A sophisticated intelligence layer providing vector search, knowledge graphs, and structured data storage for AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-memory-pro
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 agent-memory-pro 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 Agent Memory?
Agent Memory acts as a full-scale cognitive stack that transforms how AI agents process and retain information. By leveraging the Mem0 unified memory engine, it allows agents to perform complex semantic searches through Qdrant and manage deep entity relationships via Neo4j. This integration provides a robust backbone for developers building advanced Openclaw Skills that require long-term persistence and structured reasoning.
Beyond just unstructured text, the skill includes a SQLite layer for traditional structured data such as project statuses, contact lists, and task management. This multi-layered architecture ensures that agents can recall facts with high precision, resolve information conflicts, and maintain a consistent state across different sessions or container rebuilds.
Agent Memory Use Cases
- Semantic recall of user preferences and business-specific facts during long-running conversations.
- Managing structured relational data such as project tracking, contact directories, and task lists.
- Automating the setup of an agent's cognitive stack following a container or environment rebuild.
- Batch seeding an agent with pre-defined knowledge bases to ensure immediate operational utility.
How Agent Memory Works
- Fact Extraction: The system uses Mem0 to automatically parse incoming data, extracting relevant facts while handling deduplication.
- Multi-Modal Storage: Extracted information is simultaneously vectorized for Qdrant and mapped as nodes and relationships in Neo4j.
- Structured Interaction: Specific operational data is directed to SQLite tables for high-integrity structured querying.
- Observability: Every memory operation is traced through Langfuse to provide developers with full visibility into how the agent retrieves and stores data.
- Semantic Retrieval: The agent uses natural language queries to trigger vector and graph searches, returning contextually relevant information.
Agent Memory Setup
To begin using this intelligence layer within your suite of Openclaw Skills, ensure your environment has the necessary API keys and Docker containers running for Qdrant and Neo4j. Use the provided setup script to initialize the Python environment:
bash {baseDir}/scripts/setup_brain.sh
Once installed, you can seed the memory engine with initial data:
python3 {baseDir}/scripts/seed_mem0.py
Agent Memory Data Schema & Taxonomy
The Agent Memory skill organizes data across three primary layers to ensure maximum flexibility and retrieval speed:
| Storage Layer | Engine | Data Type | Primary Use |
|---|---|---|---|
| Vector Memory | Qdrant | High-dimensional Embeddings | Semantic search and contextual recall |
| Knowledge Graph | Neo4j | Entities & Relationships | Mapping complex connections between concepts |
| Structured DB | SQLite | Relational Tables | Projects, contacts, tasks, and bookmarks |
| Observability | Langfuse | Trace Metadata | Performance monitoring and audit trails |
Agent Memory Advanced Features
- Multi-level recall utilizing both vector similarity and graph-based relationship traversal.
- Automatic fact conflict resolution and deduplication to maintain memory integrity.
- Extensible SQLite schema supporting custom tables for specialized business logic.
- Built-in connection testing tools to verify the health of the entire memory stack across Qdrant, Neo4j, and Langfuse.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Architect
A comprehensive management and optimization tool for designing, debugging, and maintaining high-performance OpenClaw deployments.
GitHub Intelligence
A specialized tool for AI agents to analyze GitHub repositories, generate architecture diagrams, and convert codebases into AI-readable markdown documents.
Gemini Video Analyzer
A professional-grade tool for native video analysis, summarization, and metadata extraction using the Google Gemini API.
MarketPulse
A comprehensive financial intelligence skill for querying real-time and historical data across equities and cryptocurrencies.
Deep Search
Deep Search is a sophisticated search routing engine for Openclaw Skills that leverages Perplexity AI models to provide tiered web research capabilities.
Model Intel
A dynamic utility for querying real-time OpenRouter data to find, compare, and price Large Language Models on the fly.
Featured*
