An offline vector memory tool using ChromaDB and BGE-small-zh for secure, semantic long-term storage within Openclaw Skills.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install local-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 local-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 Local Memory?
The Local Memory skill provides a robust, privacy-first alternative to cloud-based memory systems. By leveraging ChromaDB and the BGE-small-zh-v1.5 embedding model, this Openclaw Skills extension enables AI agents to perform semantic search, long-term information retention, and precise memory management completely offline. It allows developers to maintain high data privacy while ensuring their agents have access to a rich historical context across different sessions without relying on external APIs.
Local Memory Use Cases
- Storing critical project facts or user preferences for long-term agent retrieval in Openclaw Skills.
- Performing semantic searches across historical conversation logs to find specific context.
- Managing and pruning AI memory by deleting specific data points or obsolete information.
- Building offline-capable RAG (Retrieval-Augmented Generation) workflows where data security is a priority.
How Local Memory Works
- Initialize the environment by running the setup script to download the BGE-small-zh model and necessary dependencies.
- Store new information via the memory_store script, which generates vector embeddings for the text and saves them to a local ChromaDB instance.
- Query the local database using semantic search via the memory_recall script to retrieve the most relevant historical context based on a query string.
- Manage the lifecycle of memories by removing specific entries using either a unique ID or a semantic similarity match.
Local Memory Setup
To begin using this skill within Openclaw Skills, run the installation script to download models and dependencies (takes 5-10 minutes):
python scripts/setup.py
Data is stored automatically in the data/ subdirectory. To store a memory, use:
python scripts/memory_store.py --text "content to remember" --category fact --importance 0.8
Local Memory Data Schema & Taxonomy
The skill organizes data using a vector collection in ChromaDB. Each entry is structured as follows:
| Field | Type | Description |
|---|---|---|
| text | String | The primary content of the memory |
| category | Enum | Classification: fact, preference, decision, entity, or other |
| importance | Float | Priority score from 0 to 1 |
| id | String | Unique identifier for precise management |
Local Memory Advanced Features
- Fully offline semantic search using BGE-small-zh-v1.5 embeddings optimized for Chinese and English.
- Granular memory categorization for filtered retrieval within the Openclaw Skills framework.
- Importance-based weighting to prioritize high-value information during recall.
- Dual-mode deletion support allowing for ID-based precision or semantic-based removal of obsolete data.
SKILL.md
Loading
Related Openclaw Skills
Webhook Debugger
A local development utility to receive, inspect, and replay webhook payloads for seamless API integration testing.
Zhou Hongyi AI Insight Skill
A comprehensive knowledge and protocol skill providing deep insights into Zhou Hongyi's AI strategy and the Openclaw ecosystem.
LiblibAI Image Generation Skill
A comprehensive AI image generation skill that leverages the LiblibAI platform for professional text-to-image and image-to-image workflows.
macOS Automator Services Collection
A comprehensive suite of 5 macOS Automator workflows for high-speed file processing and image manipulation directly from the Finder context menu.
Historical Weather AI Agent Skill
A specialized skill for AI agents to query detailed historical weather records including temperature, wind, and air quality for specific cities and dates.
HarmonyOS NEXT Expert
A specialized skill providing comprehensive, offline documentation and expert guidance for HarmonyOS NEXT development including ArkTS, ArkUI, and NDK.
Featured*
