 mr-2" />
+            保存所有更改
+          </Button>
+          <Button onClick={loadConfig} variant="outline" size="lg">
+            取消
+          </Button>
+        </div>
+      )}
+    </div>
+  );
+}
+
--- a/src/pages/AdminDashboard.tsx
+++ b/src/pages/AdminDashboard.tsx
@ -16,11 +16,13 @@ import {
  Clock,
  AlertTriangle,
  TrendingUp,
-  Package
+  Package,
+  Settings
 } from "lucide-react";
 import { api, dbMode, isLocalMode } from "@/shared/config/database";
 import { DatabaseManager } from "@/components/database/DatabaseManager";
 import { DatabaseStatusDetail } from "@/components/database/DatabaseStatus";
+import { SystemConfig } from "@/components/system/SystemConfig";
 import { toast } from "sonner";

 export default function AdminDashboard() {
@ -112,11 +114,11 @@ export default function AdminDashboard() {
      <div className="flex items-center justify-between">
        <div>
          <h1 className="text-3xl font-bold text-gray-900 flex items-center gap-3">
-            <Database className="h-8 w-8 text-primary" />
-            数据库管理
+            <Settings className="h-8 w-8 text-primary" />
+            系统管理
          </h1>
          <p className="text-gray-600 mt-2">
-            管理和监控本地数据库，查看存储使用情况和数据统计
+            管理系统配置、LLM设置、数据库和存储使用情况
          </p>
        </div>
        <Button variant="outline" onClick={loadStats}>
@ -212,14 +214,20 @@ export default function AdminDashboard() {
      </div>

      {/* 主要内容标签页 */}
-      <Tabs defaultValue="overview" className="w-full">
-        <TabsList className="grid w-full grid-cols-4">
+      <Tabs defaultValue="config" className="w-full">
+        <TabsList className="grid w-full grid-cols-5">
+          <TabsTrigger value="config">系统配置</TabsTrigger>
          <TabsTrigger value="overview">数据概览</TabsTrigger>
          <TabsTrigger value="storage">存储管理</TabsTrigger>
          <TabsTrigger value="operations">数据操作</TabsTrigger>
-          <TabsTrigger value="settings">设置</TabsTrigger>
+          <TabsTrigger value="settings">高级设置</TabsTrigger>
        </TabsList>

+        {/* 系统配置 */}
+        <TabsContent value="config" className="space-y-6">
+          <SystemConfig />
+        </TabsContent>
+
        {/* 数据概览 */}
        <TabsContent value="overview" className="space-y-6">
          <div className="grid grid-cols-1 lg:grid-cols-2 gap-6">
--- a/src/shared/config/env.ts
+++ b/src/shared/config/env.ts
@ -1,85 +1,98 @@
-// 环境变量配置
+// 从 localStorage 读取运行时配置
+const STORAGE_KEY = 'xcodereviewer_runtime_config';
+const getRuntimeConfig = () => {
+  try {
+    const saved = localStorage.getItem(STORAGE_KEY);
+    return saved ? JSON.parse(saved) : null;
+  } catch {
+    return null;
+  }
+};
+
+const runtimeConfig = getRuntimeConfig();
+
+// 环境变量配置（支持运行时配置覆盖）
 export const env = {
  // ==================== LLM 通用配置 ====================
  // 当前使用的LLM提供商 (gemini|openai|claude|qwen|deepseek|zhipu|moonshot|baidu|minimax|doubao|ollama)
-  LLM_PROVIDER: import.meta.env.VITE_LLM_PROVIDER || 'gemini',
+  LLM_PROVIDER: runtimeConfig?.llmProvider || import.meta.env.VITE_LLM_PROVIDER || 'gemini',
  // LLM API Key
-  LLM_API_KEY: import.meta.env.VITE_LLM_API_KEY || '',
+  LLM_API_KEY: runtimeConfig?.llmApiKey || import.meta.env.VITE_LLM_API_KEY || '',
  // LLM 模型名称
-  LLM_MODEL: import.meta.env.VITE_LLM_MODEL || '',
+  LLM_MODEL: runtimeConfig?.llmModel || import.meta.env.VITE_LLM_MODEL || '',
  // LLM API 基础URL (可选，用于自定义端点或代理)
-  LLM_BASE_URL: import.meta.env.VITE_LLM_BASE_URL || '',
+  LLM_BASE_URL: runtimeConfig?.llmBaseUrl || import.meta.env.VITE_LLM_BASE_URL || '',
  // LLM 请求超时时间(ms)
-  LLM_TIMEOUT: Number(import.meta.env.VITE_LLM_TIMEOUT) || 150000,
+  LLM_TIMEOUT: runtimeConfig?.llmTimeout || Number(import.meta.env.VITE_LLM_TIMEOUT) || 150000,
  // LLM 温度参数 (0.0-2.0)
-  LLM_TEMPERATURE: Number(import.meta.env.VITE_LLM_TEMPERATURE) || 0.2,
+  LLM_TEMPERATURE: runtimeConfig?.llmTemperature !== undefined ? runtimeConfig.llmTemperature : (Number(import.meta.env.VITE_LLM_TEMPERATURE) || 0.2),
  // LLM 最大token数
-  LLM_MAX_TOKENS: Number(import.meta.env.VITE_LLM_MAX_TOKENS) || 4096,
+  LLM_MAX_TOKENS: runtimeConfig?.llmMaxTokens || Number(import.meta.env.VITE_LLM_MAX_TOKENS) || 4096,

  // ==================== Gemini AI 配置 (兼容旧配置) ====================
-  GEMINI_API_KEY: import.meta.env.VITE_GEMINI_API_KEY || '',
+  GEMINI_API_KEY: runtimeConfig?.geminiApiKey || import.meta.env.VITE_GEMINI_API_KEY || '',
  GEMINI_MODEL: import.meta.env.VITE_GEMINI_MODEL || 'gemini-2.5-flash',
  GEMINI_TIMEOUT_MS: Number(import.meta.env.VITE_GEMINI_TIMEOUT_MS) || 25000,

  // ==================== OpenAI 配置 ====================
-  OPENAI_API_KEY: import.meta.env.VITE_OPENAI_API_KEY || '',
+  OPENAI_API_KEY: runtimeConfig?.openaiApiKey || import.meta.env.VITE_OPENAI_API_KEY || '',
  OPENAI_MODEL: import.meta.env.VITE_OPENAI_MODEL || 'gpt-4o-mini',
  OPENAI_BASE_URL: import.meta.env.VITE_OPENAI_BASE_URL || '',

  // ==================== Claude 配置 ====================
-  CLAUDE_API_KEY: import.meta.env.VITE_CLAUDE_API_KEY || '',
+  CLAUDE_API_KEY: runtimeConfig?.claudeApiKey || import.meta.env.VITE_CLAUDE_API_KEY || '',
  CLAUDE_MODEL: import.meta.env.VITE_CLAUDE_MODEL || 'claude-3-5-sonnet-20241022',

  // ==================== 通义千问 配置 ====================
-  QWEN_API_KEY: import.meta.env.VITE_QWEN_API_KEY || '',
+  QWEN_API_KEY: runtimeConfig?.qwenApiKey || import.meta.env.VITE_QWEN_API_KEY || '',
  QWEN_MODEL: import.meta.env.VITE_QWEN_MODEL || 'qwen-turbo',

  // ==================== DeepSeek 配置 ====================
-  DEEPSEEK_API_KEY: import.meta.env.VITE_DEEPSEEK_API_KEY || '',
+  DEEPSEEK_API_KEY: runtimeConfig?.deepseekApiKey || import.meta.env.VITE_DEEPSEEK_API_KEY || '',
  DEEPSEEK_MODEL: import.meta.env.VITE_DEEPSEEK_MODEL || 'deepseek-chat',

  // ==================== 智谱AI 配置 ====================
-  ZHIPU_API_KEY: import.meta.env.VITE_ZHIPU_API_KEY || '',
+  ZHIPU_API_KEY: runtimeConfig?.zhipuApiKey || import.meta.env.VITE_ZHIPU_API_KEY || '',
  ZHIPU_MODEL: import.meta.env.VITE_ZHIPU_MODEL || 'glm-4-flash',

  // ==================== Moonshot 配置 ====================
-  MOONSHOT_API_KEY: import.meta.env.VITE_MOONSHOT_API_KEY || '',
+  MOONSHOT_API_KEY: runtimeConfig?.moonshotApiKey || import.meta.env.VITE_MOONSHOT_API_KEY || '',
  MOONSHOT_MODEL: import.meta.env.VITE_MOONSHOT_MODEL || 'moonshot-v1-8k',

  // ==================== 百度文心一言 配置 ====================
-  BAIDU_API_KEY: import.meta.env.VITE_BAIDU_API_KEY || '',
+  BAIDU_API_KEY: runtimeConfig?.baiduApiKey || import.meta.env.VITE_BAIDU_API_KEY || '',
  BAIDU_MODEL: import.meta.env.VITE_BAIDU_MODEL || 'ERNIE-3.5-8K',

  // ==================== MiniMax 配置 ====================
-  MINIMAX_API_KEY: import.meta.env.VITE_MINIMAX_API_KEY || '',
+  MINIMAX_API_KEY: runtimeConfig?.minimaxApiKey || import.meta.env.VITE_MINIMAX_API_KEY || '',
  MINIMAX_MODEL: import.meta.env.VITE_MINIMAX_MODEL || 'abab6.5-chat',

  // ==================== 豆包 配置 ====================
-  DOUBAO_API_KEY: import.meta.env.VITE_DOUBAO_API_KEY || '',
+  DOUBAO_API_KEY: runtimeConfig?.doubaoApiKey || import.meta.env.VITE_DOUBAO_API_KEY || '',
  DOUBAO_MODEL: import.meta.env.VITE_DOUBAO_MODEL || 'doubao-pro-32k',

  // ==================== Ollama 本地模型配置 ====================
  OLLAMA_API_KEY: import.meta.env.VITE_OLLAMA_API_KEY || 'ollama',
  OLLAMA_MODEL: import.meta.env.VITE_OLLAMA_MODEL || 'llama3',
-  OLLAMA_BASE_URL: import.meta.env.VITE_OLLAMA_BASE_URL || 'http://localhost:11434/v1',
+  OLLAMA_BASE_URL: runtimeConfig?.ollamaBaseUrl || import.meta.env.VITE_OLLAMA_BASE_URL || 'http://localhost:11434/v1',

  // ==================== Supabase 配置 ====================
  SUPABASE_URL: import.meta.env.VITE_SUPABASE_URL || '',
  SUPABASE_ANON_KEY: import.meta.env.VITE_SUPABASE_ANON_KEY || '',

  // ==================== GitHub 配置 ====================
-  GITHUB_TOKEN: import.meta.env.VITE_GITHUB_TOKEN || '',
+  GITHUB_TOKEN: runtimeConfig?.githubToken || import.meta.env.VITE_GITHUB_TOKEN || '',

  // ==================== 应用配置 ====================
  APP_ID: import.meta.env.VITE_APP_ID || 'xcodereviewer',

  // ==================== 分析配置 ====================
-  MAX_ANALYZE_FILES: Number(import.meta.env.VITE_MAX_ANALYZE_FILES) || 40,
-  LLM_CONCURRENCY: Number(import.meta.env.VITE_LLM_CONCURRENCY) || 2,
-  LLM_GAP_MS: Number(import.meta.env.VITE_LLM_GAP_MS) || 500,
+  MAX_ANALYZE_FILES: runtimeConfig?.maxAnalyzeFiles || Number(import.meta.env.VITE_MAX_ANALYZE_FILES) || 40,
+  LLM_CONCURRENCY: runtimeConfig?.llmConcurrency || Number(import.meta.env.VITE_LLM_CONCURRENCY) || 2,
+  LLM_GAP_MS: runtimeConfig?.llmGapMs || Number(import.meta.env.VITE_LLM_GAP_MS) || 500,
  
  // ==================== 语言配置 ====================
-  OUTPUT_LANGUAGE: import.meta.env.VITE_OUTPUT_LANGUAGE || 'zh-CN', // zh-CN | en-US
+  OUTPUT_LANGUAGE: runtimeConfig?.outputLanguage || import.meta.env.VITE_OUTPUT_LANGUAGE || 'zh-CN', // zh-CN | en-US

  // ==================== 开发环境标识 ====================
  isDev: import.meta.env.DEV,