How kingjulio8238/memary works
Main Components Structure
• Chat Agent (Central component)
• Memory Module
• Memory Stream
• Entity Knowledge Store
• Knowledge Graph
• Routing Agent
b) External Services/Integrations:
• LLM Services (Ollama, GPT)
• Vision Models (LLaVA, GPT-4-Vision)
• Database Systems (FalkorDB/Neo4j)
• External APIs (Google Maps, Alpha Vantage, Perplexity)
Component Relationships
a) Primary Data Flow:
User Query → Chat Agent → Routing Agent → Knowledge Graph ↔ Memory Module
b) Memory Processing Flow:
Memory Stream → Entity Knowledge Store → Context Window Generation
c) Knowledge Graph Operations:
Query → Recursive Retrieval → Entity Extraction → Graph Storage
Specific Implementation Details
Key architectural elements:
a) Memory Module Section:
• Timeline Analysis component
• Entity Ranking System
• Knowledge Categorization
• Memory Compression
b) Knowledge Graph Section:
• Recursive Retrieval System
• Multi-hop Reasoning
• Entity Relationship Mapping
c) Context Window:
• Agent Response Handler
• Entity Relevance Processor
• Chat History Summarizer
Additional Details
Elements in separate panels or callouts:
• Memory Compression Workflow
• Entity Knowledge Store Structure
• Recursive Retrieval Process
• Multi-Graph Support
Technical Specifications
• Supported Model Types
• Database Options
• API Dependencies
• File Storage Formats
Implementation Notes
Create callouts highlighting:
• Auto-generated Memory Features
• Memory Module Integration Points
• System Improvement Metrics
• Memory Rewind Capabilities