operating-loop-workflowlisted

# $operating-loop-workflow > **One job:** Run the operating-loop seven-move loop (shape → plan → pre-flight → implement → capture) in Codex. **YOU MUST EXECUTE THIS WORKFLOW. Do not just describe it.** In the Codex runtime the operating-loop seven moves ARE the `$rpi` skill chain — there is no separate engine to install. `$rpi` composes the loop end to end. ## Execution 1. Run `$rpi --auto "<the capability/intent to drive through the loop>"`. - `$rpi` chains `$discovery` (shape → plan → pre-mortem) → `$crank` (TDD per slice across conflict-free waves) → `$validation` (acceptance roll-up + capture). - For the plan-only half (shape → plan → pre-flight, no implementation), run `$discovery "<intent>"` and stop after `$pre-mortem`. 2. Report the resulting plan, the `$pre-mortem` verdict, and the `$validation` outcome. ## Guardrails 1. Treat `$rpi` as the canonical seven-move loop in Codex. 2. Preserve each move's evidence (plan, pre-mortem verdict, validation result) before advancing.