A local vector database and semantic memory system specifically optimized for Chinese language processing in AI agent workflows.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install evrmem
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 evrmem 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 evrmem?
evrmem is a specialized local memory system designed to provide AI agents with a persistent and searchable knowledge base. By leveraging the text2vec-base-chinese embedding model and ChromaDB, it enables high-performance semantic search and storage capabilities directly on the local machine. This tool is essential for developers working with Openclaw Skills who need to implement context-aware agents that can recall past interactions, project-specific notes, and technical findings without relying on external cloud services.
The system is built for privacy and speed, offering a robust foundation for Retrieval-Augmented Generation (RAG). It transforms raw text into vector embeddings, allowing agents to find relevant information based on meaning rather than just keyword matches. Whether you are building a personal assistant or a technical documentation bot, this skill ensures your agent has a reliable and scalable long-term memory.
evrmem Use Cases
- Storing and retrieving technical insights or bug fixes across different development sessions.
- Implementing long-term memory for AI agents to remember user preferences and past conversations.
- Building a local, searchable knowledge base from Markdown notes or project documentation.
- Enhancing LLM prompts with relevant context using RAG-based retrieval to minimize hallucinations.
- Categorizing and querying information using project-specific tags and metadata filtering.
How evrmem Works
- Initialize the system to download and configure the local Chinese embedding models and vector database.
- Add text content to the memory system, which automatically generates vector embeddings and stores them in ChromaDB with metadata.
- Execute semantic searches using natural language queries to find the most relevant memories based on vector distance.
- Utilize the RAG workflow to fetch contextually related snippets and inject them into agent prompts.
- Manage the memory repository using the CLI or Python API to filter by project, tag, or similarity scores.
evrmem Setup
To get started with this skill, install the package and initialize the local environment:
pip install evrmem
evrmem init
For users in regions with restricted access to HuggingFace, use the mirror endpoint before initialization:
export HF_ENDPOINT=https://hf-mirror.com
evrmem init
You can further customize the behavior by editing the ~/.evrmem/config.yaml file to specify local model paths or device settings.
evrmem Data Schema & Taxonomy
The system organizes data into a structured format for efficient retrieval within the Openclaw Skills ecosystem:
| Attribute | Description | Data Type |
|---|---|---|
| Content | The primary text data or memory snippet | String |
| Project | Identifier for grouping related memories | String |
| Tags | Labels for granular categorization and filtering | List/String |
| Distance | Semantic similarity score (lower is closer) | Float |
| Metadata | Additional context such as timestamps or IDs | JSON Object |
evrmem Advanced Features
- Multi-device support allowing for inference on either CPU or CUDA-enabled GPUs.
- Full CLI interface for managing memories, checking system statistics, and performing manual queries.
- Air-gapped and offline support through local model path configurations and network-disable flags.
- Customizable similarity thresholds to ensure only highly relevant data is retrieved during RAG operations.
- Seamless integration hooks for Python-based agents to automate memory storage during active workflows.
SKILL.md
Loading
Related Openclaw Skills
ConsolePatrol
ConsolePatrol is an AI-first diagnostic tool that automatically scans web applications for console errors and framework-specific warnings.
Credential Auditor
An automated security auditing tool for credential matching, dictionary generation, and multi-protocol vulnerability testing.
Linux Command Guard Elite
A defense-in-depth Linux command safety skill for AI agents using allowlist-first policies and approval gates.
buddy-bills: Smart AI Expense Tracker
buddy-bills is an AI-powered accounting skill that uses natural language to track personal and family expenses, manage fixed costs, and generate detailed financial summaries.
Agent Swarm Intelligence Core
A sophisticated multi-agent orchestration framework enabling collective intelligence through coordination, self-organization, and emergent behavior.
Palliative Care Video
An AI-powered video assistant designed to help palliative care programs educate patients and overcome hospice misconceptions through high-quality short clips.
Featured*
