Import

You already have workflows. Start with those.

dagweave reads the workflows you already run, from a repo or straight off a live cluster, and hands them back as standard Argo Workflows YAML, ready to edit on the canvas.

The problem

Your workflows are already running. Not all of them are in a repo.

The ones that only exist on the cluster are hardest to hand to anyone.

dagweave pulls those out through the connector and puts them in your repo.

Getting them in

Bring them in from wherever they are

Some teams have a tidy repo of manifests. Some have twelve CronWorkflows that were kubectl-applied in 2023 and exist nowhere else.

where they come from text

git repo   ->  every manifest in the repo, in one pass
cluster    ->  what is actually running, through the connector
paste      ->  manifests, straight onto the canvas

What it can read

It opens manifests dagweave didn't write

Real manifests get written by people, over years. They use retry strategies, artifact repository references, pod spec patches, sidecars, a withParam loop three levels deep.

Your whole file opens, every template and every field.
An unusual manifest still opens, whatever is in it.
Change one field and publish. The diff your reviewer reads is that field, not a regenerated file.

dagweave writes the file out fresh, without the YAML comments. Keep your notes in the workflows.argoproj.io/description annotation.

fields from a real manifest yaml

retryStrategy:
  limit: "3"
  retryPolicy: OnTransientError
withParam: "{{steps.shards.outputs.result}}"
artifactRepositoryRef:
  configMap: artifact-repositories
  key: s3-prod

Afterwards

There is only one file, and it's yours.

An imported workflow doesn't become a dagweave object with an export button bolted on. It stays a standard Argo Workflows manifest, in your repo, in the format the controller already reads.

You're not maintaining a canvas version and a real version that drift apart. There's one file.
It runs where it ran before: your cluster, your Argo Workflows install, unchanged by the trip through the canvas.
Publish it, and a step matching something already in your registry folds into that template, not a fifth copy.
Stop using dagweave and there is nothing to migrate back.

The visual editor Where an imported workflow lands, and every field you can change on it. Connect your cluster The connector that reads live definitions back out of a running cluster. Your YAML, your repo Where it goes back to once you've changed something.

You already have workflows. Start with those.

Your workflows are already running. Not all of them are in a repo.

Bring them in from wherever they are

It opens manifests dagweave didn't write

There is only one file, and it's yours.

Bring the workflows you already run.