Labruno Agent

Labruno 🧪🔍

Labruno is an agent coordinator that creates multiple AI solutions to your coding tasks using parallel sandboxes and evaluates them to find the best implementation.


What is Labruno?

Labruno acts as an orchestrator that:

1. Takes your coding task and spins up multiple isolated sandboxes
2. Asks each sandbox to generate a unique solution using LLaMA 4
3. Executes all solutions in parallel
4. Uses an LLM as judge to evaluate and select the best implementation
5. Shows you all solutions with the winner highlighted

Think of it as having multiple AI developers working on your task simultaneously, with an expert reviewer choosing the best approach.

Features at a Glance

• 🏎️ Parallel Processing: Creates and runs multiple sandboxes concurrently
• 🧠 Multiple Solutions: Generates diverse approaches to the same problem
• 🤖 AI Evaluation: Uses an LLM to judge which solution is best
• 🔒 Secure Execution: Runs code in isolated Daytona sandboxes
• ⚡ Fast Results: Get multiple working implementations in seconds

Setup

Requirements

• Python 3.8+
• Daytona and Groq API keys

Quick Start

# Install
git clone https://github.com/nkkko/labruno.git
cd labruno
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

# Configure
cp .env.example .env
# Add your API keys to .env

# Run
python app.py

API Keys

In your `.env` file:

DAYTONA_API_KEY=your_key_here
DAYTONA_TARGET=us
GROQ_API_KEY=your_key_here

Using Labruno

1. Open http://127.0.0.1:5001 in your browser
2. Type a coding task like "write a function to find prime numbers"
3. Click "Generate and Execute Code"
4. See multiple solutions and the AI's evaluation of the best one

How Labruno Works

1. You provide a task: Ask for any coding solution
2. Concurrent sandboxes spin up: Multiple isolated environments are created in parallel