Eternal Memory

Overview

This skill provides an eternal memory system for AI agents. It combines local files, vector databases, and cloud storage to ensure memories persist forever and can be semantically searched.

Architecture

Memory Layers

1. Daily Logs (memory/YYYY-MM-DD.md)

   • Raw conversation logs
   • Automatic daily capture
   • Timestamped entries

2. Long-term Memory (MEMORY.md)

   • Curated, distilled memories
   • Important events, decisions, preferences
   • Manually and auto-updated

3. Vector Store

   • Semantic embeddings for all memories
   • Fast similarity search
   • ChromaDB (local) + Supabase (cloud)

Quick Start

Setup

# Install dependencies
source venv/bin/activate
pip install chromadb supabase sentence-transformers

# Initialize memory database
python3 scripts/setup_memory.py

Basic Usage

# Store a memory
python3 scripts/store_memory.py "User prefers Python over JavaScript"

# Search memories
python3 scripts/search_memory.py "programming languages"

# Auto-curate from daily logs
python3 scripts/curate_memory.py --date 2026-02-02

Features

1. Auto-Capture

Automatically captures conversations and important context to daily logs.

2. Semantic Search

Search memories by meaning, not just keywords. Uses embeddings to find relevant context.

3. Auto-Curation

Periodically reviews daily logs and updates MEMORY.md with important insights.

4. Privacy-First

• Local-first architecture (ChromaDB)
• Optional cloud backup (Supabase)
• All data encrypted

5. Export/Import

Easily export all memories to JSON or import from backups.

Storage Options

Local (ChromaDB)

• Free, unlimited storage
• Fast semantic search
• No internet required
• Data stays on your machine

Cloud (Supabase)

• Free tier: 500MB database
• Accessible from anywhere
• Automatic backups
• PostgreSQL + pgvector

Resources

scripts/

• setup_memory.py: Initialize memory databases
• store_memory.py: Store new memories
• search_memory.py: Semantic search across memories
• curate_memory.py: Auto-curate from daily logs
• export_memory.py: Export all memories to JSON

references/

• chromadb_guide.md: ChromaDB setup and usage
• supabase_setup.md: Supabase configuration
• embedding_models.md: Guide to choosing embedding models

assets/

• schema.sql: Supabase database schema
• config.template.json: Configuration template

Integration with OpenClaw

This skill integrates with OpenClaw's existing memory system:

• Reads from memory/YYYY-MM-DD.md and MEMORY.md
• Enhances memory_search with vector search
• Provides backup and export functionality

Next Steps

1. Run setup script to initialize databases
2. Configure Supabase (optional, for cloud backup)
3. Let the AI auto-capture memories during conversations
4. Use heartbeats to auto-curate daily logs into MEMORY.md

---

Build eternal memory. Never forget. 🧠✨