persistent-ai-memory

Overview

Persistent AI Memory System provides sophisticated memory management for AI assistants through a multi-database architecture with SQLite storage. It offers semantic search capabilities using vector embeddings, conversation tracking across platforms, and intelligent memory filtering. The system features three integration methods: OpenWebUI plugin (primary), MCP server for cross-platform compatibility, and standalone library for custom implementations. It includes advanced features like multi-user isolation with configurable user and model IDs, tool call logging, and self-reflection capabilities.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

Comparable tools

Installation

# Linux/macOS
pip install git+https://github.com/savantskie/persistent-ai-memory.git

# Windows (same command, just use Command Prompt or PowerShell)
pip install git+https://github.com/savantskie/persistent-ai-memory.git

# Via mcpo
python -m ai_memory_mcp_server

{
  "mcpServers": {
    "ai-memory": {
      "command": "python",
      "args": ["-m", "ai_memory_mcp_server"]
    }
  }
}

FAQ

What embedding providers does this system support?

The system supports Ollama (local), LM Studio (local), and OpenAI (cloud) as embedding providers, each with different speed, quality, and cost characteristics.

How does user and model ID isolation work?

All memory operations require user_id and model_id parameters to ensure multi-user safety, model tracking, and audit trails. This can be configured as strict mode (recommended for production) or use defaults (simpler for single-user setups).

Can this system be integrated with platforms other than OpenWebUI?

Yes, the system provides three integration methods: OpenWebUI plugin (primary), MCP server for any MCP-compatible assistant, and standalone library for custom Python implementations.

Compare persistent-ai-memory with