Memory Management¶

MCPOmni Connect provides flexible memory management through Redis persistence and file-based conversation history, allowing you to maintain context across sessions and save important conversations.

Memory Types¶

Type	Storage	Persistence	Use Case
Session Memory	RAM	Current session only	Active conversation
Redis Memory	Redis Database	Across sessions	Long-term context
File History	JSON Files	Manual save/load	Conversation backups

Redis-Powered Persistence¶

Overview¶

Redis memory provides automatic conversation persistence across MCPOmni Connect sessions, maintaining context and conversation history even after restarts.

Setup Redis¶

Ubuntu/DebianmacOSDockerWindows

sudo apt update
sudo apt install redis-server
sudo systemctl start redis-server
sudo systemctl enable redis-server

brew install redis
brew services start redis

docker run -d --name redis -p 6379:6379 redis:alpine

Download from Redis Windows releases or use WSL.

Configuration¶

Configure Redis connection in your .env file:

.env

# Redis Configuration (optional)
REDIS_HOST=localhost
REDIS_PORT=6379
REDIS_DB=0
REDIS_PASSWORD=your_password  # if password protected

Using Redis Memory¶

Enable/Disable Memory Persistence¶

# Toggle memory persistence
/memory

When Enabling:

✅ Memory persistence is now ENABLED using Redis
Conversations will be saved and restored across sessions.
Connected to Redis at localhost:6379 (DB: 0)

When Disabling:

❌ Memory persistence is now DISABLED
Conversations will not be saved to Redis.
Session memory will be cleared on restart.

Check Memory Status¶

/status

Output:

🚀 MCPOmni Connect Status:
- Mode: CHAT
- Connected Servers: 3
- Memory: ENABLED (Redis localhost:6379)
- Debug: OFF
- Session Duration: 45 minutes

How Redis Memory Works¶

Automatic Saving: Conversations automatically saved to Redis
Context Preservation: Maintains conversation context across restarts
Session Restoration: Previous conversations loaded on startup
Intelligent Pruning: Old conversations automatically cleaned up

Example Session with Redis¶

# Session 1
> /memory
Memory persistence is now ENABLED using Redis

> Hello, I'm working on a Python project
AI: Hello! I'd be happy to help with your Python project. What specific 
    aspect are you working on?

> I need to analyze some CSV files
AI: Great! I can help you analyze CSV files. Do you have the files ready?

# Exit MCPOmni Connect

# Session 2 (later)
> mcpomni_connect
🚀 MCPOmni Connect - Universal Gateway to MCP Servers
📚 Restored conversation from Redis (5 messages)

> What were we discussing?
AI: We were discussing your Python project, specifically analyzing CSV files.
    You mentioned you needed help with CSV analysis. Are you ready to continue?

File-Based Chat History¶

Overview¶

File-based history allows manual saving and loading of specific conversations, useful for:

Project Documentation: Save conversations related to specific projects
Backup: Create backups of important discussions
Sharing: Share conversation history with team members
Templates: Save common workflows as templates

Saving Conversations¶

Manual Save¶

/save:filename.json

Examples:

# Save with descriptive names
/save:database_migration_discussion.json
/save:bug_investigation_2024-01-15.json
/save:deployment_workflow.json

# Save with timestamps
/save:session_$(date +%Y%m%d_%H%M).json

Response:

✅ Conversation saved to database_migration_discussion.json
Saved 15 messages (3,247 tokens)
File location: ./database_migration_discussion.json

Loading Conversations¶

Manual Load¶

/load:filename.json

Examples:

# Load specific conversation
/load:database_migration_discussion.json

# Load recent session
/load:session_20240115_1430.json

Response:

✅ Conversation loaded from database_migration_discussion.json
Loaded 15 messages from previous session
Context restored: Database migration planning

File Format¶

Conversation files are saved in JSON format:

example_conversation.json

{
    "metadata": {
        "version": "1.0",
        "created_at": "2024-01-15T14:30:00Z",
        "mcpomni_version": "0.3.0",
        "total_messages": 15,
        "total_tokens": 3247,
        "session_duration": "45 minutes"
    },
    "configuration": {
        "llm_provider": "openai",
        "llm_model": "gpt-4o-mini",
        "connected_servers": ["filesystem", "database", "notifications"]
    },
    "conversation": [
        {
            "timestamp": "2024-01-15T14:30:15Z",
            "role": "user",
            "content": "Help me analyze the database schema",
            "command": null
        },
        {
            "timestamp": "2024-01-15T14:30:20Z",
            "role": "assistant",
            "content": "I'll help you analyze the database schema...",
            "tools_used": ["query_database"],
            "tokens": 156
        }
    ]
}