Introduction

The repository demonstrates two sequential agent swarms: Code Review and Medical Team. Both swarms use Ollama's API and a streaming approach for efficient communication between agents.

Prerequisites

To get started, ensure you have the following installed:

• Python 3.7+
• Ollama running locally
• The Requests library installed using pip install requests

Setting Up Ollama

To set up Ollama, follow these steps:

1. Download and Install Ollama: Download Ollama from the official website and install it on your machine.
2. Run Ollama Locally: Run Ollama using the command ollama start.
3. Verify the Server: Verify the server by visiting http://localhost:11434 in your browser.
4. Test Ollama: Test Ollama using a simple request to generate text from a prompt.

Running the Sequential Swarm

To run the sequential swarm, follow these steps for each swarm:

Code Review Sequential Swarm

• Clone the repository and modify the task description.
• Run the swarm using python run_code_review_swarm.py.
• The generated code will be saved to three separate files: draft.py, revision1.py, and revision2.py.

Medical Team Sequential Swarm

• Modify the task description to reflect the medical task.
• Run the swarm using python run_medical_team_swarm.py.

Configuration Details

The following configuration details can be modified:

• Model Name: Can be modified to use any available model.
• Prompt: Can be modified to define the task for code generation or medical diagnosis.
• Agent Flow: Can be modified to change the order in which agents process the task.

Customizing Agents

Agents can be customized by:

• Modifying their respective system prompts.
• Tweaking each agent to perform a specific role or handle different tasks.

License

The repository is licensed under the MIT License.

Getting Started

To get started, follow these steps:

1. Install Python 3.7+ and the Requests library.
2. Download and install Ollama.
3. Run Ollama locally and verify the server.
4. Clone the repository and modify the task description for the desired swarm.
5. Run the swarm using the provided Python script.
6. Review the generated code or medical diagnosis results.