A sophisticated memory lifecycle manager for AI agents based on the Ebbinghaus forgetting curve to handle knowledge retention and decay.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-ebbinghaus
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 memory-ebbinghaus 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 Memory Ebbinghaus?
Memory Ebbinghaus is an advanced utility designed to manage the long-term memory of AI agents by simulating human-like cognitive retention. By applying the Ebbinghaus forgetting curve, this Openclaw Skills integration allows agents to track memory strength decay over time, ensuring that only relevant and reinforced information remains in the active context while stale knowledge is systematically archived or deleted.
The skill provides a structured framework for memory reinforcement, where the stability of a memory item increases with each review. This prevents context bloat and ensures that AI agents maintain a high-quality, high-utility knowledge base. It is an essential tool for developers building persistent AI agents that need to distinguish between fleeting information and long-term technical findings or project history.
Memory Ebbinghaus Use Cases
- Managing agent memory files to prevent context window saturation.
- Implementing spaced repetition workflows for AI long-term knowledge retention.
- Automatically cleaning up stale knowledge and fading project details.
- Identifying which pieces of information require immediate review or reinforcement.
- Archiving historical project data to a persistent markdown-based storage system.
How Memory Ebbinghaus Works
- The skill calculates memory strength using the formula $strength = e^{(-days_elapsed / stability)}$.
- New memory items are added with an initial stability of 1.0 and maximum strength.
- A daily decay command is executed to update the strength values based on elapsed time.
- Memory items are categorized as Active, Decaying, or Fading based on their current strength score.
- Users or agents perform reviews, which reset strength to 1.0 and multiply stability by 1.5 to slow future decay.
- Fading memories are either forgotten (deleted) or moved to a permanent MEMORY.md archive.
Memory Ebbinghaus Setup
To begin using this tool within the Openclaw Skills ecosystem, initialize the database with the following command:
python3 scripts/ebbinghaus.py status
You can configure custom storage paths using environment variables:
EBBINGHAUS_DB=/path/to/memory_db.json \
EBBINGHAUS_ARCHIVE=/path/to/MEMORY.md \
python3 scripts/ebbinghaus.py status
Memory Ebbinghaus Data Schema & Taxonomy
The skill organizes data within a JSON database and a Markdown archive. The schema categorizes memories into specific domains to enhance retrieval logic:
| Category | Description |
|---|---|
| project | Specific project or task completion details |
| tech | Technical findings, code snippets, or solutions |
| person | Contextual memory regarding specific individuals |
| event | Logs of important occurrences or milestones |
| general | Miscellaneous information |
Memory health is tracked via three statuses:
- Green (Active): Strength ≥ 0.7
- Yellow (Decaying): Strength 0.3 - 0.7
- Red (Fading): Strength < 0.3
Memory Ebbinghaus Advanced Features
- Heartbeat Integration: Automated checks that alert the user when critical memories are fading.
- Dynamic Stability Scaling: Exponentially increases memory retention duration through repeated reinforcement.
- Archive Pipeline: Seamlessly moves expired active memories into a structured MEMORY.md file for long-term reference.
- Multi-Category Support: Allows for granular filtering of memory decay based on the type of information stored.
- Silent Logging: Background monitoring of decaying items to minimize user interruption while maintaining data integrity.
SKILL.md
Loading
Related Openclaw Skills
SwiftVerify
An automated verification skill for Swift and SwiftUI projects designed to enforce Apple Design Award quality standards.
Skill Creator: Build
A specialized meta-skill that generates independent project management skills to guide AI agents from initial planning to final delivery.
Multi-Platform Articles (MPA)
A professional automation tool that transforms Markdown files into styled HTML and pushes them directly to WeChat Official Account drafts with one click.
Obsidian Plugin Development
A professional-grade toolkit for building, debugging, and submitting high-quality Obsidian plugins using TypeScript and modern web technologies.
CNInfo Report Downloader
A robust CLI tool for batch-downloading formal A-share financial reports from CNInfo with built-in filtering and resume capabilities.
flightAI
An AI-powered aviation assistant for searching, booking, changing, and refunding domestic flights via Python-driven automation.
Featured*
