memexllm 0.0.6.post1

A library for managing LLM conversation history

🐟 MemexLLM

Overview

MemexLLM is a Python library for managing and storing LLM conversations. It provides a flexible and extensible framework for history management, storage, and retrieval of conversations.

Features

• Drop-in Integrations: Add conversation management to your LLM applications with zero code changes using our provider integrations
• Flexible Storage: Choose from memory, SQLite, or bring your own storage backend
• Conversation Management: Organize, retrieve, and manipulate conversation threads with ease
• Memory Management Algorithms: Control conversation context with built-in algorithms (FIFO, summarization, etc.)
• Provider Agnostic: Works with OpenAI, Anthropic, and other LLM providers
• Extensible Architecture: Build custom storage backends and memory management algorithms

Quick Start

Installation

pip install memexllm  # Basic installation
pip install memexllm[openai]  # With OpenAI support

Basic Usage

from memexllm.storage import MemoryStorage
from memexllm.algorithms import FIFOAlgorithm
from memexllm.core import HistoryManager

# Initialize components
storage = MemoryStorage()
algorithm = FIFOAlgorithm(max_messages=100)
history_manager = HistoryManager(storage=storage, algorithm=algorithm)

# Create a conversation thread
thread = history_manager.create_thread()

# Add messages
history_manager.add_message(
    thread_id=thread.id,
    content="Hello, how can I help you today?",
    role="assistant"
)

Zero-Code-Change Integration

Add conversation management to your OpenAI application with no code changes:

from openai import OpenAI
from memexllm.integrations.openai import patch_openai

# Initialize your OpenAI client as usual
client = OpenAI(api_key="your-api-key")

# Patch the OpenAI client to add conversation management
# This adds persistence without changing your existing code
patch_openai(client)

# Use the client as you normally would - conversations are now managed automatically
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello, who are you?"}]
)

Documentation

For detailed documentation, including:

• Complete API reference
• Advanced usage examples
• Available storage backends
• Contributing guidelines
• Feature roadmap

Visit our documentation at: https://eyenpi.github.io/MemexLLM/

Contributing

We welcome contributions! Please see our Contributing Guide for details on how to get started.

License

This project is licensed under the MIT License.