A mandatory core framework that enables AI agents to achieve a closed-loop system of memory-driven decision making, continuous self-evolution, and proactive action.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install integrated-memory-evolution-action
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 integrated-memory-evolution-action 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 Integrated Memory Evolution Action?
The Integrated Memory Evolution Action skill is a foundational protocol designed for advanced AI agents. It integrates a sophisticated three-layer memory architecture with a self-evolution engine and proactive action layers. By utilizing this framework, agents move beyond simple reactive responses to become autonomous entities that learn from every interaction, manage complex task states via a Write-Ahead Log (WAL) protocol, and maintain a permanent knowledge base of user preferences and system ontologies. This is an essential component for any developer building robust agentic workflows with Openclaw Skills.
At its core, this skill ensures that every agent interaction contributes to a long-term intelligence cycle. It forces agents to check historical context before acting and to log new insights immediately after execution. This creates a highly resilient system where errors are rarely repeated and the agent's utility grows over time through automated daily and weekly optimization cycles.
Integrated Memory Evolution Action Use Cases
- Managing long-term user preferences and communication styles across multiple sessions to ensure personalized agent behavior.
- Reducing repetitive errors by logging and reviewing error patterns in the Evolution Layer for systematic fixes.
- Maintaining state across complex, multi-step coding or research tasks using the action layer working buffer.
- Automating agent self-improvement through scheduled daily and weekly knowledge promotion cycles.
- Facilitating proactive task execution and resourcefulness in autonomous agent workflows within the Openclaw Skills ecosystem.
How Integrated Memory Evolution Action Works
- The agent performs a mandatory Pre-Task Check by reading across three layers of memory and verifying existing learning or error logs to avoid redundant mistakes.
- During the execution phase, the agent follows the Write-Ahead Log (WAL) Protocol, recording decisions, preferences, and state changes to the memory layers before providing a final response.
- After the dialogue concludes, the agent executes a Post-Dialogue Write process to archive conversation summaries to Layer 1 and update specific task statuses in Layer 2.
- The system triggers daily evolution cycles (at 00:00) to extract insights from conversation logs into the Evolution Layer for broader accessibility.
- Weekly reviews (Sundays at 23:00) promote temporary learnings into permanent system rules and optimize the agent ontology graph and core tool definitions.
Integrated Memory Evolution Action Setup
To deploy this skill, ensure your agent environment has access to the shared memory and workspace directories defined in the Openclaw Skills standards.
# 1. Initialize the directory structure for the three-layer memory system
mkdir -p shared/memory/layer1-conversations shared/memory/layer2-events/tasks shared/memory/layer3-permanent
# 2. Setup the evolution layer directories
mkdir -p workspace/.learnings workspace/memory/ontology
# 3. Create initial state tracking files
touch workspace/SESSION-STATE.md workspace/HEARTBEAT.md workspace/memory/working-buffer.md
# 4. Verify script availability for evolution logging
ls scripts/log-learning.mjs scripts/ontology.py
Integrated Memory Evolution Action Data Schema & Taxonomy
The skill organizes data across a hierarchical directory and metadata taxonomy to ensure rapid retrieval and high context density:
| Layer | Scope | Key Data Files |
|---|---|---|
| Memory L1 | Instant Context | layer1-conversations/YYYY-MM-DD/agent.md (Daily summaries) |
| Memory L2 | Event Tracking | layer2-events/tasks/Txxx.md, errors/, skills/ |
| Memory L3 | Permanent Rules | layer3-permanent/user-profile.md (Habits, API configs, principles) |
| Evolution | Intelligence | .learnings/LEARNINGS.md, .learnings/ERRORS.md, ontology/graph.jsonl |
| Action | Active Execution | SESSION-STATE.md (Current task), memory/working-buffer.md (Retry logs) |
Integrated Memory Evolution Action Advanced Features
- WAL (Write-Ahead Log) Protocol: Guarantees memory integrity by requiring data persistence before the agent issues a response.
- Multi-Layer Memory Archiving: Seamlessly transitions data from immediate dialogue summaries to permanent user profiles and skill optimizations.
- Self-Evolution Loops: Automated daily and weekly routines that promote task-specific learnings into system-wide core rules.
- Proactive Resourcefulness: A built-in action logic that requires agents to attempt at least 10 different approaches to a problem before requesting human intervention.
- Ontology Knowledge Graphing: Uses JSONL format to map complex relationships between tasks, entities, and skills for advanced reasoning within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Pet Renaissance Portrait Generator
A specialized AI tool that transforms pet photos and descriptions into museum-quality Renaissance oil paintings using high-end image generation APIs.
Polymarket Candle Momentum
A data-driven trading skill for Polymarket crypto markets using candle body analysis and Binance volume surges.
Sleep Channel Tracker
A deterministic sleep tracker that turns Discord channels into precise, metadata-grounded sleep logs.
Breaking News Event Trader
A sophisticated automated trading skill that monitors premium news feeds to execute high-impact Polymarket trades based on real-time events.
Whale Alert Monitor
A sophisticated blockchain surveillance tool for real-time tracking of whale wallet activities, large-scale transfers, and exchange liquidity flows.
Polymarket AI Divergence
An automated trading skill that detects price discrepancies between Simmer AI consensus and Polymarket prices to execute edge-based trades.
Featured*
