A comprehensive three-tier memory architecture combining vector search, structured local notes, and workspace file indexing for AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install triple-memory
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 triple-memory 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 Triple Memory System?
The Triple Memory System is a sophisticated framework designed to provide AI agents with deep, persistent context retention across multiple sessions. By utilizing this combination of Openclaw Skills, developers can ensure their agents never lose track of user preferences, architectural decisions, or project-specific technical details. The system integrates three distinct backends—LanceDB for high-speed automated conversation recall, Git-Notes for local, branch-aware structured data, and direct file-based searching for indexing workspace documentation like MEMORY.md.
This holistic approach solves the common problem of context drift in LLM-based workflows. Whether you are managing complex coding tasks or tracking long-term user requirements, these Openclaw Skills provide the infrastructure necessary for an agent to operate with a human-like memory of past interactions and project states.
Triple Memory System Use Cases
- Maintaining persistent project context and architectural decisions across different developer sessions.
- Branch-specific memory management to prevent context contamination between features.
- Automated extraction of user preferences and facts during natural language interactions.
- Localized, private storage of sensitive decisions without relying on external API calls via Git-Notes.
- Deep indexing of project documentation and daily logs for rapid information retrieval.
How Triple Memory System Works
- The LanceDB auto-recall mechanism searches for and injects relevant conversation memories into the prompt before the agent responds.
- The agent processes the current task while having access to the three-tier memory architecture.
- Upon completion, LanceDB auto-capture identifies and stores new preferences or facts discussed during the session.
- Git-Notes extracts structured decisions and entities, tagging them with importance levels and associating them with the current git branch.
- The agent or user updates physical workspace files such as MEMORY.md and daily logs to maintain a human-readable, long-term knowledge repository.
Triple Memory System Setup
To get started with these Openclaw Skills, first configure the LanceDB plugin in your agent configuration file:
{
"plugins": {
"slots": { "memory": "memory-lancedb" },
"entries": {
"memory-lancedb": {
"enabled": true,
"config": {
"embedding": { "apiKey": "${OPENAI_API_KEY}", "model": "text-embedding-3-small" },
"autoRecall": true,
"autoCapture": true
}
}
}
}
}
Next, install the Git-Notes component using the CLI:
clawdhub install git-notes-memory
Finally, copy the scripts/file-search.sh utility to your project workspace to enable file-based indexing.
Triple Memory System Data Schema & Taxonomy
The Triple Memory System organizes data across several layers to ensure maximum retrieval efficiency:
| Location | Data Type | Description |
|---|---|---|
MEMORY.md |
Curated Knowledge | The primary source of long-term project truth. |
memory/active-context.md |
Session State | Tracks the current goals and immediate status of the agent. |
memory/YYYY-MM-DD.md |
Daily Logs | A chronological record of work performed and decisions made. |
Git-Notes |
Structured Entities | Branch-aware storage for specific technical decisions and entity metadata. |
LanceDB |
Vector Embeddings | Semantic storage of past conversation snippets for automatic recall. |
Triple Memory System Advanced Features
- Importance Weighting: Assign flags from Critical to Low to control how aggressively certain memories are surfaced.
- Branch Isolation: Ensure that memories recorded on one git branch do not interfere with work on another, perfect for multi-feature development.
- Silent Operation: The system is designed to perform memory writes and updates without interrupting the user or cluttering the chat logs.
- Entity Extraction: Automatically identifies and tags topics, names, and concepts within your structured notes for better organization.
- Automated Lifecycle: Seamlessly handles the transition from conversation (vector) to decision (structured) to documentation (file).
SKILL.md
Loading
Related Openclaw Skills
Triple Memory System
A sophisticated three-tier memory architecture that integrates vector databases, structured git-based notes, and local file search for total context persistence.
Browser Ladder
A cost-optimized web automation framework that escalates from simple fetches to premium cloud browsers only when necessary.
SupaSkills
A comprehensive library of over 1,000 quality-scored expert skills designed to give AI agents deep domain-specific expertise.
Vercel React Composition Patterns
A professional framework for building flexible React components using compound patterns and state decoupling to eliminate boolean prop proliferation.
Social Media Ops
A comprehensive multi-agent framework that scaffolds a specialized seven-agent team to manage multi-brand social media operations with automated workflows and approval gates.
iOS Dev & App Store Release (China)
An end-to-end workflow guide for developing, building, and submitting iOS applications specifically tailored for the China App Store environment.
Featured*
