A sophisticated multi-layered memory architecture for AI agents, optimized for Openclaw Skills to maintain persistent context and decision history across sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install elite-longterm-memory-backup
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 elite-longterm-memory-backup 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 Elite Longterm Memory?
Elite Longterm Memory is a professional-grade memory system designed to solve the context window limitations and 'forgetfulness' of modern AI agents. By integrating six distinct memory strategies—including Write-Ahead Logging (WAL), semantic vector search, and structured knowledge graphs—it creates a durable and evolving 'brain' for your development environment. This skill ensures that your AI coding assistants like Cursor, Claude, or ChatGPT retain critical architectural decisions and user preferences.
As a cornerstone of the Openclaw Skills ecosystem, this memory system prioritizes reliability through its multi-tiered approach. It moves beyond simple chat history by organizing data into active working memory, warm semantic stores, and permanent cold archives. This structure allows agents to perform at a higher level of autonomy, reducing repetitive instructions and preventing the loss of work during session compactions or restarts.
Elite Longterm Memory Use Cases
- Maintaining persistent project context and task status across different IDE sessions.
- Recording and retrieving specific user preferences such as framework choices, naming conventions, or UI libraries.
- Tracking complex architectural decisions to prevent the AI from suggesting previously discarded approaches.
- Reducing token consumption by automatically extracting and storing facts instead of raw history.
- Synchronizing agent knowledge across multiple devices or team members via cloud backup integrations.
How Elite Longterm Memory Works
- Context Loading: Upon session start, the agent reads the SESSION-STATE.md file for immediate tasks and performs a semantic search across Openclaw Skills memory sources to recall relevant history.
- Write-Ahead Logging: Before responding to any user prompt that involves a decision or new fact, the agent updates its hot memory to ensure durability against crashes.
- Semantic Retrieval: The system uses LanceDB to perform vector searches, automatically injecting relevant historical context into the prompt based on the current conversation topic.
- Structured Archiving: Permanent decisions and technical learnings are committed to a Git-Notes based knowledge graph, providing a version-aware historical record.
- Log Distillation: At the conclusion of a session, the agent summarizes the day's progress into a curated MEMORY.md file and generates specific daily logs for future reference.
Elite Longterm Memory Setup
To implement this advanced memory architecture for Openclaw Skills, follow these installation and configuration steps:
- Initialize the session state and directory structure:
# Create the hot memory file
cat > SESSION-STATE.md << 'EOF'
# SESSION-STATE.md — Active Working Memory
## Current Task
[None]
EOF
# Create the long-term storage directory
mkdir -p memory
- Configure semantic search in your
~/.openclaw/openclaw.jsonfile:
{
"memorySearch": {
"enabled": true,
"provider": "openai",
"sources": ["memory"]
}
}
- Initialize the cold storage backend using Git-Notes:
git init
python3 skills/git-notes-memory/memory.py -p . sync --start
- Ensure your
OPENAI_API_KEYis exported in your environment to enable vector embeddings.
Elite Longterm Memory Data Schema & Taxonomy
The system organizes data into a hierarchical taxonomy to optimize for both retrieval speed and long-term durability:
| Layer | Storage Method | Data Type |
|---|---|---|
| Hot RAM | SESSION-STATE.md | Active tasks, blockers, and pending actions |
| Warm Store | LanceDB Vectors | Semantic facts, preferences, and important context |
| Cold Store | Git-Notes | Version-tracked architectural decisions and logic |
| Curated Archive | MEMORY.md | Distilled long-term wisdom and project summaries |
| Daily Logs | memory/YYYY-MM-DD.md | Granular chronological history of session activities |
Elite Longterm Memory Advanced Features
- WAL Protocol integration to guarantee that state is saved before the agent attempts to generate a response.
- Automated fact extraction via Mem0, resulting in up to an 80% reduction in context window token usage.
- Branch-aware memory that tracks context and decisions specific to different git branches within a repository.
- Cross-device synchronization using SuperMemory API for seamless agent transitions between machines.
- Weekly memory hygiene workflows to prune irrelevant vector data and consolidate daily logs into the primary archive.
SKILL.md
Loading
Related Openclaw Skills
Cute Kitten Generator
A high-performance ComfyUI workflow integration for generating and sharing professional-grade animal photography.
Email Analyzer
A precision-engineered IMAP tool for analyzing and bulk-cleaning mailboxes based on customizable keyword logic and strict safety protocols.
163 Email Management Skill
A comprehensive Python-based CLI tool for full 163.com mailbox management, including secure IMAP/SMTP integration and attachment handling.
Leo X-Post
A specialized AI agent skill designed for Leo to automate X (Twitter) posting, media uploads, and replies via Python.
Humanizer: AI Writing Refinement
A sophisticated writing editor that identifies and removes mechanical AI-generated patterns to produce natural, human-sounding text.
Memory Crystal
A private, sovereign memory layer that enables AI agents to capture, search, and share context across all Openclaw Skills and development tools.
Featured*
