🧠 Subsystem Deep Dive

memdir/

📁 memdir/ 📄 7 files

Tutorial: memdir

This project implements a persistent, file-based memory system that allows an AI assistant to recall user preferences and project context across sessions. It uses a Two-Tier Architecture—keeping a lightweight index (MEMORY.md) in context while retrieving detailed topic files on demand—to manage token limits efficiently. The system enforces strict security sandboxing and distinguishes between Private (user-specific) and Team (shared) scopes, ensuring the AI respects boundaries while maintaining temporal freshness to avoid citing outdated code.

flowchart TD A0["Two-Tier Storage Architecture (Index vs. Detail)"] A1["Contextual Recall Mechanism"] A2["Structured Memory Taxonomy"] A3["Scoped Persistence (Private vs. Team)"] A4["Path Security and Sandboxing"] A5["Temporal Freshness & Verification"] A3 -->|"Locates storage for"| A0 A2 -->|"Defines content format for"| A0 A0 -->|"Supplies file manifest to"| A1 A5 -->|"Flags stale data during"| A1 A4 -->|"Enforces boundaries on"| A3

Chapters

  1. Structured Memory Taxonomy
  2. Scoped Persistence (Private vs. Team)
  3. Two-Tier Storage Architecture (Index vs. Detail)
  4. Contextual Recall Mechanism
  5. Temporal Freshness & Verification
  6. Path Security and Sandboxing

Generated by Code IQ

Files in this section

01_structured_memory_taxonomy.md
Chapter 1: Structured Memory Taxonomy
Welcome to the memdir project! If you are new here, you might be wondering: "How do we stop an AI from remembering usel…
02_scoped_persistence__private_vs__team_.md
Chapter 2: Scoped Persistence (Private vs. Team)
In the previous chapter, Structured Memory Taxonomy, we learned what the AI is allowed to remember (User, Feedback, Pro…
03_two_tier_storage_architecture__index_vs__detail_.md
Chapter 3: Two-Tier Storage Architecture (Index vs. Detail)
In the previous chapter, Scoped Persistence (Private vs. Team), we solved the problem of who sees a memory by splitting…
04_contextual_recall_mechanism.md
Chapter 4: Contextual Recall Mechanism
In the previous chapter, Two-Tier Storage Architecture (Index vs. Detail), we learned how to keep the AI's "backpack" l…
05_temporal_freshness___verification.md
Chapter 5: Temporal Freshness & Verification
In the previous chapter, Contextual Recall Mechanism, we built a "smart librarian" that finds the right memory file for…
06_path_security_and_sandboxing.md
Chapter 6: Path Security and Sandboxing
In the previous chapter, Temporal Freshness & Verification, we learned how to trust the content of a memory by checking…