bart-loop, your Claude/Opencode workflow

bart-loop is a method for developing with AI that uses a simple loop to keep giving an AI agent a task until it's finished. It's inspired by a character from The Simpsons and focuses on trying again and again rather than getting it perfect the first time. The idea is that repeating a task helps improve it over time. When the AI makes mistakes, it's seen as useful information that helps guide the process. The person guiding the AI needs to be good at writing clear instructions, because the AI basically reflects the skill of the person using it. The loop handles trying again automatically.

Benefits

• Persistent Improvement:The loop keeps working on a task, refining it with each attempt.
• Learning from Mistakes:Failures are treated as valuable data to guide the AI.
• Automated Retries:The system automatically handles retrying tasks, saving the user effort.
• Structured Development:It provides a clear framework for iterative AI development.

Use Cases

• Coding Projects:Can be used to build features, write tests, and refactor code.
• Complex Task Breakdown:Useful for breaking down large projects into smaller, manageable phases.
• Bug Fixing:Can be set up to repeatedly try and fix bugs until resolved.
• Developing New Languages:Has been used to create entire programming languages.

Vibes

This technique has shown impressive results, including creating six code repositories in one night at a hackathon, securing a $50,000 contract for a low cost, and developing a full programming language over three months.