A robust long-term memory system for AI agents using MongoDB and Voyage AI for semantic search and persistent context.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-memory-skill
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 openclaw-memory-skill 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 OpenClaw Memory?
OpenClaw Memory is a sophisticated plugin designed to provide AI agents with a permanent, searchable memory bank. By leveraging MongoDB for storage and Voyage AI for semantic embeddings, it allows agents to move beyond transient conversation windows and recall specific decisions, user preferences, and project insights across multiple sessions. This skill is a high-value addition to the Openclaw Skills library for developers building context-aware applications.
The system goes beyond simple text storage by understanding meaning through vector embeddings. It integrates deeply with the agent's lifecycle through automated hooks, ensuring that important facts are captured without manual intervention. Whether you are tracking a complex database schema decision or a user's coding style preference, OpenClaw Memory ensures your agent remains informed and consistent over time, making it one of the most powerful Openclaw Skills for long-term project management.
OpenClaw Memory Use Cases
- Recalling technical decisions and architecture patterns from previous development sessions.
- Persisting user-specific preferences like programming language choices or UI framework favorites.
- Building a searchable knowledge base of domain-specific facts discovered during research tasks.
- Maintaining session continuity by bootstrapping the agent with relevant historical context upon startup.
- Automatically capturing insights and project learnings without manual user prompting.
How OpenClaw Memory Works
- The agent identifies a fact, decision, or preference worth remembering during a live conversation.
- Data is sent to the memory daemon, where Voyage AI generates semantic embeddings for the content to enable vector search.
- The information is stored in a MongoDB collection with associated metadata and tags for targeted retrieval.
- When relevant context is needed, the memory_search tool performs a semantic query to find the most similar records based on meaning rather than just keywords.
- Automated lifecycle hooks inject discovered memories back into the agent's active context window to guide future responses and maintain continuity.
OpenClaw Memory Setup
To get started with this entry in the Openclaw Skills collection, follow these installation steps:
# Install the memory plugin
openclaw plugins install openclaw-memory
# Navigate to the plugin directory
cd openclaw-memory
# Install dependencies and start the memory daemon
pnpm install && pnpm dev:daemon
You can also run the system using Docker by executing docker compose up -d. Ensure you have MongoDB 8.0+ running and your Voyage AI API key configured in your ~/.openclaw/openclaw.json file.
OpenClaw Memory Data Schema & Taxonomy
OpenClaw Memory uses a structured approach to organize agent knowledge, ensuring interoperability with other Openclaw Skills. Below is the primary metadata structure:
| Field | Description |
|---|---|
| text | The core content or insight of the memory record. |
| tags | Categories like preference, decision, or project names used for filtering. |
| score | Semantic similarity ranking (0.0 to 1.0) returned during searches. |
| ttl | Time-to-live for memory expiration, allowing for temporary or permanent storage. |
| createdAt | ISO timestamp of when the memory was captured for temporal tracking. |
OpenClaw Memory Advanced Features
- Web-based dashboard for visualizing memory graphs and relationship mapping in real-time.
- Nine-stage reflection pipeline for conflict resolution, confidence scoring, and duplicate detection.
- Semantic relationship tracking using graph edges such as SUPPORTS, CONTRADICTS, and DERIVES_FROM.
- Automated context enrichment that appends relevant memories to tool outputs like Bash or Grep results.
- Lifecycle hooks including auto-remember and memory-bootstrap for zero-touch context management.
SKILL.md
Loading
Related Openclaw Skills
NetPad
A robust toolset for programmatically managing web forms, data submissions, and fine-grained organizational access control.
Find Skills for ClawHub
A discovery and management tool designed to find and install specialized capabilities from the official ClawHub skill registry.
Model Router Premium
An intelligent request router that selects the most cost-effective LLM based on task complexity and model metadata.
Exchange Rates (XE.com)
A headless browser-based tool to fetch real-time mid-market currency exchange rates from XE.com with automatic API fallback.
Voyage AI
A comprehensive CLI-based skill for Voyage AI embeddings, reranking, and MongoDB Atlas Vector Search integration.
Rednote (Xiaohongshu) Automation Skill
A robust automation tool for searching, extracting, and interacting with the Xiaohongshu (Little Red Book) platform via browser automation.
Featured*
