TTD-DR: Test-Time Diffusion Deep Research

A sophisticated AI-powered research system that implements test-time diffusion methodology for comprehensive, multi-faceted research using Kimi API and LangGraph orchestration.

Overview

TTD-DR (Test-Time Diffusion Deep Research) is an advanced research automation system that leverages the power of Kimi AI and LangGraph workflows to conduct deep, iterative research with self-evolution capabilities. The system implements test-time diffusion principles to iteratively refine research quality through multiple rounds of gap analysis and content enhancement.

Features

• Test-Time Diffusion Methodology: Iterative research refinement with quality scoring
• Multi-Engine Search Integration: Supports Tavily, DuckDuckGo, and Naver search APIs
• Self-Evolution System: Automatic gap identification and content enhancement
• Multi-Faceted Research: Handles complex, multi-dimensional research queries
• Kimi API Integration: Optimized for Moonshot AI's Kimi model
• Comprehensive State Management: Full research lifecycle tracking with LangGraph

Quick Start

See the full documentation and usage examples in the project files.

License

MIT License