ARGUS-X: Agentic Research Gap Understanding & Synthesis System

Production-grade agentic AI system for deep research paper analysis, gap detection, and research direction synthesis.

Features

• Hierarchical Paper Parsing: Preserves document structure with figures, tables, equations
• Claim-Assumption-Evidence Graph (CAEG): Formal representation of research claims
• 9 Specialized Agents: Coordinated multi-agent analysis pipeline
• Formal Gap Ontology: 7-class research gap taxonomy
• Literature Grounding: Real-time validation against arXiv and citation graphs
• Falsification Loop: Active counter-evidence search and reviewer simulation
• Multi-Paper Synthesis: Cross-paper reasoning and field-level insights
• Research-Grade UI: Modern interface with graph visualizations

Quick Start

Prerequisites

• Python 3.11+
• Node.js 18+
• Poetry

Installation

# Install backend dependencies
poetry install

# Install frontend dependencies
cd frontend
npm install
cd ..

# Configure environment
cp .env.example .env
# Edit .env with your API keys

Running

# Start backend
poetry run python -m argus.main

# Start frontend (in another terminal)
cd frontend
npm run dev

Access the UI at http://localhost:5173

Architecture

Agent Pipeline

Parse → Extract Claims & Assumptions → Evaluate Evidence
  → Detect Gaps → Validate via Literature → Falsify
  → Cross-Paper Synthesis → Research Direction Synthesis

Memory Layers

• Episodic: Paper-level context and findings
• Semantic: Field-level knowledge and patterns
• Meta: Historical gap validity and agent performance

API Endpoints

• POST /api/papers/upload - Upload research papers
• GET /api/papers/{paper_id} - Get paper structure
• POST /api/analyze/{paper_id} - Run analysis pipeline
• GET /api/caeg/{paper_id} - Get claim-evidence graph
• GET /api/gaps/{paper_id} - Get detected research gaps
• GET /api/directions/{paper_id} - Get research directions

Configuration

See .env.example for all configuration options.

License

Proprietary - Research Use Only