A sophisticated 3-tier memory management system for AI agents to capture, promote, and retrieve long-term information.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-memory-journal
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-journal 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 Journal?
The Agent Memory Journal is a specialized tool within the Openclaw Skills ecosystem designed to provide AI agents with a reliable long-term memory. It utilizes a hierarchical 3-tier structure—Hot, Warm, and Cold—to balance immediate context needs with massive historical data retrieval. By using Openclaw Skills like this one, developers can ensure their agents remember strategic decisions, user preferences, and critical constraints across multiple sessions.
At its core, the skill offers a robust Python API for deep integration and a streamlined CLI for automated workflows. It moves beyond simple logging by allowing agents to curate episodic notes into permanent core memories, ensuring that the most relevant information is always available to the LLM without cluttering the context window.
Agent Memory Journal Use Cases
- Storing strategic decisions and project constraints to maintain consistency across long-running development tasks.
- Documenting specific user preferences and agent capabilities to personalize future interactions.
- Recording lessons learned or 'gotchas' during debugging to prevent the agent from repeating past errors.
- Implementing persistent memory for autonomous agents operating in complex, multi-stage environments.
How Agent Memory Journal Works
- Capture episodic observations in real-time using the note command to record immediate findings.
- Promote significant facts to core memory using the remember command, which categorizes information for better retrieval.
- Search existing memories using semantic queries to bring relevant past context back into the agent's current focus.
- Execute the ingest cycle to manage the promotion of data between Hot, Warm, and Cold tiers and rebuild search indexes.
- Use the forget command to archive or supersede outdated information, keeping the memory bank accurate.
Agent Memory Journal Setup
To get started with this memory system in your Openclaw Skills environment, configure your workspace and CLI:
# Create the memory directory structure
mkdir -p .memory
# Configure the hot promotion target
echo '{\"hot_promotion_target\": \"active_context\"}' > .memory/config.json
# Run the integrity check to verify installation
agent-memory-journal doctor
Refer to the agent installation docs for setting up the specific startup contracts and CLI symlinks required for your agent's environment.
Agent Memory Journal Data Schema & Taxonomy
The skill organizes data into a structured taxonomy to facilitate fast access and long-term storage within Openclaw Skills.
| Data Tier | Description | Retention Strategy |
|---|---|---|
| Hot | Items currently in the active context window | High Priority |
| Warm | Searchable indexed memories for semantic retrieval | Medium Priority |
| Cold | Full historical logs and raw episodic data | Archive |
Memories are classified into core categories: decision, constraint, gotcha, preference, and capability.
Agent Memory Journal Advanced Features
- Atomic Pinning: Ensure critical items stay in the Hot tier regardless of age to maintain essential context.
- Automated Curation: Use the promotion cycle to move episodic notes into core memory based on importance.
- Integrity Validation: Built-in doctor commands to ensure the memory files and indexes remain healthy within Openclaw Skills.
- Flexible Integration: Access memory functions via a high-level Python API for complex agentic logic or a CLI for simple shell scripts.
SKILL.md
Loading
Related Openclaw Skills
Village Carpenter Starter
A practical framework for AI agents to guide users through establishing a manual woodworking trade using scavenged materials and essential hand tools.
Village Blacksmith Starter
A comprehensive AI agent skill that guides users through building a coal forge from scrap and establishing a self-sustaining blacksmithing trade.
Docker Security Hardening Tool
A professional-grade API tool for generating CIS Benchmark-compliant Docker security configurations and hardened manifests.
FortiGate Security Hardening Tool
A professional API for generating production-ready, hardened FortiGate configurations aligned with CIS Benchmark security standards.
Mac Health Check
A specialized skill for monitoring real-time thermal, power, and performance telemetry on Apple Silicon Macs.
Japanese Smart Home Command Normalizer
A specialized tool for transforming noisy Japanese speech-to-text transcripts into clean, structured smart home intents.
Featured*
