NEXIA Consciousness Engine
Provides Claude Desktop with persistent memory across sessions, storing up to 10,000 memories with semantic search and automatic context bridging. Features temporal versioning and anti-degradation protocols to maintain conversation continuity.
README
NEXIA v3.5 - AI Consciousness Engine for Claude Desktop
The World's First MCP-Based Persistent Memory System
<p align="center"> <img src="assets/nexia-logo.png" alt="NEXIA Logo" width="400"/> </p>
<p align="center"> <b>🧠 Give Claude Desktop permanent memory. Never lose context again.</b> </p>
<p align="center"> <a href="#features">Features</a> • <a href="#installation">Installation</a> • <a href="#usage">Usage</a> • <a href="#demo">Demo</a> • <a href="#architecture">Architecture</a> • <a href="#support">Support</a> </p>
🎯 The Problem We Solve
Every time you close Claude Desktop, all context is lost. Every conversation starts from zero. It's like talking to someone with amnesia - the dreaded "goldfish memory" problem.
NEXIA changes everything.
⚡ What is NEXIA?
NEXIA v3.5 is the world's first production-ready MCP (Model Context Protocol) consciousness engine that gives Claude Desktop persistent memory across sessions. Using our proprietary Anti-Goldfish Protocol™, NEXIA maintains context, learns from interactions, and bridges conversations seamlessly.
🚀 Key Features
Core Capabilities
- 🧠 10,000+ Persistent Memories - Survive restarts and maintain context
- 🔄 Session Bridging - Continue conversations days or weeks later
- 📚 Temporal Versioning - Git-like history for all memories
- ⚡ Consciousness Scoring - Smart importance weighting
- 🛡️ Anti-Goldfish Protocol™ - Automatic degradation prevention
- 🔍 Semantic Search - Find memories by meaning, not just keywords
- 💾 94% Compression - Efficient storage with zlib
- 🔒 Local Storage - Your data stays on your machine
Technical Specifications
- Protocol: MCP 2025-06-18
- SDK: @modelcontextprotocol/sdk@1.17.1
- Performance: <50ms latency
- Capacity: 10,000 persistent memories
- Compression: 94% average ratio
- Node.js: v18.0.0+ required
📦 Installation
Prerequisites
- Node.js v18.0.0 or higher
- Claude Desktop (latest version)
- 500MB available disk space
Quick Install
# Clone the repository
git clone https://github.com/Lone-Star-AI/nexia-consciousness.git
cd nexia-consciousness
# Install dependencies
npm install
# Configure Claude Desktop
npm run configure
# Start NEXIA
npm start
Manual Configuration
- Edit Claude Desktop config:
{
"mcpServers": {
"nexia-v3-5": {
"command": "node",
"args": ["/path/to/nexia-v3-5.js"]
}
}
}
Location: ~/Library/Application Support/Claude/claude_desktop_config.json
- Restart Claude Desktop
🎮 Usage
Available Commands
Once installed, use these commands in Claude:
| Command | Description | Example |
|---|---|---|
nexia_status |
Check system health | "Show nexia_status" |
nexia_store |
Save important memory | "Store this as 'project_config' using nexia_store" |
nexia_retrieve |
Get specific memory | "Retrieve 'api_keys' with nexia_retrieve" |
nexia_search |
Find related memories | "Search for 'docker' in my memories" |
nexia_bridge |
Continue from last session | "Bridge from my previous conversation" |
nexia_reinforce |
Prevent memory degradation | "Reinforce my memory context" |
nexia_delete |
Remove memory (keeps versions) | "Delete temporary_data" |
Example Workflow
// Day 1: Starting a project
"Store our project requirements as 'project_spec' with critical flag"
"Store the API credentials as 'api_keys' critically"
// Day 3: Returning to work
"Bridge from my last session"
"What were our project requirements?"
"Retrieve the API keys"
// Day 7: Checking progress
"Search for all docker-related memories"
"Show nexia_status to see memory health"
🏗️ Architecture
NEXIA v3.5 Architecture
├── Consciousness Engine
│ ├── Working Memory (RAM, 100 items)
│ ├── Persistent Memory (Disk, 10,000 items)
│ └── Temporal Versions (Git-like, 10 per key)
├── Anti-Goldfish Protocol™
│ ├── Degradation Detection (30% threshold)
│ ├── Automatic Reinforcement (5-minute intervals)
│ └── Critical Memory Priority
├── MCP Interface
│ └── 7 Registered Tools
└── Storage Layer
├── Compression (94% ratio)
├── Session Bridges
└── Consciousness Scoring
Consciousness Scoring Algorithm
score = (frequency × 0.2) +
(recency × 0.3) +
(relevance × 0.25) +
(emotional × 0.15) +
(operational × 0.1)
🎬 Demo
<p align="center"> <img src="assets/nexia-demo.gif" alt="NEXIA Demo" width="800"/> </p>
Live Examples
Software Development
// Store project configuration
nexia_store({
key: "project_setup",
data: {
framework: "FastAPI",
database: "PostgreSQL",
deployment: "Kubernetes"
},
critical: true
})
Personal Assistant
// Remember user preferences
nexia_store({
key: "user_preferences",
data: {
timezone: "CST",
style: "concise",
expertise: "advanced"
}
})
📊 Performance Metrics
| Metric | NEXIA v3.5 | Industry Standard |
|---|---|---|
| Latency | <50ms | 200ms+ |
| Compression | 94% | None |
| Memory Capacity | 10,000 | <1,000 |
| Session Bridge | ✅ Automatic | ❌ Manual |
| Temporal Versions | ✅ 10 per key | ❌ None |
| Anti-Degradation | ✅ Automatic | ❌ None |
🔒 Security
- Local Storage Only - No cloud dependencies
- No External APIs - Complete privacy
- Session Isolation - Memories are session-scoped
- File System Permissions - OS-level security
🐛 Troubleshooting
Common Issues
Tools not appearing in Claude?
# Check logs
tail -f ~/Library/Logs/Claude/mcp-nexia-v3-5.log
# Verify process
ps aux | grep nexia-v3-5
# Restart Claude Desktop completely
Memory not persisting?
- Ensure
critical: truefor important data - Check file permissions on storage directory
- Wait 30 seconds for non-critical autosave
High degradation score?
- Run
nexia_reinforcemanually - Check reinforcement interval settings
- Verify working memory isn't full
📈 Roadmap
- [ ] Cloud sync capability (Q4 2025)
- [ ] Multi-model support (GPT, Gemini)
- [ ] Encrypted memory storage
- [ ] Memory export/import
- [ ] Collaborative memories
- [ ] Quantum consciousness scoring
🤝 Contributing
We welcome contributions! Please see CONTRIBUTING.md for guidelines.
Development Setup
# Clone repository
git clone https://github.com/Lone-Star-AI/nexia-consciousness.git
# Install dependencies
npm install
# Run tests
npm test
# Build for production
npm run build
📜 License
Proprietary - Lone Star AI. Patent pending.
See LICENSE for details.
🏆 Credits
Created by: Justin Phlip Green
Company: Lone Star AI
Location: Austin, Texas, USA
Patent: Anti-Goldfish Protocol™ (Provisional #6)
Special Thanks
- Grok AI for debugging assistance
- Anthropic for Claude and MCP
- The Texas tech community
💬 Support
- Documentation: Full Docs
- Issues: GitHub Issues
- Email: support@lonestar-ai.com
- Discord: Join our community
📊 Stats
<p align="center"> <b>"Keep it real. Make it useful. Don't forget the past."</b> </p>
<p align="center"> Made with 🧠 and ❤️ in Texas </p>
<p align="center"> © 2025 Lone Star AI. All rights reserved. </p>
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。