Skip to content

5-memory-integration.md

Latest commit

 

History

History
75 lines (53 loc) · 2.55 KB

5-memory-integration.md

File metadata and controls

75 lines (53 loc) · 2.55 KB
Raw

MCP Setup Guide: Memory System Integration

How Memory MCP integrates with our five-component memory system

🔗 When to Use Memory MCP vs Wiki-Links

Use Wiki-Links For:

  • Document navigation - Clicking between [[memory-human]], [[memory-ai]], etc.
  • Static references - Links that don't change often
  • Human reading flow - When you want linear, narrative connections

Use Memory MCP For:

  • Dynamic relationships - Connections that evolve over time
  • Complex associations - Multi-way relationships between concepts
  • Searchable knowledge - Finding patterns across all memory
  • AI analysis - When AI needs to understand relationship networks

🛠️ Simple MCP Patterns

Create Key Entities

# Add major project concepts to knowledge graph
mcp_memory_create_entities([
  {name: "CEIS-NIA-2025", entityType: "project", observations: ["Net Impact Analysis implementation for 2025"]},
  {name: "TWaNG-methodology", entityType: "method", observations: ["Propensity score weighting approach"]},
  {name: "Phase-1-complete", entityType: "milestone", observations: ["Memory hub established with wiki-links"]}
])

Track Decision Evolution

# Connect decisions over time  
mcp_memory_create_relations([
  {from: "Memory-System-Evaluation", to: "Phase-1-Implementation", relationType: "led-to"},
  {from: "Phase-1-Implementation", to: "Wiki-Link-Network", relationType: "created"},
  {from: "Wiki-Link-Network", to: "Enhanced-Navigation", relationType: "enables"}
])

Search Knowledge Network

# Find all project decisions
mcp_memory_search_nodes("decision implementation phase")

# Find methodology connections  
mcp_memory_search_nodes("TWaNG methodology propensity score")

🎯 Best Practices

Keep It Simple

  • Use MCP for relationships, wiki-links for navigation
  • Don't duplicate - if it's in [[project-memory]], just reference it in MCP
  • Focus on concepts and connections, not detailed content

Complement VS Code

  • MCP for discovery ("What connects to X?")
  • Wiki-links for navigation ("Go to document Y")\
  • VS Code search for content ("Find all mentions of Z")

Human-Friendly Workflow

  1. Document in [[memory-human]] or [[memory-ai]]
  2. Navigate with wiki-links
  3. Connect concepts with Memory MCP when needed
  4. Search with VS Code or MCP depending on need

This approach keeps each tool focused on what it does best while integrating with our five-component memory system.