[English | 中文]

Deep Research is an advanced, AI-powered research assistant built with Nuxt 3. It leverages Large Language Models (LLMs) and recursive web search techniques to automate the process of deep information retrieval, analysis, and report generation.

Give it a topic, and it will:

1. Plan: Generate a research framework and search keywords.
2. Search: Execute parallel web searches and crawl content using industry-leading tools.
3. Analyze: Read, deduplicate, and synthesize information from multiple sources.
4. Report: Produce a structured, comprehensive research report with citations.
5. Visualize: Display the entire research thought process in an interactive flow.

• 🚀 Safe & Secure: All configuration and API keys are stored locally in your browser (Client Mode), or securely on your server (Server Mode).
• 🔍 Recursive Deep Search: Goes beyond surface-level results by recursively exploring related topics and following leads.
• 🕙 Real-time Feedback: Watch the AI think, plan, and search in real-time with a transparent execution flow.
• 📊 Visual Knowledge Graph: Interactive tree visualization of the research path and gathered information.
• 📄 Professional Reports: Exports findings as polished Markdown or PDF reports.
• 🤖 Multi-Model Support: Works with OpenAI, DeepSeek, SiliconFlow, OpenRouter, Ollama, and more.
• 🌍 Multi-Language: Supports research and reporting in multiple languages (English, Chinese, Dutch, etc.).
• 🐳 Easy Deployment: Docker-ready with support for both simplified Server Mode (env vars) and flexible Client Mode.

Built on a modern, robust tech stack:

• Framework: Nuxt 4 (Vue 3)
• UI Library: Nuxt UI & Tailwind CSS
• State Management: Pinia
• AI Integration: Vercel AI SDK
• Search Integration: Tavily & Firecrawl

You can run Deep Research in two modes: Server Mode (API keys set on server) or Client Mode (users enter API keys in browser).

Users don't need to configure anything. You provide the API keys via environment variables.

docker run -p 3000:3000 \
  -e NUXT_PUBLIC_SERVER_MODE=true \
  -e NUXT_AI_API_KEY=your-ai-api-key \
  -e NUXT_WEB_SEARCH_API_KEY=your-search-api-key \
  -e NUXT_PUBLIC_AI_PROVIDER=openai-compatible \
  -e NUXT_PUBLIC_AI_MODEL=gpt-4o-mini \
  -e NUXT_PUBLIC_WEB_SEARCH_PROVIDER=tavily \
  anotia/deep-research-web:latest

Users configure their own API keys in the browser settings.

docker run -p 3000:3000 --name deep-research-web -d anotia/deep-research-web:latest

1. Clone the repository:

git clone https://github.com/AnotiaWang/deep-research-web-ui.git cd deep-research-web-ui

2.  **Install dependencies**:
    ```bash
pnpm install

3. Start development server:

pnpm dev

    Access the app at `http://localhost:3000`.

4.  **Build for production**:
    ```bash
pnpm build

Variable	Description	Default
Server Mode
NUXT_PUBLIC_SERVER_MODE	Enable server-side configuration mode	false
NUXT_AI_API_KEY	API Key for the AI Provider	-
NUXT_AI_API_BASE	Base URL for the AI Provider	-
NUXT_WEB_SEARCH_API_KEY	API Key for Search Provider (Tavily/Firecrawl)	-
Public Settings
NUXT_PUBLIC_AI_PROVIDER	AI Provider (openai, deepseek, siliconflow, custom)	openai-compatible
NUXT_PUBLIC_AI_MODEL	Model ID (e.g., gpt-4o, deepseek-chat)	gpt-4o-mini
NUXT_PUBLIC_AI_CONTEXT_SIZE	Max context window size	128000
NUXT_PUBLIC_WEB_SEARCH_PROVIDER	Search Provider (tavily, firecrawl, google-pse)	tavily
NUXT_PUBLIC_WEB_SEARCH_CONCURRENCY_LIMIT	Max parallel search requests	2
NUXT_PUBLIC_WEB_SEARCH_SEARCH_LANGUAGE	Default search language	en

AI Models:

• OpenAI (and compatibles)
• DeepSeek
• SiliconFlow
• OpenRouter
• Ollama (Local)
• InfiniAI

Web Search:

• Tavily (Recommended)
• Firecrawl
• Google Programmable Search Engine (PSE)

/
├── app/                      # Nuxt Frontend
│   ├── components/           # Vue Components (ResearchForm, Reports, etc.)
│   ├── composables/          # Shared logic (AI providers, Search logic)
│   ├── stores/               # Pinia State (Config, History)
│   └── pages/                # Application Routes
├── server/                   # Nitro Backend
│   └── api/
│       ├── research.post.ts  # Core recursive research logic
│       └── report.post.ts    # Report generation endpoint
├── lib/                      # Core Business Logic (Shared)
│   ├── ai/                   # AI Provider abstractions & Prompts
│   └── core/                 # Deep Research algorithm implementation
└── docs/                     # Detailed Documentation

• Recursive deep search & analysis
• Multi-language support
• Docker deployment
• Real-time visualization
• Enhanced Reporting: PDF/Markdown export improvements, data visualization integration.
• Advanced Agents: Multi-agent collaboration for complex tasks.
• Data Sources: Support for PDF upload and academic paper analysis (RAG).
• User System: Authentication and cloud history sync.

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the Project
2. Create your Feature Branch (git checkout -b feature/AmazingFeature)
3. Commit your Changes (git commit -m 'Add some AmazingFeature')
4. Push to the Branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

Distributed under the MIT License. See LICENSE for more information.

Unlock Reliable Proxy Services with Swiftproxy

With Swiftproxy, you can access high-performance, secure proxies to enhance your web automation, privacy, and data collection efforts. Our services are trusted by developers and businesses to scale scraping tasks and ensure a safe online experience. Get started today at Swiftproxy.net. Use the coupon GHB5 to get 10% off!