NexDR is an advanced deep research agent built on the Nex-N1 agentic model and the NexAU agent framework. It autonomously decomposes research tasks, conducts parallel investigations, and generates structured research reports or visually rich presentation slides.
English | 简体中文
In the DeepResearch Bench evaluation, NexDR's comprehensive performance surpasses OpenAI Deep Research, demonstrating leading advantages in information accuracy, report structure, and multimodal visualization capabilities.
- Python 3.12 or higher
- pip or uv package manager
# Clone the repository
git clone https://github.com/nex-agi/NexDR.git
cd NexDR
# Install with uv
uv sync# Clone the repository
git clone https://github.com/nex-agi/NexDR.git
cd NexDR
# Install dependencies
pip install -e .Create a .env file in the project root with your API keys:
# Required for LLM capabilities (required)
LLM_API_KEY=your_openai_api_key_here
LLM_BASE_URL=your_openai_base_url_here
LLM_MODEL=model_name_you_used
# For web search and Serper document scraping (required)
SERPER_API_KEY=your_serper_api_key_here
# For document parsing (choose at least one provider)
JINA_API_KEY=your_jina_api_key_here
DOC_READER_PROVIDERS=jina,serper
# For Langfuse tracing (optional)
LANGFUSE_PUBLIC_KEY=your_langfuse_public_key
LANGFUSE_SECRET_KEY=your_langfuse_secret_key
LANGFUSE_HOST=https://cloud.langfuse.comRun a research task and generate a markdown report:
python quick_start.py \
--query "What are the latest developments in quantum computing?" \
--report_format markdown \
--output_dir workspaces/my_researchGenerate an HTML presentation instead:
python quick_start.py \
--query "Analyze the impact of AI on healthcare" \
--report_format html \
--output_dir workspaces/healthcare_researchGenerate a markdown and html report:
python quick_start.py \
--query "Analyze the impact of AI on healthcare" \
--report_format markdown+html \
--output_dir workspaces/healthcare_research--query: Your research question (required)--report_format: Output format -markdownorhtml(default:markdown)--output_dir: Directory to save results (default:workspaces/workspace_TIMESTAMP)
After running a research task, your workspace will contain:
workspaces/my_research/
├── logs_TIMESTAMP.log # Detailed execution logs
├── markdown_report.md # Research report (if markdown format)
├── citations.json # Citation references (if markdown format)
├── html_report.html # HTML presentation (if html format)
└── final_state.json # Execution metadata and statistics
uv run streamlit run demo_app.py"Please conduct an in-depth analysis of the success of Black Myth: Wukong, covering all aspects including development, release, and reviews, and create a presentation report."
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
⭐ If you find NexDR useful, please consider giving it a star on GitHub! ⭐