Something went wrong. Try again.
[READ-ONLY] Mirror of https://github.com/takeruhukushima/MAGIagent.
Something went wrong. Try again.
Files
- Python 100%
README.md
MAGI: Manus Artificial General Intelligence for academic research assistance with human-in-the-loop AI agents #
MAGI (Manus Artificial General Intelligence) is a system designed to assist with materials science research by coordinating multiple specialized AI agents under human supervision.
Overview #
MAGI consists of a Master Research Agent that coordinates several specialized sub-agents:
- Master Research Agent: Coordinates the workflow between sub-agents and serves as the main interface for users.
- Planning Agent: Handles research planning, literature review, and experimental design.
- Execution Agent: Manages experiment execution, simulation runs, and data collection.
- Analysis Agent: Performs data analysis, visualization, and interpretation.
- Knowledge Agent: Manages research artifacts, references, and knowledge retrieval.
Installation #
-
Clone the repository:
git clone https://github.com/yourusername/gemini-ai-agent.git cd gemini-ai-agent/agents/academic-research -
Create and activate a virtual environment:
python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate -
Install dependencies:
pip install -r requirements.txt -
Set up environment variables: Create a
.envfile in the project root with your Google Cloud credentials:GOOGLE_APPLICATION_CREDENTIALS="path/to/your/service-account-key.json"
Usage #
from MAGI.agent import root_agent
# Initialize the MAGI system
agent = root_agent
# Start a research task
response = agent.process("Help me design an experiment to study the properties of graphene oxide.")
print(response)
Architecture #
Features #
- Human-in-the-loop: Critical decisions require human approval
- Modular Design: Easy to extend with new agents and tools
- Knowledge Management: Built-in vector database for research artifacts
- Reproducibility: Full audit trail of all research activities
License #
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.