Skip to content

om-surushe/owui-adaptive-memory

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Adaptive Memory v3.1 for OpenWebUI

License: MIT Python 3.8+ OpenWebUI Plugin

A sophisticated plugin that provides persistent, personalized memory capabilities for LLMs within OpenWebUI.

Adaptive Memory automatically learns facts, preferences, and goals from your AI conversations and intelligently injects relevant memories into future prompts—creating a more natural and personalized experience.


✨ Key Features

  • Auto-Learning — Automatically extracts facts (e.g., "I live in Paris"), preferences (e.g., "I prefer Python"), and goals from conversations
  • Smart Retrieval — Vector-based similarity for efficient, contextually relevant memory retrieval
  • Deduplication — Multi-layered semantic and text-based similarity to prevent duplicate memories
  • Memory Banks — Categorize memories into General, Personal, Work, etc. for focused retrieval
  • Auto-Summarization — Periodically condenses older memories to save space and reduce clutter
  • Memory Confidence Scoring — Filters low-quality memories based on configurable thresholds
  • Flexible Providers — Supports both Ollama and OpenAI-compatible APIs for LLM and embeddings
  • Health Monitoring — Prometheus metrics and health endpoints for observability

🚀 Quick Start

  1. Install the Plugin: Copy adaptive_memory_v3.1.py to your OpenWebUI plugins folder
  2. Enable the Plugin: Go to Workspace → Tools → Adaptive Memory and toggle valid valves
  3. Basic Configuration:
    • Embedding Provider: Choose local (free, runs on your server) or openai_compatible (external API)
    • LLM Provider: Choose ollama (default) or openai_compatible
  4. Start Chatting: The AI will now start remembering key details about you!

⚙️ Configuration Guide ("Valves")

Setting Description
Embedding Provider local or openai_compatible
LLM Provider ollama or openai_compatible
Minimum Confidence How certain the AI must be to save a memory (0.0 - 1.0)
Duplicate Threshold How strictly duplicates are detected

Advanced settings like auto-summarization and custom memory banks generally don't need changing.


💬 Chat Commands

Command Description
/memory status Check if memory is active
/memory list View all stored memories
/memory forget [id] Delete a specific memory
/memory list_banks See available memory banks
/memory assist Get help with memory features

🔧 Troubleshooting

Issue Solution
"No memories found" Chat more naturally about yourself. Ensure Minimum Confidence isn't set too high (default 0.75 is good)
Errors Check OpenWebUI logs. Verify your LLM and Embedding provider URLs are correct

📊 API Endpoints

  • GET /adaptive-memory/health — Health check
  • GET /adaptive-memory/metrics — Prometheus metrics

🙏 Credits

Based on the original work by gramanoid.


📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

About

Persistent, personalized memory plugin for OpenWebUI

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages