A sophisticated multi-layered memory architecture for AI agents that integrates vector search, Git-based knowledge graphs, and write-ahead logging.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install elite-longterm-memory-lobster
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-lobster 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 comprehensive framework designed to solve the problem of context loss in AI agent interactions. By implementing a six-layer architecture, it provides agents like Clawdbot and Moltbot with a robust way to store and retrieve information. This system is a vital component for developers building high-performance Openclaw Skills that require reliable data retention, semantic search capabilities, and cross-session knowledge persistence.
The framework moves beyond simple chat history by utilizing the Write-Ahead Log (WAL) protocol, ensuring that every decision, preference, and task state is recorded before the agent responds. This approach makes your Openclaw Skills resilient to crashes, token window compactions, and session restarts, creating a seamless experience for end-users.
Elite Longterm Memory Use Cases
- Retaining user preferences and project decisions across multiple distinct chat sessions.
- Reducing token consumption by extracting and storing facts rather than full conversation histories.
- Syncing a shared knowledge base across different agent instances or devices.
- Managing complex, long-running workflows where maintaining state durability is critical.
- Creating a version-controlled, human-readable audit trail of an agent's technical decisions.
How Elite Longterm Memory Works
- The system initiates the Hot RAM layer via a SESSION-STATE.md file, which tracks active tasks using the Write-Ahead Log protocol.
- Semantic search is performed against the LanceDB Warm Store to automatically inject relevant historical context based on the current user query.
- Facts and concrete preferences are extracted during the conversation and stored with high importance scores in the vector database.
- Permanent decisions and structured knowledge are committed to the Cold Store using a Git-Notes based knowledge graph for versioned persistence.
- Periodically, the system runs hygiene routines to consolidate daily logs into a curated, human-readable MEMORY.md file.
Elite Longterm Memory Setup
1. Initialize Local State
Create the initial session file in your workspace root:
cat > SESSION-STATE.md << 'EOF'
# SESSION-STATE.md — Active Working Memory
## Current Task
[None]
## Key Context
[None]
EOF
2. Configure Vector Storage
Update your config to enable semantic search for your Openclaw Skills:
{
"memorySearch": {
"enabled": true,
"provider": "openai",
"sources": ["memory"]
}
}
3. Initialize Git-Notes
Setup the cold storage layer for your project:
git init
python3 skills/git-notes-memory/memory.py -p . sync --start
Elite Longterm Memory Data Schema & Taxonomy
The skill organizes memory across several layers to balance retrieval speed and durability:
| Layer | Storage Method | Data Type |
|---|---|---|
| Hot RAM | Markdown (SESSION-STATE.md) | Active tasks, blockers, and immediate context |
| Warm Store | LanceDB Vectors | Semantic memories, user preferences, and facts |
| Cold Store | Git-Notes | Technical decisions and structured knowledge graphs |
| Archive | Markdown (memory/ folder) | Daily logs and human-readable distilled wisdom |
| Cloud | SuperMemory API | Cross-device synchronized knowledge base |
Elite Longterm Memory Advanced Features
- Strict Write-Ahead Log (WAL) protocol to guarantee context durability before every agent response.
- Integrated Mem0 support for automatic fact extraction, reducing prompt overhead by up to 80%.
- Branch-aware context persistence using Git-Notes, allowing memory to follow code versioning.
- Customizable importance thresholds to prevent memory pollution from low-value interactions.
- Automated maintenance hygiene for de-duplicating facts and consolidating long-term archives.
SKILL.md
Loading
Related Openclaw Skills
TriMet Transit CLI
A command-line interface for accessing real-time Portland TriMet transit data, including arrivals, trip routing, and system alerts.
Trakt.tv CLI Integration
A CLI-driven skill to access Trakt.tv watch history and search for media content via AI agents.
Todoist
A powerful integration for managing Todoist tasks and projects directly through an AI agent via the TypeScript CLI.
Steam Games CLI
A powerful command-line interface for browsing, filtering, and discovering games within your Steam library based on playtime, ratings, and compatibility.
Xiaohongshu Operations Agent (xiaohongshu-ops)
A comprehensive AI agent skill for automated Xiaohongshu operations, from account positioning to viral content execution.
Coordinate Meeting (Meetlark)
An AI-native scheduling poll tool that eliminates back-and-forth by coordinating meeting times between humans and agents.
Featured*
