Simple Good Llm Memory Server

Quick Start

1. Start Services

docker-compose up -d

This starts:

• ChromaDB (port 8000) - Vector storage for semantic search
• Neo4j (ports 7474, 7687) - Knowledge graph for relationships

2. Configure in llm client

Type /mcp in llm client and add:

{
  "mcpServers": {
    "skynet-memory": {
      "command": "node",
      "args": ["/Users/PROECT_DIRECTORY/dist/index.js"],
      "env": {
        "CHROMA_URL": "http://localhost:8000",
        "NEO4J_URI": "bolt://localhost:7687",
        "NEO4J_USER": "neo4j",
        "NEO4J_PASSWORD": "password123"
      }
    }
  }
}

3. Restart llm client

Changes take effect after restart.

Features

Dual Storage Architecture

• ChromaDB: Vector embeddings for semantic search
• Neo4j: Graph relationships between memories, tags, sessions

Hybrid Search

• Semantic search via embeddings (Google text-embedding-004)
• Keyword fallback when semantic results are poor
• Intelligent merging and ranking

Memory Operations

• Save: Store with tags, importance (1-10), context, session
• Search: Semantic + keyword hybrid with filters
• Update: Modify existing memories
• Delete: Remove with automatic graph cleanup
• Related: Find semantically similar memories
• Stats: Analytics and tag distribution
• Time Range: Temporal queries with pagination

Resilience

• Automatic retry queues for failed syncs
• Health checks and graceful degradation
• Falls back to hash-based embeddings if API unavailable

Architecture

┌─────────────────────────────────────────┐
│         MCP Server (index.js)           │
│  - 8 Tools exposed via MCP protocol     │
└────────────┬────────────────────────────┘
             │
┌────────────┴────────────────────────────┐
│   Conscious Memory Service              │
│  - Hybrid search logic                  │
│  - Importance scoring                   │
│  - Session management                   │
└────────┬─────────────┬──────────────────┘
         │             │
         ▼             ▼
┌─────────────┐  ┌──────────────────┐
│  ChromaDB   │  │  Neo4j KG        │
│  Vectors    │  │  Relationships   │
└─────────────┘  └──────────────────┘

8 MCP Tools

1. save_memory - Store information
2. search_memories - Search with filters
3. search_memories_by_time_range - Temporal queries
4. update_memory - Modify memories
5. delete_memory - Remove memories
6. get_memory_tags - List all tags
7. get_related_memories - Find similar
8. get_memory_stats - Analytics

Environment Variables

# ChromaDB
CHROMA_URL=http://localhost:8000
MEMORY_COLLECTION_NAME=skynet_memories

# Neo4j
NEO4J_URI=bolt://localhost:7687
NEO4J_USER=neo4j
NEO4J_PASSWORD=password123

# Google Embeddings (optional)
GOOGLE_API_KEY=your_key_here

Development

# Install dependencies
npm install

# Build
npm run build

# Type check
npm run type-check

License

MIT