Skip to content

openai-agents-session

Redis and DynamoDB session backends for openai-agents-python

PyPI version Python Versions License: MIT

Overview

openai-agents-session provides production-ready session storage backends for the OpenAI Agents Python SDK. It enables persistent conversation history across multiple interactions with your AI agents.

Features

  • Redis Backend: High-performance in-memory storage with optional TTL
  • DynamoDB Backend: Serverless, scalable cloud storage with automatic expiration
  • Async Native: Built with asyncio for non-blocking operations
  • Type Safe: Full type hints and PEP 561 compliant
  • Zero Config: Works out of the box with sensible defaults

Quick Example

import redis.asyncio as redis
from agents import Agent, Runner
from openai_agents_session import RedisSession

# Create Redis client
client = redis.from_url("redis://localhost:6379")

# Create session with 1-hour TTL
session = RedisSession(
    session_id="user-123",
    redis_client=client,
    ttl=3600,
)

# Use with OpenAI Agent
agent = Agent(name="Assistant", instructions="You are a helpful assistant.")
runner = Runner(agent=agent, session=session)

# Conversation persists across calls
result = await runner.run("Hello! My name is Alice.")
result = await runner.run("What's my name?")  # Agent remembers "Alice"

Installation

pip install "openai-agents-session[redis]"

pip install "openai-agents-session[dynamodb]"

pip install "openai-agents-session[all]"

Why Use This?

The default openai-agents-python stores conversation history in memory, which means:

  • History is lost when your application restarts
  • Each instance has its own isolated memory
  • No way to resume conversations across sessions

With openai-agents-session, you get:

  • Persistent storage that survives restarts
  • Shared state across multiple application instances
  • Automatic expiration with configurable TTL
  • Scalable backends for production workloads

Next Steps