Memory Defrag 🧠🔧

---
name: memory-defrag
description: AI that reorganizes your notes, finds duplicates, suggests consolidation - keep your second brain clean
homepage: https://github.com/clawdbot
user-invocable: true
command-dispatch: defrag
metadata.clawdbot: {"version": "2.0.0", "hef_generation": 6}
---

Memory Defrag 🧠🔧

> _"A cluttered mind can't think clearly. Neither can cluttered notes."_

Memory Defrag is your AI-powered note janitor. It scans your memory files, finds duplicates and related content, suggests consolidations, and helps reorganize your second brain.

What's New in Gen 6

### 🔗 Reference Graph
- Detects `[[wiki-links]]`, `@mentions`, and file paths
- Maps connections between notes
- Identifies truly isolated content

### 📈 Growth Trends
- Tracks memory growth over time
- Visualizes word count and health trends
- Helps identify documentation patterns

### 🌱 Freshness Decay
- Exponential decay model for content freshness
- Highlights stale areas needing attention
- Smarter orphan detection

### 🔒 Safe Operations
- Automatic backups before any changes
- Easy restore from any backup
- Non-destructive archive system

### 🔍 N-gram Similarity
- Character n-gram matching for better duplicate detection
- Catches rephrased content that keyword matching misses
- Combined with keyword Jaccard for hybrid accuracy

What It Does

### 🔍 Duplicate Detection
- N-gram + keyword similarity (hybrid approach)
- Finds near-duplicates across files
- Groups related content that should be merged

### 📊 Content Analysis
- Maps your knowledge topology
- Identifies orphaned notes (stale + no connections)
- Finds dense clusters (over-documented areas)
- Spots sparse zones (under-documented areas)
- Tracks content freshness with decay model

### 🔗 Reference Connectivity
- Detects `[[wiki-links]]` between notes
- Tracks `@mentions` and file references
- Shows connection graph statistics

### 🗂️ Smart Suggestions
- Merge: Combine similar content
- Archive: Move stale content safely
- Split: Break up oversized files
- Link: Create topic indexes

Commands

Quick Start

User: /defrag scan

Memory Defrag 🧠🔧 Scan Complete

📊 Stats:
• Files scanned: 12
• Sections analyzed: 847
• Total words: 42,350
• Duplicates found: 23
• Clusters detected: 8
• Orphans found: 15

🟢 Health Score: 78/100
🌱 Avg Freshness: 72%

💡 Next Steps:
• /defrag duplicates — see duplicate content
• /defrag suggest — get consolidation ideas

Example Workflow

1. Health Check

User: /defrag health

🏥 Memory Health Report

🟢 Overall Score: 78/100

📊 Stats:
• 847 sections analyzed
• 42,350 total words
• 23 duplicate groups
• 8 topic clusters
• 12 pending suggestions

🌱 Freshness: 72%
🔗 Connectivity: 234/847 sections linked (28%)

⚠️ Issues:
  🟡 23 duplicate sections found
  🟡 15 orphaned notes detected

2. Find Duplicates

User: /defrag duplicates

🔍 Duplicate Content Found

Group 1: 🔗 (87% similar)
  • 2025-01-29.md:234-267 — HEF Evolution Tasks
  • 2025-01-30.md:89-120 — HEF Evolution Log
  • 2025-01-31.md:45-78 — Today's HEF Work
  → Suggested: merge to memory/hef-evolution.md

Group 2: 🔗 (92% similar)
  • 2025-01-28.md:156-198 — N'Ko Vocabulary Notes
  • 2025-01-31.md:892-934 — N'Ko Words Learned
  → Suggested: merge to memory/nko-vocabulary.md

💡 Use /defrag suggest for actionable suggestions

3. Preview & Apply

User: /defrag preview sug_merge_dup_1

👁️ Preview: sug_merge_dup_1

Type: merge
Description: Merge 3 similar sections about 'HEF Evolution'
Target: memory/hef-evolution.md

Would affect:
  • 2025-01-29.md — HEF Evolution Tasks
  • 2025-01-30.md — HEF Evolution Log
  • 2025-01-31.md — Today's HEF Work

Preview:
# Merged: HEF Evolution

## From 2025-01-29.md
### HEF Evolution Tasks
[content would be merged here]

...

💡 Use /defrag apply sug_merge_dup_1 to execute

User: /defrag apply sug_merge_dup_1

✅ Suggestion Applied

Created merged file: memory/hef-evolution.md
🔒 Backup created: 20250131_143022

📝 Original sections preserved. Remove from source files manually after review.

Use /defrag restore 20250131_143022 to undo if needed.

4. Track Trends

User: /defrag trends

📈 Memory Growth Trends

01-25 [████████░░] 38,200w 🟢
01-26 [████████░░] 39,100w 🟢
01-27 [█████████░] 40,500w 🟢
01-28 [█████████░] 41,200w 🟡
01-29 [██████████] 42,350w 🟢
01-30 [██████████] 42,800w 🟢
01-31 [██████████] 43,100w 🟢

📈 Change over 7 days:
• Words: +4,900
• Sections: +42

5. Check References

User: /defrag refs

🔗 Reference Connectivity

Most Connected:
  • MEMORY.md — Main Index (12 refs)
  • projects.md — Active Projects (8 refs)
  • goals.md — 2025 Goals (6 refs)

⚠️ 613 unconnected sections
Consider adding links to integrate these into your knowledge graph.

📊 Overall: 234/847 sections connected (28%)

Configuration

Create `[home-path]`:

{
  "scan_paths": [
    "[home-path]
    "[home-path]
  ],
  "similarity_threshold": 0.70,
  "orphan_days": 30,
  "max_file_kb": 50,
  "freshness_half_life_days": 30,
  "ngram_size": 3
}

Similarity Detection

Uses hybrid approach:
- N-grams (70
- Keywords (30

Freshness Model

Content freshness decays exponentially:
- Half-life: 30 days (configurable)
- Fresh (>70
- Aging (40-70
- Stale (<40

Backup System

All destructive operations create automatic backups:
- Stored in `[home-path]`
- Named with timestamp: `YYYYMMDD_HHMMSS`
- Easy restore: `/defrag restore <backup_id>`

Integration

### With Dream Weaver
- Scans dream journals for duplicates
- Suggests merging related dreams
- Identifies stale dreams for composting

### With MEMORY.md
- Monitors growth trends
- Suggests archiving old entries
- Recommends topic-based splits

### With Daily Memory
- Detects repeated patterns
- Suggests templates
- Identifies routine vs. unique events

Files

Philosophy

> Your second brain should be as organized as you want your first brain to be.
>
> But organization isn't about perfection—it's about findability.
> Memory Defrag doesn't impose structure; it reveals the structure that's already there,
> and suggests improvements that match how you actually think.

---

_Version 2.0.0 — HEF Generation 6_ 🧠🔧