self-improve:ralph-loop

Iterative code refinement is automated by running a stable prompt repeatedly against a workspace, allowing file changes to accumulate between iterations until a completion promise or iteration limit is met.

A Ralph loop is an iterative execution mechanism that repeatedly applies a prompt to a workspace, accumulating file changes across multiple runs to progressively refine deterministic code generation tasks.

Set up repeated prompt execution by providing a workspace, a prompt file, and a completion promise, which allows the system to run stable prompts while evolving file states accumulate changes.

Yes, you can control automated code testing loops using iteration limits, which halt execution alongside defined completion promises or cancellation requests to prevent infinite refinement cycles.

Use iterative model runs for code generation when deterministic tasks benefit from refinement, requiring repeated execution with stable prompts where each iteration builds upon previous file changes to meet performance benchmarks.

Iterative workspace execution supports structured iteration reports and case-driven todo lists, providing organized outputs that track the progress of file changes accumulated across refinement loops.