Skip to main content
Operator Field Notes

The Sample Post: How Ralph Workflow Stays Boring on Purpose

A walkthrough of the Ralph Workflow dispatch model — why every loop has the same five phases, why the runtime keeps you out of the agent's way, and how that boring consistency is the actual product.

Ralph Workflow blog ralph-workflow dispatch composable-loops field-manual

Codeberg-first

Ralph Workflow is free and open source. Inspect the primary repo on Codeberg before you install — or jump to the GitHub mirror.

Most AI coding tools want to surprise you. Ralph Workflow wants to be predictable enough that you stop noticing it. That is the whole pitch — and it is intentional, not accidental.

The dispatch model

Every Ralph Workflow loop moves through the same five phases: scope, plan, build, gate, dispatch. The CLI prints them in the terminal. The audit log records them. The UI shows them on the dossier ribbon. There is no clever sixth phase, no hidden mode, no surprise fork in the road.

If you have seen one Ralph Workflow run, you have seen every Ralph Workflow run. That is by design.

Why predictable is a feature

Predictability is what lets the loop run unattended. If the operator can predict what the agent will do at each phase, they can leave the loop running overnight without babysitting it. The next morning, the operator reads the morning-after dossier, decides whether the merged PR is acceptable, and either ships it or rolls it back.

That is the product. Not the model. Not the IDE. Not the cloud. The product is the boring, repeatable handoff between operator and agent.

What you do with this

Install Ralph Workflow once. Author one loop. Run it overnight. Read the dossier in the morning. Repeat. The first run feels slow. The tenth run feels like part of your job.

That is what we are building.

What is Loop Engineering?

Run an AI coding agent unattended, let it iterate on its own output, and wake up to finished software. The pattern is emerging across 30+ independent projects — here's what it is and why it works.

loop-engineering ai-coding

Best evaluator path

Turn the idea into a real overnight test, not another saved tab.

Codeberg-first: open the primary repo, star it to track releases, choose one bounded backlog task, run it tonight, and ask one question tomorrow morning — would I merge this? GitHub stays available as the mirror.

Open the primary Codeberg repo

Read the public source before you install anything.

Pick a first task

Use the guide to choose a bounded backlog item that is honest to review.

Install and run Ralph Workflow

Keep the machine awake, then decide in the morning whether the diff is good enough to merge.