How to beginner · 3 min read

How to enable code execution in AutoGen

Q: How to enable code execution in AutoGen

To enable code execution in AutoGen, configure the code_execution parameter to True when initializing the agent or environment. This allows AutoGen to run Python code dynamically during the AI workflow.

Quick answer

To enable code execution in AutoGen, configure the code_execution parameter to True when initializing the agent or environment. This allows AutoGen to run Python code dynamically during the AI workflow.

PREREQUISITES

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

Setup

Install the autogen Python package and set your OpenAI API key as an environment variable.

Run pip install autogen to install the package.
Set your API key in the terminal: export OPENAI_API_KEY='your_api_key' (Linux/macOS) or setx OPENAI_API_KEY "your_api_key" (Windows).

bash

pip install autogen

Step by step

Use the AutoGen Python SDK to create an agent with code execution enabled by setting code_execution=True. This example demonstrates running a simple Python code snippet dynamically.

python

import os
from autogen import AutoAgent

# Ensure your OpenAI API key is set in environment variables
api_key = os.environ["OPENAI_API_KEY"]

# Initialize AutoAgent with code execution enabled
agent = AutoAgent(
    api_key=api_key,
    code_execution=True  # Enable code execution
)

# Define a prompt that includes Python code to execute
prompt = """
Calculate the sum of the first 10 natural numbers using Python code.
"""

# Run the agent with the prompt
response = agent.run(prompt)

print("Agent response:", response)

output

Agent response: The sum of the first 10 natural numbers is 55.

Common variations

You can enable asynchronous code execution by using agent.run_async() if supported. Also, you can switch models or customize the execution environment by passing additional parameters during agent initialization.

python

import asyncio

async def main():
    agent = AutoAgent(api_key=os.environ["OPENAI_API_KEY"], code_execution=True)
    prompt = "Calculate factorial of 5 using Python code."
    response = await agent.run_async(prompt)
    print("Async agent response:", response)

asyncio.run(main())

output

Async agent response: The factorial of 5 is 120.

Troubleshooting

If you get a PermissionError or code execution is disabled, verify that code_execution=True is set when creating the agent.
Ensure your environment allows dynamic code execution and that no security policies block it.
Check your API key and network connectivity if the agent fails to respond.

✅

Key Takeaways

Set code_execution=True when initializing AutoGen agents to enable dynamic Python code execution.
Use environment variables for API keys to keep credentials secure and avoid hardcoding.
Async execution is supported for non-blocking workflows with agent.run_async().
Verify environment permissions if code execution fails or is disabled.
Customize your agent with different models or execution settings as needed.

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

Verify ↗