memory-maintenance

⚠Review·Scanned 2/18/2026

This skill automates review, archiving, and cleanup of agent memory files and suggests updates to MEMORY.md. It runs shell scripts (e.g., review.sh, apply.sh), reads GEMINI_API_KEY/sources .env, and invokes gemini --model ..., enabling external LLM calls and filesystem move/delete operations.

from clawhub.ai·v1.0.0·37.6 KB·0 installs
Scanned from 1.0.0 at 275ca5b · Transparency log ↗
$ vett add clawhub.ai/maxlauriehutchinson/memory-maintenanceReview findings below

Memory Maintenance Skill

Intelligent memory management for OpenClaw agents.

What It Does

  1. Reviews daily notes → Suggests MEMORY.md updates
  2. Checks directory health → Finds fragmentation, naming issues
  3. Auto-cleans → Archives old reviews, enforces retention

Quick Start

# Install
bash scripts/install.sh

# Run review
bash scripts/review.sh

# Check status
bash scripts/status.sh

# Apply changes (after review)
bash scripts/apply.sh --safe YYYY-MM-DD

Recommended Model

  • Primary: gemini-2.5-flash (fast, efficient)
  • Fallback: gemini-2.5-flash-lite (if needed)

Architecture

Uses LLM for review, runs on cron schedule, human approval for content changes, auto-cleanup for maintenance tasks.

Safety

  • Content suggestions → Never auto-applied
  • Safe maintenance (archiving) → Auto with --safe
  • Risky operations → Require --all + confirmation
  • Trash recovery → Deleted files recoverable

Files

scripts/
  review.sh       # Main daily review
  apply.sh        # Apply approved changes
  cleanup.sh      # Archive & retention
  status.sh       # Check system health
  install.sh      # One-time setup

config/
  settings.json   # User configuration

SKILL.md          # Full documentation

Author

Max Hutchinson — Building AI agent infrastructure.

License

MIT