NexHTML - NexAU-based HTML Agent

Intelligent AI Agent Collection Based on NexAU Framework, Specializing in HTML Generation, Academic Poster Creation, Data Visualization, and More

English | 中文

Quick Start • Agent Overview

---

📋 Table of Contents

• Introduction
• Case Studies
• Features
• Quick Start
• Agent Overview

---

🎯 Introduction

NexHTML is an AI Agent development platform built on the Nexau Framework, integrating multiple specialized AI Agents designed to solve automation needs in real-world scenarios.

---

✨ Features

🌐 WebDevAgent - HTML Code Generation System

Professional HTML code generation and optimization AI system for creating high-quality frontend pages.

Core Features:

• 🖼️ Smart Image Search - Integrated Unsplash API with automatic image search and annotation
• 🎨 Visual Enhancement - AI-driven design suggestions and code optimization
• 📝 Auto Annotation - VLM-powered descriptive text generation for images
• 🔄 Iterative Optimization - Multi-round conversational code improvement

Use Cases:

• Landing Page Rapid Prototyping
• Marketing Page Generation
• Course Materials and Presentation Pages

📊 Paper2PosterAgent - Academic Poster Generation System

Automatically converts academic papers (PDF) into visually stunning academic posters in HTML.

Core Features:

• 📄 PDF Parsing - MinerU-based conversion from PDF to structured Markdown with image extraction
• 🖼️ Smart Image Annotation - VLM automatically generates titles and descriptions for paper figures
• 🏛️ Logo Management - Auto-extract institutional information and match university/organization logos
• 📱 QR Code Integration - Extract arXiv links and generate access QR codes
• 📐 Layout Optimization - AI-driven multi-column layout balancing and height detection
• 🎨 Poster Rendering - Generate professional HTML academic posters with preview screenshots

Use Cases:

• Conference Poster Generation
• Academic Presentation Materials
• Research Showcase and Exhibition

📈 DatavisSearchAgent - Data Visualization & Analysis System

End-to-end data visualization agent that transforms data topics into interactive dashboards with charts and insights.

Core Features:

• 🔍 Systematic Data Retrieval - Multi-source search with keyword expansion and cross-validation
• 🧹 Data Engineering - Structured extraction, cleaning, and quality assessment
• 🐍 Python Analysis - Stateful Python execution with pandas/numpy for data exploration
• 📊 Interactive Dashboards - Plotly-based HTML dashboards with real-time CSV data loading
• 🌐 HTTP Service - Non-blocking local server for dashboard display
• 📦 Kaggle Integration - Direct dataset download from Kaggle platform

Use Cases:

• Trend Analysis and Reporting
• Dataset Exploration and Visualization
• Business Intelligence Dashboards
• Academic Data Analysis

---

📸 Case Studies

WebDev Case 1

Prompt: View Complete Prompt →

WebDev Case 2

Prompt: View Complete PRD Document →

Paper2Poster Case 1

English Academic Poster

Chinese Academic Poster

---

🚀 Quick Start

1. Requirements

• Python: 3.13+
• System: macOS / Linux / Windows
• Tools: Git, uv (recommended) or pip

2. Clone Project

# Clone main project
git clone https://github.com/nex-agi/NexHTML.git
cd NexHTML

# Initialize submodules
git submodule update --init --recursive

3. Install Dependencies

Using uv (Recommended)

# One-command installation (installs everything including MinerU and Nexau)
uv pip install -e .

Using pip

# One-command installation
pip install -e .

4. Configure Environment Variables

Copy and edit .env file:

cp .env.example .env
vim .env

Required Configuration:

# Core LLM Configuration
LLM_MODEL=your_model_name
LLM_BASE_URL=https://api.openai.com/v1
LLM_API_KEY=your_api_key_here

5. Launch Agent

Launch WebDevAgent

Prerequisites:

Before launching WebDevAgent, you need to:

1. Apply for Unsplash API Key - Visit Unsplash Developers to register and obtain your API key for image search functionality
2. Configure VLM for Image Captioning - Set up a Vision Language Model to generate image descriptions
3. Update .env file with your credentials:

   UNSPLASH_ACCESS_KEYS=your_unsplash_key
   IMAGE_CAPTIONER_MODEL=your_vlm_model
   IMAGE_CAPTIONER_BASE_URL=your_vlm_url
   IMAGE_CAPTIONER_API_KEY=your_vlm_key

Launch Command:

uv run python src/WebDevAgent/start.py

Launch Paper2PosterAgent

Prerequisites:

Before launching Paper2PosterAgent, you need to:

1. Configure VLM (Vision Language Model) - Set up a Vision Language Model for image analysis and caption generation
2. Update .env file with your VLM credentials:

   VLM_MODEL=your_vlm_model
   VLM_BASE_URL=your_vlm_url
   VLM_API_KEY=your_vlm_key

Launch Commands:

# First start MinerU service (PDF parsing)
# In another terminal:
uv run mineru-api

# Start Agent
uv run src/Paper2PosterAgent/start.py

Launch DatavisSearchAgent

Prerequisites:

Before launching DatavisSearchAgent, you need to:

1. Apply for Serper API Key - Visit Serper.dev to register and obtain your API key for web search functionality
2. Configure Kaggle API Credentials (Optional) - For downloading datasets from Kaggle
3. Update .env file with your credentials:

   SERPER_API_KEY=your_serper_api_key
   
   # Optional: For Kaggle dataset downloads
   KAGGLE_USERNAME=your_kaggle_username
   KAGGLE_KEY=your_kaggle_api_key

Launch Command:

uv run python src/DatavisSearchAgent/start.py

6. Optional: Configure Langfuse Monitoring

To enable observability and monitoring with Langfuse, add the following to your .env file:

LANGFUSE_SECRET_KEY=sk-lf-xxx
LANGFUSE_PUBLIC_KEY=pk-lf-xxx
LANGFUSE_HOST=http://your-langfuse-host:port

See Configuration Guide → for more configuration details.

---

🎯 Agent Overview

For detailed information about each Agent (configuration, tools, workflows), please see Agent Overview Documentation →

---

🙏 Acknowledgments

We extend our gratitude to NexAU Framework, MinerU, Paper2Poster, Unsplash, Langfuse, and other projects for providing their codebases and service support.

---

⭐ If this project helps you, please give it a Star! ⭐