Catalog

Pick a catalog step. Skip the manifest.

The catalog covers the parts every pipeline needs: S3, Postgres, kubectl, Helm, docker builds, LLM calls. Pick one, fill in a form, and it's a working step on your canvas.

The problem

The S3 upload step is already written.

Upload a file, query Postgres, post to Slack. The catalog has them, versioned and pinned by digest.

Catalog items

The steps are already written

A catalog step arrives on the canvas with its fields already declared, straight from its own schema.

Every image is pinned by digest, not a floating tag. A workflow you shipped keeps the image it shipped with.
A new digest arrives as a new item version, so an update is something you pick up rather than something that lands on you.
It's yours to edit, not a black box.

what a pick actually doestext

pick     ->  s3-upload  v1.0.0
fill     ->  bucket, key, file   (from the item's own schema)
write    ->  a WorkflowTemplate in your registry
compile  ->  templateRef: s3-upload-v1

Blueprints

A whole workflow, not just a step

Catalog items are single steps. A blueprint is a complete starter workflow, laid out on the canvas with its dependencies already drawn.

They're ordinary graphs. Every step is editable as soon as it opens.

Marketplace

The templates your team writes

Publish a WorkflowTemplate your team wrote. Another team installs it into their own registry, where it becomes an ordinary versioned template of theirs.

Install what another team already got working.

publish and installtext

publish  ->  your WorkflowTemplate, versioned, visible
install  ->  a copy written into your registry, yours to edit
compile  ->  templateRef, same as any other template

Start with a step that's already written.

dagweave is in early access. Leave an email and we will bring you on as we open it up.