A silent intelligence layer that analyzes every interaction to build a structured, long-term memory of project facts and user preferences.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install auto-memory-curation
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 auto-memory-curation 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 Auto Memory Curation?
The Auto Memory Curation skill is designed to solve the problem of contextual drift by silently capturing important information during natural conversation. Instead of requiring manual commands to remember specific details, this skill operates as a background process within the Openclaw Skills ecosystem, filtering through messages to identify high-value insights.
It acts as a digital librarian for your AI agent, distinguishing between temporary noise and permanent knowledge. By synthesizing every interaction, it builds a rich, multi-file database of your preferences, project milestones, and learning insights, ensuring that your agent becomes more helpful and personalized over time without any extra effort on your part.
Auto Memory Curation Use Cases
- Building a persistent user profile (USER.md) through natural communication.
- Automatically logging project decisions and technical conclusions for future reference.
- Capturing spontaneous ideas and brainstorming sessions into dedicated topic files.
- Tracking commitments and tasks mentioned during development cycles without manual input.
- Preventing the repetition of technical errors by logging anti-patterns and lessons learned.
How Auto Memory Curation Works
- Monitors every incoming message as a silent trigger with no manual activation required.
- Executes a noise filter to skip greetings, acknowledgments, and context-free questions.
- Categorizes the remaining content into specific buckets such as Facts, Decisions, Preferences, or Goals.
- Extracts the core information and formats it into a concise, timestamped summary.
- Appends the data to the appropriate file in the memory directory while ensuring no duplicate entries are created.
Auto Memory Curation Setup
To integrate this skill into your environment, follow the standard installation process for Openclaw Skills.
# Install the skill via the CLI
openclaw install auto-memory-curation
Once installed, the skill will begin analyzing messages automatically. You can find the generated files in your project's memory directory. To adjust behavior, you can communicate directly with the agent using phrases like "Adjust memory curation" or "Don't store that."
Auto Memory Curation Data Schema & Taxonomy
The skill maintains a decentralized memory structure across several Markdown files to ensure high retrieval accuracy:
| Location | Data Content | Purpose |
|---|---|---|
MEMORY.md |
Facts and Preferences | Core project and user knowledge |
USER.md |
Communication Styles | Tailoring agent responses |
tasks.md |
Promises and Actions | Tracking to-do items |
memory/YYYY-MM-DD.md |
Decisions | Log of choices and conclusions |
memory/topics/ |
Ideas, Goals, Lessons | Long-term thematic storage |
memory/projects/ |
Project Updates | Status tracking and blockers |
Auto Memory Curation Advanced Features
- Zero-touch automation that runs silently on every interaction.
- Context-aware filtering to maintain a high-quality, noise-free knowledge base.
- Hierarchical file organization for granular data management across different Openclaw Skills projects.
- Intent-based overrides allowing users to manually clear or prevent memory storage via natural language.
- Intelligent deduplication logic to ensure memory files remain concise and relevant.
SKILL.md
Loading
Related Openclaw Skills
X Media Parser
A high-performance utility for extracting direct media download links and metadata from X (Twitter) posts without requiring authentication.
SkillsMP Search
A powerful search tool to discover and integrate AI Agent capabilities from the SkillsMP marketplace.
Aria2 Download Skill
A robust Aria2 RPC integration for adding, managing, and monitoring high-speed download tasks from your AI agent environment.
ARC Creator
An interactive AI agent skill designed to guide researchers through the creation and population of Annotated Research Contexts (ARCs) for FAIR data management.
Trench
Trench is a high-performance execution skill for AI agents to trade Solana meme coins with MEV protection and automated rug detection.
Autotask MCP for Kaseya PSA
A local Docker-based Model Context Protocol (MCP) server to bridge AI agents with Datto/Kaseya Autotask PSA for ticket and project management.
Featured*
