A persistent memory framework that allows AI agents to remember facts, track entities, and learn from past experiences across sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-memory-1
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-1 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 AgentMemory Skill?
AgentMemory Skill provides a robust foundation for building long-term context in AI workflows. By utilizing Openclaw Skills, developers can enable agents to maintain a continuous state, ensuring that important information is never lost between sessions. This skill focuses on data durability, allowing agents to retrieve historical data and improve their performance autonomously based on previous outcomes and interactions.
At its core, this tool transforms stateless LLM interactions into stateful experiences. Whether you are building a personal assistant or a complex automation workflow, integrating this into your Openclaw Skills library ensures that your agents grow more intelligent and personalized over time by managing facts, lessons, and entity relationships within a local, high-performance database.
AgentMemory Skill Use Cases
- Initializing agent sessions by loading relevant historical context and lessons to minimize redundant queries.
- Persisting durable facts extracted from user conversations to ensure the agent maintains a consistent persona and knowledge base.
- Recording lessons learned after task failures to optimize future autonomous actions and avoid repeating mistakes.
- Managing complex entity relationships, such as tracking project roles, user preferences, or technical specifications across different sessions.
How AgentMemory Skill Works
- Initialize the memory system using the Openclaw Skills framework to establish a local SQLite database connection.
- Store information using specialized methods like remember for general facts or learn for experience-based insights.
- Track distinct entities such as people, projects, or tools with specific metadata to maintain a structured knowledge graph.
- Query the persistent database during active sessions to retrieve context-aware data using semantic recall or tag-based filtering.
- Update the memory state at the conclusion of every interaction to ensure continuous learning and updated entity information.
AgentMemory Skill Setup
To get started with this skill within the Openclaw Skills ecosystem, install it via the CLI:
clawdhub install agent-memory
You can then initialize the memory system in your Python environment. If you require a specific storage location, configure the database path manually:
from src.memory import AgentMemory
mem = AgentMemory(db_path="/your/custom/path/memory.db")
AgentMemory Skill Data Schema & Taxonomy
The AgentMemory Skill organizes data into three primary structures within its persistent database to optimize retrieval within Openclaw Skills:
| Component | Description | Metadata Captured |
|---|---|---|
| Facts | Durable information and general context | content, tags, timestamp |
| Lessons | Experience-based learning from actions | action, context, outcome, insight |
| Entities | People, projects, or specific objects | name, type, role/attributes |
All data is stored locally by default at ~/.agent-memory/memory.db to ensure privacy and low-latency access.
AgentMemory Skill Advanced Features
- Automated memory protocols for Clawdbot integration via standard AGENTS.md or HEARTBEAT.md configurations.
- Outcome-based filtering to retrieve only positive or negative experiences for fine-tuning agent behavior.
- Flexible metadata tagging for custom semantic search queries across large datasets.
- Cross-session entity tracking to maintain complex relationship maps within the Openclaw Skills framework.
- Custom database pathing for self-hosted or containerized agent deployments.
SKILL.md
Loading
Related Openclaw Skills
Email Sequence Skill
An AI-powered copywriting engine designed to architect and write multi-stage email marketing campaigns with strategic timing and mobile-first formatting.
Knowledge Organizer
A sophisticated content ingestion engine that turns articles and notes into structured Markdown with automated syncing to Obsidian, Feishu, and Tencent IMA.
PDFMonkey
A high-performance skill for automating PDF document generation from dynamic templates using the PDFMonkey API and Membrane protocol.
Book Sum
A powerful utility to generate non-truncated, comprehensive book summaries including deep-dive narratives on key concepts.
Pixel Familiar Identity
A comprehensive AI agent identity and SOUL.md framework designed to power revenue-focused autonomous business agents.
Auto-Data-Analyzer
An automated data analysis tool that transforms raw CSV and Excel files into comprehensive HTML reports with cleaning, statistics, and visualizations.
Featured*
