Generate a program.md for your task and start autonomous research in minutes
Brief description of what you're optimizing
The file the agent will modify
How to execute the training script
The metric to optimize (lower is better or higher is better)
How to extract the metric from run.log
Number of parallel experiment workers (0 = sequential only)
Files or resources the agent should NOT modify (one per line)
Select the categories relevant to your task
Click "Generate program.md" to see the output...
The starter kit includes maze.py (knowledge base), a program.md template, and setup instructions. Everything you need to start autonomous research.