/optimize-memory-bank

Optimize Memory Bank for Redundancy

Safety: Create Backup

CRITICAL: Before any changes:

cp -r .memory_bank .memory_bank.backup.$(date +%Y%m%d_%H%M%S)

Backup created at: .memory_bank.backup.YYYYMMDD_HHMMSS

Optimization Process

1. Load redundancy patterns ONCE (reuse for all files):

   • Load common redundancy patterns and detection rules
   • Prepare optimization strategies for identified patterns

2. Scan files: Find all .memory_bank/**/*.md files using Glob

3. For EACH file (sequential, no subagents):

   a. Read and analyze:

   • Read file content
   • Count lines (before_lines)
   • Report: 🔍 Checking [filename] (X lines)...

   b. Check redundancy:

   • Apply redundancy pattern detection
   • Calculate redundancy percentage

   c. Optimize if needed:

   • If redundancy >10%:
     • Apply optimization fixes for detected patterns
     • Preserve unique content (never lose information)
     • Target: 30-50% reduction
     • Overwrite file with optimized version
     • Count new lines (after_lines)
     • Calculate reduction: ((before - after) / before) * 100
     • Report: ✅ Optimized [filename]: X → Y lines (-Z%)
   • If redundancy ≤10%:
     • Keep original unchanged
     • Report: ✅ [filename] already optimal

   d. Track metrics: Add to running totals (files_optimized, total_lines_saved)

4. Final report: Show summary with before/after metrics

Expected Results

Typical outcomes:

• 30-50% reduction in verbose files
• Preserved content quality
• Improved consistency
• Faster context loading

Report example:

✓ Scanned 28 files
✓ Optimized 12 files (43%)
✓ Saved 1,850 lines (38% average reduction)
✓ Backup: .memory_bank.backup.20251114_143022

Top improvements:
  - tech_stack.md: 648 → 301 lines (-54%)
  - ai-agent-handbook.md: 404 → 251 lines (-38%)
  - architecture.md: 520 → 310 lines (-40%)