ralph-loop

You can automate iterative tasks with fresh context by executing each iteration cleanly, using the filesystem as memory to track progress in PROGRESS.md and PRD.md. This approach prevents context pollution and ensures reliable results for multi-step software development.

Git versioning supports automated task execution by committing every completed task iteration to Git. This creates an audit trail and provides rollback capabilities, allowing you to revert to previous states if a multi-step task fails or produces errors.

The best way to manage external state for multi-step tasks is using the filesystem as memory. By storing requirements in PRD.md and tracking progress in PROGRESS.md, you maintain state across iterations without relying on the context window, ensuring reliable task automation.

Yes, you need Git installed to run this iterative task automation. The Skill requires Git and appropriate filesystem access to commit completed tasks, provide rollback capabilities, and maintain the audit trail required for multi-step software development workflows.

To start an iterative execution loop for a software module, initiate the task with a command specifying your objective, such as building an authentication module. The automation then handles the iterative execution with fresh context and filesystem memory tracking.

You should not use fresh context execution for simple, single-step tasks that do not require iterative development or external state tracking. It is designed specifically for complex, multi-step tasks needing rollback capabilities and filesystem memory to avoid context pollution.