OpenManus-Gemini (grego) ♊

A fork of the FoundationAgents/OpenManus project, modified to use Google's Gemini models as the primary LLM. Integrating litellm allows for easy switching between AI providers like OpenAI or Anthropic. It retains all core agent capabilities—such as web searching, file management, and code execution—powered by the Gemini API.

About This Project

This project adapts the original OpenManus autonomous agent to leverage the power of Google's Gemini models. The core logic has been refactored to use the litellm library, providing a unified and flexible interface for connecting to various Large Language Model providers. This allows you to easily switch between models from Google, OpenAI, Anthropic, Ollama, and more with a simple configuration change.

The agent is capable of complex, multi-step tasks that involve:

• Web Research: Searching the internet to find information.
• Code Generation: Writing Python scripts to perform tasks.
• File System Operations: Creating, reading, and modifying files and directories.
• Task Automation: Chaining these abilities together to achieve high-level goals.

Getting Started

Follow these instructions to set up and run the project on your local machine.

Prerequisites

• Conda for environment management.
• Python 3.12
• A Google AI API key.

Installation

1. Clone the repository:

   git clone [https://github.com/SuperAndy777/grego.git](https://github.com/SuperAndy777/grego.git)
   cd grego

2. Create and activate the Conda environment:

   conda create -n open_manus python=3.12
   conda activate open_manus

3. Install the required packages:

   pip install -r requirements.txt

4. Install Playwright browsers: The agent uses Playwright for web browsing.

   playwright install

Configuration

You must configure the application with your API key before running it.

1. Create your configuration file: Copy the example configuration file. This project uses a Gemini-specific template.

   cp config/config.gemini.example.toml config/config.toml

2. Add your API key: Open config/config.toml with a text editor:

   nano config/config.toml

   Inside the file, find the line api_key = "YOUR_GOOGLE_API_KEY" under both the [llm.default] and [llm.google] sections and replace the placeholder with your actual Google AI API key.

   Note: The config.toml file is included in .gitignore to prevent you from accidentally committing your secret API key.

Usage

Once the installation and configuration are complete, you can run the agent.

1. Activate the Conda environment (if you haven't already):

   conda activate open_manus

2. Run the main script:

   python main.py

3. Provide a prompt: When prompted, give the agent a specific, detailed task. A good prompt should have a clear goal and define what "done" looks like.

   Example Prompt: "Find the top 3 trending topics on Google Trends for the United States. For each topic, perform a web search to find a recent news headline. Compile the topics and their corresponding headlines into a single report named 'trends_report.txt'."

The agent will then begin its execution loop. Any files it creates will be saved in the workspace/ directory.

Acknowledgements

• This project is a fork of the original OpenManus project.
• Special thanks to the LiteLLM team for creating a fantastic multi-provider LLM library.