ConnectOnion
The simplest way to build AI agents with Python functions
⚡
2 Lines of Code
Turn any Python function into an AI tool
🔍
Built-in Debugging
@xray decorator shows agent thinking
📊
Auto Tracking
Automatic behavior recording and history
Installation
Terminal
$ pip install connectonion
Simple Example
That's it! Agents need just: name + tools (your existing Python functions). Everything else is optional.
simple_example.py
from connectonion import Agent
def get_weather(city: str) -> str:
"""Get current weather for a city."""
return f"Weather in {city}: sunny, 72°F"
# Create agent
agent = Agent("assistant", tools=[get_weather])
# Use the agent
response = agent.run("What's the weather in NYC?")
print(response) # Output: "Weather in NYC: sunny, 72°F"Advanced Configuration
ConnectOnion agents support multiple configuration options for different use cases:
advanced_config.py
from connectonion import Agent
# Full configuration example
agent = Agent(
name="advanced_assistant",
model="gpt-5-mini", # Model selection (default)
api_key="sk-...", # Optional API key
system_prompt="Be concise", # Custom personality
tools=[calculate, search], # Multiple tools
)
# Available Models:
# - gpt-5-nano: Fastest, most cost-effective
# - gpt-5-mini: Balanced performance (default)
# - gpt-5: Most capable, highest quality
# API Key Options:
# 1. Environment variable (recommended):
# export OPENAI_API_KEY="sk-..."
# 2. Pass directly:
# agent = Agent(api_key="sk-...")
# System Prompt Options:
# 1. Inline string:
# system_prompt="You are helpful"
# 2. File path:
# system_prompt="prompts/agent.md"
# 3. None (uses default):
# system_prompt=None🚀 Model Selection
Choose based on your needs:
gpt-5-nano: Fast responsesgpt-5-mini: Balanced (default)gpt-5: Best quality
🔑 API Key Setup
Two ways to configure:
- • Environment variable
- • Direct parameter
- • Custom LLM provider
💬 System Prompts
Define agent personality:
- • Inline strings
- • External files (.md, .txt)
- • Dynamic loading