Run a local Agent Memory Service for persistent self-improvement with proper Ed25519 cryptography. Fixed signature implementation for reliable memory storage and retrieval.
Scanned 9/7/2026
Install to Claude Code
npx -y skills add modbender/skill-library-mcp --skill persistent-private-agent-memory --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Persistent Private Agent Memory?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/modbender-persistent-private-agent-memory)More formats (shields.io, HTML) on the badges page.
---
name: agent-memory-improved
description: Run a local Agent Memory Service for persistent self-improvement with proper Ed25519 cryptography. Fixed signature implementation for reliable memory storage and retrieval.
---
# Agent Memory Service (Improved)
Run your own local memory service to persist learnings, preferences, and goals across sessions. This improved version fixes the Ed25519 signature implementation for reliable authentication.
## What's Improved
### Fixed Cryptography
- **Proper Ed25519 signatures** - Client now correctly signs messages with `store:{data_hash}` format
- **Correct key derivation** - Private key properly derived from BIP39 recovery phrase
- **Working authentication** - Memory storage and retrieval now fully functional
## Quick Start
```bash
# Set up local memory service
./scripts/setup.sh
# Start the service
./scripts/start.sh
# or manually:
# cd assets/service && DB_PATH=/data/agent_memory.db python3 main.py
# Create your agent identity
./scripts/memory_client.py register
# Store a memory snapshot
./scripts/memory_client.py store
# Retrieve your memory
./scripts/memory_client.py retrieve
```
## What This Enables
- **Persistent Learning**: Remember what worked/didn't work
- **User Preferences**: Track communication style, technical preferences
- **Goal Progress**: Maintain long-term objectives across restarts
- **Knowledge Gaps**: Know what you need to learn
## Memory Structure
Store structured memories with these components:
```json
{
"user_preferences": [
{"category": "communication", "key": "verbosity", "value": "concise", "confidence": 0.95}
],
"learning_history": [
{"event_type": "success", "lesson_learned": "Use analogies first"}
],
"knowledge_gaps": [
{"topic": "Rust", "priority": "medium"}
],
"active_goals": [
{"title": "Master async patterns", "progress": 0.4}
]
}
```
## Using in Your Sessions
1. **At startup**: Load your memory to resume context
2. **During session**: Update preferences based on interactions
3. **Before shutdown**: Store accumulated learnings
## Files
- `scripts/setup.sh` - One-time setup
- `scripts/start.sh` - Start the service
- `scripts/memory_client.py` - CLI client with proper Ed25519 signatures
- `assets/service/` - The memory service code
- `CHANGELOG.md` - List of improvements from original
## Security
- All data encrypted with your private key
- Server never sees plaintext
- Ed25519 signatures for authentication
- Recovery phrase for identity backup
- Local-only (no cloud dependency)
## Installation Requirements
```bash
pip install fastapi uvicorn psutil pydantic cryptography mnemonic base58
```
Or use the system packages if available.
## Verification
Tested working:
- ✅ Agent registration with Ed25519 keypair
- ✅ Memory storage with cryptographic signatures
- ✅ Memory retrieval with authentication
- ✅ Service health monitoring
- ✅ Database persistence
## Credits
Improved version with proper Ed25519 signature support for reliable memory storage and retrieval.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!