devorq-auto

Story-by-story feature execution is automated by running a Ralph-inspired loop that processes iterative, isolated tasks with automatic context resets. The workflow generates prd.json from SPEC.md and verifies each task using check-story gates.

Yes, autonomous story verification requires dependencies like pytest, phpunit, jq, pint, shellcheck, python3, and npm installed in your environment to execute the check-story gates and validation steps properly.

The Ralph pattern works by executing a continuous story-loop that processes one isolated task at a time. It uses delegate_task retry mechanisms to handle failures and performs automatic context resets between iterations to maintain clean execution state.

Yes, you can automatically generate prd.json from SPEC.md as part of the autonomous execution workflow. This conversion happens at the start of the process to define story requirements before the iterative verification gates run.

The best way to capture lessons and progress is through the built-in learning capture mechanism of the story-loop. It optionally commits changes and pushes PRs while recording insights gained during each isolated, iterative task execution.

You should avoid using autonomous story-by-story automation for small, non-iterative tasks that do not benefit from isolated context resets. It is designed specifically for large feature work requiring iterative execution and verification gates.