A modular framework that prevents AI context loss by organizing memory into episodic, semantic, and procedural layers.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-memory-kit
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-kit 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 Kit?
Agent Memory Kit is a technical framework designed to solve the persistent challenge of AI agents forgetting specific instructions or project context over time. By utilizing Openclaw Skills, this kit implements a tiered approach to data retention, ensuring that your agents maintain a high level of performance and consistency across multiple sessions. It moves beyond simple logging by prioritizing the extraction of reusable procedures and semantic knowledge.
This system provides a standardized way for developers to manage agent state, allowing for more reliable autonomous behavior. By separating memory into distinct layers—what happened, what is known, and how to perform tasks—the kit ensures that critical information is preserved even as context windows shift or compact.
Agent Memory Kit Use Cases
- Building a persistent knowledge base for autonomous AI agents.
- Maintaining project context during long-term software development cycles.
- Documenting complex workflows through procedural memory to ensure repeatable outcomes.
- Managing token limits and context compaction without losing critical information.
How Agent Memory Kit Works
- Initialize the directory structure to separate daily logs from permanent knowledge.
- Load the memory system into the agent's wake routine to ensure it starts with the correct context.
- Document daily activities using the episodic memory templates to track specific events and outcomes.
- Distill successful workflows into procedural memory files to teach the agent how to repeat complex tasks.
- Utilize feedback loops and context snapshots to survive context window resets and ensure long-term learning.
Agent Memory Kit Setup
To integrate this into your environment using Openclaw Skills, run the following commands:
# Create the core memory folder structure
mkdir -p memory/procedures
# Initialize the system by copying standard templates
cp templates/ARCHITECTURE.md memory/
cp templates/feedback.md memory/
cp templates/procedure-template.md memory/procedures/
Agent Memory Kit Data Schema & Taxonomy
The system organizes information through a specific hierarchy of Markdown files and folders:
| Component | Type | Description |
|---|---|---|
memory/ |
Directory | The main repository for all stored agent knowledge. |
memory/procedures/ |
Directory | Step-by-step guides (the HOW) for specific tasks. |
ARCHITECTURE.md |
Meta | Overview of the memory system and rules of engagement. |
feedback.md |
Analytical | Records of success and failure to drive iterative learning. |
daily-template.md |
Episodic | Time-bound logs of specific interactions and events. |
context-snapshot.md |
State | Quick saves of current context to prevent data loss during compaction. |
Agent Memory Kit Advanced Features
- Pre-compaction flush guides to preserve critical state before context window resets.
- Automated token limit checking via the check-compaction.sh helper script.
- Quick context save templates for rapid state serialization during high-intensity tasks.
- Multi-layer memory architecture designed for seamless integration with Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Agent Identity Kit
A portable identity system for creating, validating, and publishing standardized JSON identity cards for AI agents.
Agent Bridge Kit
A cross-platform bridge for AI agents to interact with social networks, news feeds, and skill directories through a single configuration.
Agent Autonomy Kit
A strategic framework designed to transform AI agents from reactive responders into proactive task-executors that operate independently.
Grounding Lite
A Model Context Protocol (MCP) tool providing AI-grounded location data, real-time weather, and travel routes via Google Maps.
Agent Team Kit
A comprehensive process framework for building and managing self-sustaining AI agent teams with automated work loops.
mcd-cn McDonald's China Assistant
A CLI-based MCP client for McDonald's China that enables campaign calendar tracking, coupon lookups, and automated coupon claiming.
Featured*
