EPIC: Implement NodeSpace Business Logic Layer

[malibio]

Overview

This epic tracks the complete implementation of the NodeSpace Business Logic Layer - a Rust-based system with libsql that provides core data management, node operations, and AI integration.

Architecture

Pure JSON schema architecture with:

• packages/core: Universal node model, database layer (libsql + Turso), business services
• packages/nlp-engine: AI/ML services (embeddings, intent classification via llama.cpp-rs)

Key design principles:

• Pure JSON Schema: ALL entity data in properties field - no complementary tables, no ALTER TABLE migrations
• Schema-as-Node: Schemas stored as nodes with node_type = "schema", id = type_name
• Zero Migration Risk: Critical for desktop apps (10,000+ user machines, no rollback possible)
• Rule-Based Indexing: JSON path indexes created based on query frequency (not LLM-driven)
• Local-First: Embedded Turso with sync via embedded replicas
• AI-Native: Built-in vector embeddings (F32_BLOB) and semantic search
• Multi-Platform: Shared logic for desktop (Tauri) and AI agents (MCP stdio server)

Implementation Phases

Phase 1: Core Package Foundation

• Initialize packages/core with Cargo workspace configuration #98
• Define Node data structures and models #99 (Pure JSON properties field)
• Implement NodeBehavior trait system for extensible node types #100
• Set up database connection with libsql #101 (libsql, NOT sqlx)
• Initialize database schema (Pure JSON, no migrations) #102 (single nodes table + node_mentions)

Phase 2: Node Types & Services

• EPIC: Implement built-in node behaviors #103 (schema nodes)
• Implement NodeService with CRUD operations #104 (JSON path queries)
• Add mentions relation management and backlink queries #105
• Implement Schema Management Service with Migration Infrastructure #106 (schemas are nodes, no dynamic tables)
• Semantic search: validation, benchmarks, and documentation #107 (F32_BLOB)

Phase 3: NLP Engine Package

• Initialize packages/nlp-engine with embedding service #108 (llama.cpp-rs)
• Add intent classification for natural language commands #109

Phase 4: Integration & Testing

• Create Tauri command layer for desktop app integration #110
• Implement reactive state synchronization #111
• Add MCP stdio server for AI agent access #112
• Create comprehensive integration tests #113
• Add comprehensive error handling and logging #114
• Performance optimization and production readiness #115

Success Criteria

• Pure JSON schema with no ALTER TABLE risk
• Built-in node types as schema nodes
• Turso + JSON path indexes performing well
• Vector embeddings (F32_BLOB) for semantic search
• Rule-based IndexManager for dynamic indexes
• Auto-reference detection (primitive vs entity types)
• Tauri integration
• MCP stdio server
• Reactive state synchronization
• User-defined schemas without table creation
• Performance optimized (0-100K nodes free tier)

Documentation

• docs/architecture/business-logic/overview.md
• docs/architecture/business-logic/database-schema.md
• docs/architecture/business-logic/node-behavior-system.md
• docs/architecture/data/storage-architecture.md
• docs/architecture/core/technology-stack.md

[malibio]

Epic complete. The Business Logic Layer is fully implemented:

• SurrealDB backend with nodes table
• NodeBehavior trait system with Text, Header, Task, CodeBlock, QuoteBlock, OrderedList, Date, Schema behaviors
• NodeService with CRUD operations
• QueryService for SQL queries
• Embedding architecture for semantic search
• Domain event system
• MCP server integration
• Tauri commands layer