LoopFlow automates Claude Code with looping workflows

LoopFlow turns Claude Code into an autonomous system that executes tasks on a set scenario. Users describe the goal, agent sequence, and validation conditions in a YAML file. The system then iteratively works toward the result until successful or budget exhaustion.

The LoopFlow architecture addresses three automation challenges: self-checking, cost control, and memory loss. It uses separate agents for code generation and validation, requires a mandatory PASS verdict from the validator, and keeps a Markdown history log to preserve context between runs.

The tool supports git workflows for change isolation and allows setting a dollar budget for each step. LoopFlow operates without daemons or cloud services.