A lightweight, SQLite-powered memory system for AI agents that handles long-term storage, semantic retrieval, and natural memory decay.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install amarin-memory
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 amarin-memory 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 Amarin Memory?
Amarin Memory provides a robust, persistent adaptive memory layer designed specifically for autonomous agents within the Openclaw Skills ecosystem. Unlike simple key-value stores, it uses sqlite-vec for high-performance semantic search and implements biological-inspired features like temporal decay, where less relevant information fades over time unless reinforced. This ensures your agents maintain a relevant and manageable context without needing a massive external vector database.
By integrating this skill, developers can give their agents a consistent identity and the ability to recall user preferences or past interactions across multiple sessions. As an essential part of the Openclaw Skills collection, Amarin Memory handles automatic deduplication and novelty scoring, ensuring that only the most valuable and unique information is prioritized during retrieval.
Amarin Memory Use Cases
- Storing user preferences and long-term habits across multiple agent sessions.
- Retrieving relevant context from past conversations to improve response accuracy.
- Maintaining a stable agent identity and persona through core memory blocks.
- Optimizing agent performance by automatically decaying unimportant or outdated information.
- Implementing multi-agent memory isolation using specific tags and blocks within the Openclaw Skills framework.
How Amarin Memory Works
- The agent identifies a valuable piece of information or context during a conversation.
- The information is stored using a CLI command, where it is automatically deduplicated against existing memories.
- An embedding is generated via an external service like Ollama and stored in a local SQLite database for semantic search.
- When the agent needs to recall information, it performs a search that combines semantic similarity with an importance score.
- Periodic maintenance cycles run a decay function to lower the importance of unused memories, keeping the memory pool efficient.
Amarin Memory Setup
To get started with this skill from the Openclaw Skills library, ensure you have Python 3.11+ installed.
First, install the package using your preferred manager:
uv pip install amarin-memory
Next, initialize the local database and vector index:
python3 scripts/setup.py
Ensure an embedding service like Ollama is running, and if necessary, set your environment variable:
export OLLAMA_URL="http://localhost:11434"
Amarin Memory Data Schema & Taxonomy
Amarin Memory organizes data within a local SQLite database located at ~/.amarin/agent.db to keep Openclaw Skills data portable and private.
| Component | Description |
|---|---|
| Memories | Main storage table containing the text, embeddings, and importance scores. |
| Tags | Metadata used to categorize and filter memories during retrieval. |
| Core Blocks | Static, high-priority information like agent persona or user bio. |
| Importance | A scale from 0.0 to 1.0 reflecting the significance of the data. |
| Decay Factor | A temporal value that reduces importance over time for non-protected memories. |
Amarin Memory Advanced Features
- Automatic deduplication using a 0.85 cosine similarity threshold to prevent redundant storage.
- Novelty-based surprise scoring that boosts the importance of unique information.
- Manual memory protection to prevent critical data from being affected by temporal decay.
- Semantic KNN search powered by sqlite-vec for lightning-fast retrieval within the Openclaw Skills environment.
- Soft-delete functionality allowing for memory forgetting with a recorded reason for future auditing.
SKILL.md
Loading
Related Openclaw Skills
Apple Silicon AI Fleet Manager
A high-performance local AI inference engine that aggregates the compute power of M-series Macs into a unified, scalable cluster.
Automation Suggestion
An intelligent system that analyzes manual task patterns to propose high-value automations without manual intervention or nagging.
Amazon PPC
A comprehensive framework for architecting and optimizing Amazon Sponsored Ads based on margin-driven logic and tiered campaign structures.
Amazon SEO
A comprehensive optimization framework for Amazon product listings designed to increase organic rank and conversion rates through strategic keyword placement.
Loyalty Designer
A strategic framework for designing margin-positive loyalty programs that increase customer lifetime value through points, tiers, and referral mechanics.
n8n Workflow Management
A comprehensive management tool for n8n to programmatically create, test, and optimize automated workflows.
Featured*
