A sophisticated memory lifecycle management system for AI agents featuring Bayesian decay and automated quality gates.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-guardian-agent
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-guardian-agent using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is memory-guardian?
The memory-guardian skill is a robust framework designed to manage the lifecycle of an AI agent's workspace memory. By utilizing a dual-layer Bayesian decay model and a four-state quality gate, it ensures that your agent's knowledge base remains relevant, compact, and high-quality. This Openclaw Skills utility prevents memory bloat and maintains the integrity of references across storage files and associated directories, allowing for smarter and more context-aware agent interactions through precise lifecycle control.
memory-guardian Use Cases
- Maintaining long-term workspace memory for coding agents using Openclaw Skills.
- Automatically pruning stale or low-importance information through Bayesian decay.
- Diagnosing and fixing memory bloat in large-scale agent projects.
- Managing complex case lifecycles from ingestion to retirement or archival.
- Ensuring consistent memory quality across multi-agent workflows.
How memory-guardian Works
- The system performs a status check using memory_status to verify the integrity of references and meta-information.
- New information is ingested via memory_ingest, where importance and tags are assigned.
- A dual-layer signal loop tracks memory access via logs and daily notes to inform the decay engine.
- Scheduled maintenance runs via run_batch to sync file changes and apply Bayesian decay to outdated memories.
- The quality gate monitors metrics like retirement rates and case signals to ensure workspace health.
- Compaction processes are triggered when file sizes exceed defined thresholds to keep the storage optimized.
memory-guardian Setup
To install the memory-guardian Openclaw Skills package, ensure your MG_WORKSPACE environment variable is set. You can run the initial setup and verify the installation with the following commands:
# Check the current status and verify file integrity
python3 scripts/memory_guardian.py status --workspace /path/to/workspace
# Manually trigger a maintenance batch
python3 scripts/memory_guardian.py run
Ensure a cron task is created to handle daily maintenance via the run_batch tool.
memory-guardian Data Schema & Taxonomy
| Component | Description |
|---|---|
| MEMORY.md | The primary compressed storage for active memories and case summaries. |
| memory/ | Directory containing detailed memory files and historical data. |
| meta.json | Central metadata store tracking decay, scores, and quality gate states. |
| access_log.jsonl | Append-only log of memory access events for signal processing. |
| MG_WORKSPACE | Environment variable defining the root of the managed memory system. |
memory-guardian Advanced Features
- Dual-layer Bayesian decay engine for sophisticated information pruning.
- Four-state quality gate to manage case lifecycles including active, frozen, retired, and ignored states.
- Automated compaction logic that triggers based on MEMORY.md file size thresholds.
- L3 confirmation and manual review hooks for high-stakes memory retirement and audit.
- Quiet degradation support where health checks auto-adjust based on signal freshness.
- Integration with cron for fully automated Openclaw Skills maintenance loops.
SKILL.md
Loading
Related Openclaw Skills
Product Vision and Strategy Assessment
A comprehensive diagnostic tool that evaluates product vision, strategy, and principles to ensure teams operate as missionaries with a focused, customer-obsessed direction.
Memory Optimization for Python
A specialized skill designed to diagnose memory bottlenecks and transform Python code into high-efficiency, low-footprint applications.
Anjuke Automation Manager
An end-to-end automation skill designed to manage Anjuke real estate accounts, from property publishing to AI-driven description optimization.
Router9 AI Skill
A comprehensive multi-tool integration providing transcription, text-to-speech, computer vision, and cloud storage via a unified CLI.
Concrete Language Rewriter
The Concrete Language Rewriter converts abstract, theoretical, or jargon-heavy drafts into vivid, sensory language that sticks in the reader's mind.
Value Testing Technique Selection
A strategic framework for selecting the right testing methodology to validate if customers will actually choose, buy, and use a product before investing in full-scale development.
Featured*
