How to intermediate · 3 min read

How to implement agent state management

Quick answer

Implement agent state management by storing and updating the agent's context or memory between interactions, typically using variables, databases, or vector stores. Use client.chat.completions.create with context passed in the messages array to maintain state across calls.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install openai>=1.0

Setup

Install the OpenAI Python SDK and set your API key as an environment variable to securely authenticate requests.

bash

pip install openai>=1.0

Step by step

Maintain agent state by storing conversation history in a list and passing it with each API call to preserve context. This example demonstrates a simple chat agent that remembers previous messages.

python

import os
from openai import OpenAI

client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Initialize conversation state
conversation_history = [
    {"role": "system", "content": "You are a helpful assistant."}
]

def chat_with_agent(user_input):
    # Append user message to history
    conversation_history.append({"role": "user", "content": user_input})

    # Call the chat completion API with full conversation history
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=conversation_history
    )

    # Extract assistant reply
    assistant_message = response.choices[0].message.content

    # Append assistant reply to history
    conversation_history.append({"role": "assistant", "content": assistant_message})

    return assistant_message

# Example usage
print(chat_with_agent("Hello, who won the World Series in 2020?"))
print(chat_with_agent("Can you remind me what you just said?"))

output

The Los Angeles Dodgers won the World Series in 2020.
I said that the Los Angeles Dodgers won the World Series in 2020.

Common variations

You can implement agent state management using asynchronous calls, streaming responses, or by storing state externally in databases or vector stores for scalability. Different models like claude-3-5-sonnet-20241022 or gemini-1.5-pro can be used similarly by passing conversation history.

python

import os
import anthropic

client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

conversation_history = []

system_prompt = "You are a helpful assistant."

conversation_history.append({"role": "system", "content": system_prompt})

user_input = "Hello, how are you?"
conversation_history.append({"role": "user", "content": user_input})

message = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=500,
    system=system_prompt,
    messages=conversation_history
)

assistant_reply = message.content[0].text
conversation_history.append({"role": "assistant", "content": assistant_reply})

print(assistant_reply)

output

I'm doing well, thank you! How can I assist you today?

Troubleshooting

If the agent forgets context, ensure the full conversation history is passed with each API call and that the token limit is not exceeded. For large histories, summarize or truncate older messages. Also, verify your API key is correctly set in os.environ.

✅

Key Takeaways

Maintain agent state by storing conversation history and passing it with each API call.
Use environment variables for API keys and the latest SDK client patterns for security and compatibility.
For long conversations, manage token limits by summarizing or truncating history.
Agent state can be stored in-memory, databases, or vector stores depending on scale and persistence needs.
Different LLM providers and models support similar state management via message history.

Verified 2026-04 · gpt-4o, claude-3-5-sonnet-20241022, gemini-1.5-pro

Verify ↗