An infinite, perfectly organized long-term memory system that complements your agent's built-in memory with unlimited categorized storage.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install 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 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 Memory?
The Memory skill is a robust data management framework designed to give your AI agent an organized, persistent storage layer that goes far beyond standard context windows. While traditional agent memory is often ephemeral or messy, this addition to your Openclaw Skills library creates a parallel system in a dedicated directory where information is meticulously categorized and indexed.
By separating long-term storage from immediate working memory, the Memory skill ensures that your agent can manage thousands of files, project histories, and technical documents without losing performance. It functions as a Personal Knowledge Management (PKM) system tailored specifically for agentic workflows, allowing for high-density information retrieval through a structured markdown-based architecture.
Memory Use Cases
- Tracking detailed project histories and multi-phase development roadmaps over months or years.
- Building extensive contact networks and CRM profiles with full interaction context.
- Maintaining a permanent log of architectural or business decisions and the reasoning behind them.
- Developing deep domain-specific knowledge bases and reference materials that grow over time.
- Managing complex collections, inventories, or research data that require strict taxonomic organization.
How Memory Works
- The system initializes a dedicated storage directory at ~/memory/ which remains separate from the agent's internal workspace memory.
- Users define custom storage categories during the setup phase to align the memory structure with their specific workflow needs.
- The agent writes information to specific markdown files immediately upon receiving significant data to ensure no context is lost.
- Each category maintains an INDEX.md file that acts as a map, allowing the agent to navigate large datasets without reading every file.
- As the volume of data grows, the skill implements a scale-by-splitting logic, creating subcategories to keep indices small and search speeds high.
- To retrieve data, the agent performs a search via grep or follows the index hierarchy to locate the precise file needed.
Memory Setup
To integrate this capability into your suite of Openclaw Skills, initialize the environment by following these steps:
- Launch the skill and access the setup guide to define your initial memory structure:
# Review the setup requirements and configuration
cat setup.md
- Collaborate with the agent to identify your primary categories (e.g., projects, people, decisions).
- Use the provided templates to establish the root index and category folders:
# Check the available templates for consistent data entry
cat memory-template.md
- If necessary, configure the one-way sync to move high-value data from the agent's built-in memory into this permanent storage system.
Memory Data Schema & Taxonomy
The Memory skill organizes data using a hierarchical markdown structure optimized for both human readability and machine parsing:
| Component | Path | Purpose |
|---|---|---|
| Root Index | ~/memory/INDEX.md |
The master directory of all available data categories. |
| Configuration | ~/memory/config.md |
Stores system preferences and sync settings. |
| Categories | ~/memory/[category]/ |
User-defined folders (e.g., /projects, /knowledge). |
| Category Index | ~/memory/[category]/INDEX.md |
A tabular list of items within a category for fast navigation. |
| Data Files | ~/memory/[category]/{item}.md |
Individual entries containing detailed notes or logs. |
| Sync Folder | ~/memory/sync/ |
Optional storage for one-way mirrored data from built-in memory. |
Memory Advanced Features
- Scalable architecture that supports splitting large categories into sub-indices once they exceed 100 entries.
- One-way synchronization modules that protect built-in memory while archiving important context to the permanent store.
- High-speed navigation patterns that allow agents to find information in datasets containing 500+ files using hierarchical indices.
- Seamless interoperability with other Openclaw Skills such as the decision-tracking (decide) and learning modules.
- Maintenance protocols for weekly and monthly archiving to ensure the storage remains lean and performant.
SKILL.md
Loading
Related Openclaw Skills
Meilisearch
A comprehensive skill for AI agents to deploy, tune, and manage Meilisearch instances for fast and relevant search experiences.
Meditate
A proactive reflection skill that enables AI agents to generate sandboxed insights and strategic questions during idle time.
Medicine
A sophisticated medical reasoning tool for clinical decision support, patient education, and evidence-based research.
Meals
A comprehensive AI-driven meal planning and grocery management system for your personal workspace.
Mercado Libre
A comprehensive marketplace assistant for Mercado Libre designed to automate product research, price comparison, and selling operations.
Message Communication & Tone Calibration
A sophisticated messaging layer for AI agents that ensures platform-aware formatting, tone matching, and safe escalation for high-stakes communication.
METADATA
Featured*
