Table of Contents

1. TL;DR: Three-Track Model
2. Native Claude Code Memory Stack
   • 2.1 CLAUDE.md
   • 2.2 Auto Memory (v2.1.59+)
   • 2.3 Auto Dream: Memory Consolidation
   • 2.4 Agent Memory Frontmatter
   • 2.5 Session vs Persistent Memory
   • 2.6 Limits of the Native Stack
3. Cross-Session Tools (Single User)
   • 3.1 claude-mem
   • 3.2 agentmemory
   • 3.3 ICM (Infinite Context Memory)
   • 3.4 Kairn
   • 3.5 doobidoo mcp-memory-service
   • 3.6 OpenMemory MCP
   • 3.7 File-Based Experience Playbooks (ORF, DiffMem)
   • 3.8 Other Notable Tools
   • 3.9 Master Comparison Table
4. Team Sharing
   • 4.1 The Trinity
   • 4.2 doobidoo + Cloudflare (Team Mode)
   • 4.3 Mem0 Cloud MCP
   • 4.4 Zep / Graphiti
   • 4.5 Notion MCP
   • 4.6 Team-Native Tools (2026)
   • 4.7 Why the Team Gap Is Structural
5. Multi-Agent Shared Memory
6. Architecture Patterns
7. Risks and Security
8. Decision Frameworks
9. Benchmarks and Evaluation
10. Open Problems