panda-agi 0.4.1

PandaAGI SDK

🐼 PandaAGI SDK - An SDK for AGI (Agentic General Intelligence)

The PandaAGI SDK provides a simple, intuitive API for building general AI agents in just a few lines of code. It abstracts away the complexity of Agentic Loops and provides a powerful interface for you to build autonomous agents. Each agent can be configured to run in a custom environment, interacting with the web, your file system, writing code, and running shell commands.

Installation

pip install panda-agi

Or with uv:

uv add panda-agi

🔧 Getting started

First of all, make sure you have a API key. You can get one for free here. Make sure to set it as an environment variable:

export PANDA_AGI_KEY=your_api_key

or set it in the .env file:

PANDA_AGI_KEY=your_api_key

Once you have the API key, you can start using the SDK:

import asyncio
from panda_agi import Agent
from panda_agi.envs import LocalEnv

async def main():
    # Create a custom environment for the agent
    agent_env = LocalEnv("./my_agent_workspace")
    
    # Create the agent
    agent = Agent(environment=agent_env)
    
    # Run the agent with a task
    response = agent.run("Tell me a joke about pandas")
    print(response.output)

    # Other possible tasks
    response = agent.run("Make a report of the real estate market in Germany")
    # -> will generate a reporrt in the provided workspace folder

    response = agent.run("Can you analyze our sales and create a dashboard?")
    # -> will generate a dashboard in the provided workspace folder starting from a csv file in the workspace folder

    response = agent.run("Can you create a website for our company?")
    # -> will generate a website in the provided workspace folder

    # Disconnect when done
    await agent.disconnect()

if __name__ == "__main__":
    asyncio.run(main())

In case you want to enable te web search, you will also need a Tavily API key. You can get one for free here. Then set it as an environment variable or set it in the .env file:

TAVILY_API_KEY=your_api_key

📱 Running with the UI

In case you don't want to build an app from scratch, we provide a UI that you can use to run your agents.

Running it is as simple as:

# Run the UI
cd examples/ui
./start.sh

This will start a docker container with the UI running. You can access it at http://localhost:3000 and start using it.

🛠️ Features

• Simple, intuitive API for interacting with PandaAGI agents
• Support for local and Docker environments
• Asynchronous event-based communication
• Pydantic models for type safety

📚 Documentation

For complete documentation, visit our documentation site.

🛠️ Development

Prerequisites

• Python 3.8+
• uv

Setup

1. Clone the repository
2. Install dependencies:

uv pip install -e ".[dev]"

Testing

Run tests with pytest:

uv run pytest

📝 License

MIT License