TeenAGI

A Python package for building AI agents capable of making multiple function calls in sequence. Inspired by BabyAGI but with more advanced capabilities.

Table of Contents

• Installation
• Usage
  • Setting Up API Keys
  • Python API
  • Command Line Interface
  • Function Registration
• How It Works
• Development
  • Setup Development Environment
• License
• Contributing

Installation

pip install teenagi

Usage

Setting Up API Keys

TeenAGI requires API keys to be stored in a .env file in your project directory. Create a file named .env with the following content:

# For OpenAI GPT models
OPENAI_API_KEY=your_openai_api_key_here

# For Anthropic Claude models
ANTHROPIC_API_KEY=your_anthropic_api_key_here

This file should be kept secure and never committed to version control.

Python API

from teenagi import TeenAGI, create_agent

# TeenAGI will automatically load API keys from your .env file
agent = TeenAGI(
    name="Research Assistant",
    provider="openai",  # "openai" or "anthropic"
    model="gpt-4.1
)

# Add capabilities to the agent
agent.learn("can search the web for information")
agent.learn("can summarize long documents")

# Get a response that might require multiple function calls
response = agent.respond("Find and summarize recent research on climate change")
print(response)

# Alternative: use the factory function with Anthropic
specialized_agent = create_agent(
    name="DataAnalyst",
    provider="anthropic",
    model="claude-3-opus-20240229"
)

Command Line Interface

TeenAGI includes a CLI for quick interactions. It will automatically use the API keys from your .env file:

# Basic usage
teenagi --name "ResearchBot" --capabilities "can search the web" "can summarize text" "Find recent papers on quantum computing"

# Using with OpenAI (requires OPENAI_API_KEY in .env)
teenagi --provider openai --model "gpt-4" "Explain quantum entanglement"

# Using with Anthropic (requires ANTHROPIC_API_KEY in .env)
teenagi --provider anthropic --model "claude-3-sonnet-20240229" "Explain quantum entanglement"

Function Registration

TeenAGI can call actual Python functions. Here's how to register and use functions:

from teenagi import TeenAGI
import datetime

# Create an agent
agent = TeenAGI(name="AssistantBot", provider="anthropic")

# Register functions using the decorator syntax
@agent.register_function(description="Get the current date and time")
def get_current_time() -> str:
    """Get the current date and time."""
    now = datetime.datetime.now()
    return now.strftime("%Y-%m-%d %H:%M:%S")

@agent.register_function(description="Get weather information for a location")
def get_weather(location: str) -> dict:
    """
    Get current weather for a location.
    
    Args:
        location: City name or location
    """
    # In a real app, you'd call a weather API
    return {
        "location": location,
        "temperature": 72,
        "condition": "Sunny",
        "humidity": 65
    }

# Add capabilities to the agent
agent.learn("can check the current time")
agent.learn("can get weather information")

# Let the agent use the functions
response = agent.respond("What time is it now and what's the weather in New York?")
print(response)

The agent will:

1. Determine which functions to call based on the user's request
2. Call the appropriate functions with the right parameters
3. Incorporate the function results into the final response

When calling functions, TeenAGI will:

• Automatically parse and validate function arguments
• Handle type conversions
• Provide helpful error messages
• Support complex parameter types
• Allow multiple function calls in sequence

How It Works

TeenAGI enables LLMs to use registered Python functions based on the user's request. Here's how it works:

sequenceDiagram
    participant User
    participant TeenAGI
    participant LLM as LLM (OpenAI/Anthropic)
    participant Functions as Registered Functions
    
    User->>TeenAGI: respond("Get weather in London and calculate 15*7")
    TeenAGI->>LLM: Send prompt with available functions
    LLM-->>TeenAGI: Function call: get_weather("London")
    TeenAGI->>Functions: Execute get_weather("London")
    Functions-->>TeenAGI: Return weather data
    TeenAGI->>LLM: Send weather data, continue with request
    LLM-->>TeenAGI: Function call: calculate("15*7")
    TeenAGI->>Functions: Execute calculate("15*7")
    Functions-->>TeenAGI: Return 105
    TeenAGI->>LLM: Send calculation result, request final response
    LLM-->>TeenAGI: Generate final response with all data
    TeenAGI->>User: Return complete response

The process follows these steps:

1. Registration: Register Python functions with the agent
2. Request: User sends a natural language request
3. Function Selection: LLM determines which functions to call based on the request
4. Execution: TeenAGI executes the selected functions with the provided arguments
5. Iteration: Results are fed back to the LLM, which may call additional functions
6. Response: After all necessary functions are called, a final response is generated

This approach allows TeenAGI to perform real actions and access external data sources while maintaining a simple, natural language interface.

Development

Setup Development Environment

# Clone the repository
git clone https://github.com/jordan/teen-agi.git
cd teen-agi

# Install dependencies with Poetry
poetry install

# Create a .env file with your API keys
echo "OPENAI_API_KEY=your_key_here" > .env
# or
echo "ANTHROPIC_API_KEY=your_key_here" > .env

# Run tests
poetry run pytest

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.