gemini-starter-agent 0.1.2

A CLI tool to bootstrap Gemini agents with OpenAI Agent SDK using UV.

Gemini Starter Agent - Simplifying Gemini Chat Completions in the OpenAI Agents SDK

Gemini Starter Agent is a Python CLI tool to quickly bootstrap AI agents using the Gemini API and OpenAI Agent SDK. It leverages UV for project scaffolding, virtual environment management, and dependency handling.

Features

• Quick scaffolding of a Gemini AI agent project.

• Automatic creation of virtual environments.

• Easy dependency installation (openai-agents, python-dotenv).

• Generates a ready-to-run main.py for your agent.

• Friendly CLI prompts for agent name, purpose, and API key.

• PEP-621 compliant pyproject.toml scripts for easy CLI execution.

• Cross-platform support.

Installation

Install the package via pip:

pip install gemini-starter-agent

Run the package

gemini-starter-agent

Usage

Run the CLI to generate a new Gemini agent:

`pip install gemini-starter-agent   `

You will be prompted for:

• Project name (default: agent)

• Gemini API key

• Gemini model (choose from default or enter your own)

• Agent name (default: Helpful Assistant)

• Agent purpose/instructions

After completion, your project structure will look like:

/your-project-name/
├── src/
│   └── your_project_name/
│       ├── __init__.py
│       └── main.py
├── .env
├── pyproject.toml
└── uv.lock

Example main.py

Here is how the generated main.py will look:

import asyncio
import os
from dotenv import load_dotenv
from agents import Agent, Runner, RunConfig, OpenAIChatCompletionsModel, set_tracing_disabled
from openai import AsyncOpenAI

# Load environment variables
load_dotenv()

GEMINI_MODEL = os.getenv("GEMINI_MODEL")
GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
BASE_URL = os.getenv("BASE_URL")

# Disable tracing for cleaner output
set_tracing_disabled(True)

client: AsyncOpenAI = AsyncOpenAI(api_key=GEMINI_API_KEY, base_url=BASE_URL)
model: OpenAIChatCompletionsModel = OpenAIChatCompletionsModel(GEMINI_MODEL, client)

agent: Agent = Agent(
    name="{agent_name}",
    instructions="{agent_purpose}",
    model=model,
)

async def main() -> None:
    prompt = "What is Agentic AI? The output format should be in haiku" # enter a prompt here
    result = await Runner.run(agent, prompt, run_config=RunConfig(model))
    print(result.final_output)

def start():
    asyncio.run(main())

Running Your Agent

Change into the project folder:

Run the agent using UV scripts:

cd [project_name]
uv run helpful-assistant               

Environment Variables

The .env file is automatically generated and contains:

GEMINI_API_KEY=your_api_key_here  
GEMINI_MODEL=your_model_here  
BASE_URL=https://generativelanguage.googleapis.com/v1beta/openai/`

Contributing

• Submit bug reports or feature requests via GitHub Issues.

• Fork the repository and create pull requests for improvements.

• Ensure code style consistency and add documentation for new features.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Author

Marjan Ahmed

• Email: [e-mail]

• GitHub: https://github.com/marjan-ahmed