Universal Living Memory

Universal AI connection framework with three symbiotic libraries: API, Webhook, and MCP.

🚀 Features (v1.1.0 - Universal)

Universal Connection Framework

• 🔑 API Connection Library: Connect to ANY AI model or API service

  • OpenAI, Anthropic, Google, Grok, Together AI, Mistral, etc.
  • Custom authentication (Bearer, API Key, Custom headers)
  • Multi-model support per connection
  • Dynamic capability detection

• 🪝 Webhook Library: Event-driven integrations

  • Zapier, n8n, Make, custom webhooks
  • Event filtering (completion, error, all)
  • Multiple HTTP methods (POST, GET, PUT)
  • Automatic triggering on events

• 🔧 MCP Server Library: Model Context Protocol integrations

  • Filesystem access
  • Database connections
  • Custom tool servers
  • Extensible capabilities

Core Capabilities

• Hardware Discovery: Automatically detects system resources
• Vector Memory: Persistent memory using Qdrant
• Resource Management: Dynamic client hydration and collapse
• Cloud Spot Pricing: Monitor cheapest GPU options
• Self-Termination: Automatic shutdown after inactivity
• Docker-Based: Easy deployment

Quick Start

# 1. Clone the repository
git clone https://github.com/brian95240/universal-living-memory.git
cd universal-living-memory

# 2. Configure credentials
cp .env.example .env
# Edit .env with your API keys (store them in Vaultwarden first!)

# 3. Run genesis to configure system
python genesis.py

# 4. Launch infrastructure
docker compose up -d

# 5. Check health
curl http://localhost:8000/health

Universal Connection Management

Add API Connection

curl -X POST http://localhost:8000/v1/connections/api \
  -H "Content-Type: application/json" \
  -d '{
    "conn_id": "together_ai",
    "name": "Together AI",
    "base_url": "https://api.together.xyz/v1",
    "auth_type": "bearer",
    "api_key_value": "your_api_key_here",
    "models": ["mistralai/Mixtral-8x7B-Instruct-v0.1"],
    "capabilities": ["chat", "completion"],
    "enabled": true
  }'

Add Webhook

curl -X POST http://localhost:8000/v1/connections/webhook \
  -H "Content-Type: application/json" \
  -d '{
    "webhook_id": "zapier_hook",
    "name": "Zapier Integration",
    "url": "https://hooks.zapier.com/hooks/catch/...",
    "method": "POST",
    "events": ["completion", "error"],
    "enabled": true
  }'

Add MCP Server

curl -X POST http://localhost:8000/v1/connections/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "server_id": "filesystem",
    "name": "Filesystem MCP",
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"],
    "capabilities": ["read", "write", "list"],
    "enabled": true
  }'

Architecture

• FastAPI Orchestrator: Main API server with universal routing
• PostgreSQL: Relational data storage
• Qdrant: Vector database for semantic memory
• Connection Library Manager: Three symbiotic libraries (API, Webhook, MCP)
• Universal Adapter: Dynamic connection handling for any service
• Memory Engine: FastEmbed-based semantic memory
• Cloud Delta Engine: Spot pricing aggregation
• Lifecycle Monitor: Automatic resource cleanup

API Endpoints

Connection Management

# API Connections
POST /v1/connections/api          # Add API connection
GET /v1/connections/api           # List API connections
DELETE /v1/connections/api/{id}   # Remove API connection

# Webhooks
POST /v1/connections/webhook      # Add webhook
GET /v1/connections/webhook       # List webhooks
DELETE /v1/connections/webhook/{id} # Remove webhook

# MCP Servers
POST /v1/connections/mcp          # Add MCP server
GET /v1/connections/mcp           # List MCP servers
DELETE /v1/connections/mcp/{id}   # Remove MCP server

# Universal Operations
GET /v1/connections/all           # Get all connections
GET /v1/connections/search?q=     # Search connections
GET /v1/connections/stats         # Get statistics
POST /v1/connections/confirm      # AI confirmation

Chat Completion

POST /v1/chat/completions         # Chat with any provider

Cloud Pricing

GET /v1/cloud/pricing             # Get spot pricing
GET /v1/cloud/cheapest            # Get cheapest GPU

Health & Status

GET /health                       # System health
GET /                             # API info

Supported Services

AI Models (API Library)

• OpenAI (GPT-4, GPT-3.5)
• Anthropic (Claude)
• Google (Gemini)
• X.AI (Grok)
• Together AI (Mixtral, Llama)
• Mistral AI
• Cohere
• Replicate
• Any OpenAI-compatible API

Webhooks (Webhook Library)

• Zapier
• n8n
• Make (Integromat)
• IFTTT
• Custom webhooks
• Slack webhooks
• Discord webhooks

MCP Servers (MCP Library)

• Filesystem
• PostgreSQL
• SQLite
• GitHub
• Google Drive
• Custom MCP servers

Configuration Files

Connection libraries are stored in JSON format:

• config/api_library.json - API connections
• config/webhook_library.json - Webhooks
• config/mcp_library.json - MCP servers

All libraries support hot-reload without restart.

Security

• Store all credentials in Vaultwarden
• Use environment variables for API keys
• Never commit credentials to git
• See SECURITY.md for details

Contributing

See CONTRIBUTING.md for development guidelines.

License

AGPL-3.0 + Commercial (see LICENSE)

Related Projects

• Genesis Studio - Voice-guided client with connection management

Changelog

v1.1.0 (Universal)

• Added Universal Connection Framework
• Added API Connection Library
• Added Webhook Library
• Added MCP Server Library
• Added Universal Adapter for dynamic connections
• Enhanced future-proof architecture

v1.0.1 (Hyper-Dynamic)

• Added runtime provider registration
• Added cloud spot pricing discovery
• Added self-termination for idle systems

v1.0.0 (Golden Master)

• Initial release
• Multi-provider AI integration
• Vector memory with Qdrant