Arkef
Design and validate complex software architectures with spec-driven nodes in a visual graph database - catch architectural flaws before writing code.
Project Description
Arkef is a cutting-edge GraphRAG-powered application prototype builder that leverages deep Neo4j integration to create contextually-aware AI-generated user journeys, wireframes, and full page structures. The system builds a knowledge graph that maintains relationships between all application components, enabling intelligent code generation with full contextual awareness.
The application is production-ready for Scalingo deployment with web: yarn start -p $PORT.
Key deployment characteristics:
Next.js 15.4.3 with Turbopack for optimized builds
Server-side rendering (SSR) for API routes
Parse Server integration for persistent data storage
Neo4j Aura cloud-ready connection with environment-based configuration
SSE (Server-Sent Events) streaming for real-time wireframe generation feedback
🧠Deep MCP Integration with GraphRAG
The project implements a sophisticated GraphRAG architecture using Neo4j as the knowledge graph backbone with a hierarchical structure: Journey → Wireframe → Page → Section → Component / Style.
Prior Work
Created During the Hackathon (100% New)
Neo4j GraphRAG Architecture:
Complete knowledge graph schema design (Journey → Wireframe → Page → Section → Component)
neo4j-repository.ts — Generic repository layer with Cypher queries
neo4j.ts — Driver connection and context retrieval functions
/api/neo4j/context — Context and breadcrumb API endpoint
All graph relationships (HAS_PAGE, HAS_SECTION, LEADS_TO, DEPENDS_ON, etc.)
AI-Powered Generation System:
Claude Haiku 4.5 integration via @ai-sdk/anthropic
/api/journey/generate — Streaming user journey generation
/api/wireframe/generate — Real-time wireframe generation with SSE
System prompts for contextual code generation
Section binding comments parsing for Neo4j sync
Frontend Application:
ReactFlow-based journey visualization canvas
Brain Modules system (AI chat, notes, debugging per node)
Page Editor Panel with section management
Streaming code preview components
Scalingo Deployment Configuration:
Production-ready Procfile and environment setup
Neo4j Aura cloud integration