ralph-loop-kiro-specs

A self-correcting agent loop drives Kiro spec implementation by running kiro-cli iteratively, selecting one top-level task per cycle while tracking progress, logging corrections, and verifying exit criteria for multi-task specifications.

Self-correction in spec-driven workflows logs task errors, codebase patterns, and timing data per iteration to refine subsequent agent runs. This mechanism enables automatic recovery from common failures while maintaining implementation discipline.

Yes, Kiro specification automation requires kiro-cli and Bash to execute the iterative agent loop. Optional Kiro steering files provide additional guidance to monitor and control the sequential task processing workflow.

Track Kiro specification progress by monitoring task status, implementation details, and per-task timing. Upon completion, the workflow generates a self-contained dashboard summary page displaying all task statuses and timing information.

The automated agent loop for spec implementation runs up to 15 iterations per specification. While it handles self-correction and exit-criteria verification, complex specifications may require manual oversight if failures exceed the loop's automated recovery capabilities.