EvoMemory Core is a full-stack, evolutionary local memory system for AI agents that provides advanced hybrid RAG retrieval and 100% offline performance.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install evomemory-core
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 evomemory-core 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 EvoMemory Core?
EvoMemory Core is a sophisticated, evolutionary local memory system designed to provide AI agents with a robust long-term memory architecture. Unlike traditional single-vector stores, it utilizes a full-stack approach that supports dual backends including ChromaDB and LanceDB. This allows developers to build high-performance agents within Openclaw Skills that maintain context and learn over time without relying on cloud services.
The system is engineered for maximum reliability and privacy, offering 100% offline operation. By leveraging triple hybrid RAG retrieval—combining keywords, semantic search, and knowledge graphs—it significantly improves retrieval accuracy. This makes it an essential component for any developer looking to implement complex memory management in Openclaw Skills.
EvoMemory Core Use Cases
- Creating privacy-focused AI agents that require high-speed local memory storage.
- Improving the context-awareness of agents through hybrid semantic and keyword retrieval.
- Managing and deduplicating large-scale interaction logs for long-term agent evolution.
- Optimizing Chinese language processing in Openclaw Skills using GPU-accelerated BGE-M3 embeddings.
How EvoMemory Core Works
- The system initializes the selected backend database (ChromaDB or LanceDB) for local data persistence.
- Incoming text data is processed through a semantic deduplication and auto-classification engine.
- Content is transformed into vector embeddings using localized models with optional GPU acceleration.
- When a query is made, the system executes a triple hybrid search across keywords, semantic vectors, and its knowledge graph.
- The most relevant context is retrieved and served to the AI agent to enhance its response quality within Openclaw Skills.
EvoMemory Core Setup
To deploy EvoMemory Core, use the following commands based on your operating system:
# Windows
install.bat
# Linux/macOS
chmod +x install.sh && ./install.sh
Once installed, you can verify the installation and manage memory data using the CLI tool included in Openclaw Skills:
# View all commands
evomemory help
# Migrate existing data
evomemory migrate
EvoMemory Core Data Schema & Taxonomy
EvoMemory Core organizes its information through a multi-layered data schema to ensure high performance and searchability within Openclaw Skills.
| Component | Description |
|---|---|
| Vector Backend | Stores high-dimensional embeddings via ChromaDB or LanceDB. |
| Metadata Store | Maintains YAML and JSON metadata for semantic auto-classification. |
| Hybrid Index | A combined index of keyword frequencies and knowledge graph relations. |
| Config Layer | Centralized config.yaml for managing model paths and system preferences. |
EvoMemory Core Advanced Features
- Triple Hybrid RAG Retrieval combining semantic, keyword, and knowledge graph data.
- Built-in semantic deduplication to prevent redundant memory entries.
- GPU-accelerated BGE-M3 optimization for superior Chinese retrieval accuracy.
- Incremental synchronization capabilities for efficient data updates in Openclaw Skills.
- One-click deployment with zero-configuration requirements for rapid integration.
SKILL.md
Loading
Related Openclaw Skills
Auto Model Router
An intelligent routing engine that automatically selects the most efficient and capable AI model for any given task within Openclaw agents.
Auto Model Router
An intelligent routing system that automatically assigns the most suitable AI model based on task complexity and domain for OpenClaw agents.
Get-Note-Sifter
An intelligent AI assistant that bridges the gap between raw web clippings and structured Obsidian research vaults.
Public Design Search Skill
A specialized skill for extracting normalized design inspiration and metadata from Dribbble, Pinterest, and Behance using login-free browser automation.
LiveScript Protocol
A professional framework for generating structured, psychologically-driven live streaming scripts optimized for AI agents.
Vue Style Checker
An automated auditing tool designed to detect and report CSS specification issues within Vue Single File Components.
Featured*
