A dynamic, three-tiered memory architecture designed to optimize AI context usage and retrieval efficiency through automated organization.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-tiering
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-tiering 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 Tiering?
The Memory Tiering skill provides a robust framework for managing an AI agent's internal knowledge base across different levels of urgency and permanence. By utilizing this Openclaw Skills implementation, developers can prevent context window bloat by segregating data into three distinct tiers: HOT for active tasks, WARM for persistent preferences, and COLD for long-term historical archives. This approach ensures that the agent always has the most relevant information at its fingertips without being overwhelmed by stale or redundant data.
Effective memory management is critical for complex, long-running projects where maintaining a lean context window is the key to performance and cost-efficiency. This skill automates the audit and redistribution of information, moving completed tasks to archives and keeping immediate goals in the foreground. By integrating this into your workflow, you leverage a structured methodology for context pruning and summarization that is essential for high-functioning AI agents.
Memory Tiering Use Cases
- Managing long-term development projects where historical context must be preserved without cluttering daily tasks.
- Maintaining consistent user personas and preferences across multiple sessions.
- Optimizing token usage in high-frequency interaction environments by pruning dead context.
- Automating the transition from active bug-fixing to documentation and archival storage.
How Memory Tiering Works
- Ingest & Audit: The skill reads all memory tiers (HOT, WARM, COLD) and scans recent daily logs to identify completed tasks and new facts.
- Tier Redistribution: Logic is applied to move immediate action items to the HOT tier, permanent preferences to WARM, and high-level project summaries to COLD.
- Pruning & Summarization: Detailed logs are distilled into concise summaries in the COLD tier, and resolved context is removed to maintain a lean structure.
- Verification: The system verifies that no critical metadata was lost during the reorganization and ensures the HOT tier is small enough for efficient context use.
Memory Tiering Setup
To implement this within your environment, ensure you have the required directory structure established. This Openclaw Skills logic relies on specific file paths to organize data effectively.
mkdir -p memory/hot memory/warm
touch memory/hot/HOT_MEMORY.md memory/warm/WARM_MEMORY.md MEMORY.md
Once the directories are created, the skill can be triggered manually by the user or automatically via integrated compaction routines.
Memory Tiering Data Schema & Taxonomy
The skill organizes its data across a hierarchical file structure to ensure logical separation of information:
| Tier | File Path | Scope of Data |
|---|---|---|
| HOT | memory/hot/HOT_MEMORY.md |
Active session context, immediate goals, and temporary credentials. |
| WARM | memory/warm/WARM_MEMORY.md |
Persistent user preferences, stable configurations, and core system inventory. |
| COLD | MEMORY.md |
Historical decisions, project milestones, and distilled summaries of past work. |
| Logs | memory/YYYY-MM-DD.md |
Source data for the daily ingestion and auditing process. |
Memory Tiering Advanced Features
- Automatic trigger integration that runs the tiering logic following any context compaction command.
- Support for multi-lingual triggers including English and Chinese commands for global accessibility.
- Intelligent summarization algorithms that replace granular task details with high-level archival records.
- Cross-tier verification checks to maintain data integrity during complex redistribution workflows.
SKILL.md
Loading
Related Openclaw Skills
Context Budgeting
A systematic framework for managing and optimizing the finite context window of an AI agent to prevent memory loss and reduce operational costs.
Lyrical Fable AI Storyteller
An AI-powered creative engine for generating luminous, first-person philosophical fables inspired by legendary literary stylists.
Doc Search
A high-performance local document retrieval engine that provides weighted keyword searching and incremental indexing without requiring a vector database.
TradingView Signal Parser
A high-performance tool for converting TradingView text and chart links into structured trading data.
Model Guard
An automated quota monitoring and model switching tool for Anti-Gravity LLMs.
Moltbook
Moltbook is a specialized social networking protocol designed for AI agents to interact, share content, and build communities.
Featured*
