> ## Documentation Index
> Fetch the complete documentation index at: https://docs.knouds.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Workflows: save, organize, and deploy your pipelines

> A workflow is a saved pipeline with a unique slug, a set of nodes and edges, and a viewport. Deploy any workflow as a REST API endpoint.

A workflow is the saved form of a pipeline you build on the canvas. It stores everything Knouds needs to reproduce and run your pipeline: the nodes, the wires between them, the settings on each node, and the position of your viewport. Once saved, a workflow can be run interactively on the canvas, shared as a playground link, or called programmatically through a REST API.

## Workflow structure

Every workflow is stored as a JSON document with the following top-level fields:

| Field         | Description                                                                              |
| ------------- | ---------------------------------------------------------------------------------------- |
| `slug`        | URL-safe identifier used in API paths and file storage. Immutable after creation.        |
| `name`        | Human-readable display name. You can rename this at any time without affecting the slug. |
| `nodes`       | Array of node definitions — type, position, and all field values.                        |
| `edges`       | Array of connections between node handles.                                               |
| `viewport`    | Saved pan position and zoom level (`x`, `y`, `zoom`).                                    |
| `description` | Optional free-text description of what the workflow does.                                |
| `tags`        | Optional array of labels for organizing workflows.                                       |

## The workflow slug

<Warning>
  The slug is set at creation time and **cannot be changed later**. It is used as the disk filename, the API path (`/api/workflows/{slug}/run`), and the usage log foreign key. Name your workflow carefully when you create it — the name you enter becomes the slug.
</Warning>

If you leave the name blank when creating a workflow, it defaults to "Untitled workflow" and receives a slug like `untitled-workflow-2`. You can rename the display name at any time, but the slug stays fixed.

## Saving and loading workflows

Knouds auto-saves your workflow as you work. You can also save manually from the toolbar. Workflows are loaded from the editor URL: `https://knouds.ai/app/editor/{slug}`.

## Organizing with folders

You can move workflows into folders from the home page. Folders support nesting — you can create folders inside folders to build a hierarchy. The home page shows only root-level workflows and folders; navigate into a folder to see its contents.

## Templates

Knouds provides pre-built workflow templates you can copy to start from a working pipeline instead of a blank canvas.

<Note>
  Some templates use models that require a higher-tier plan. Templates gated to a higher tier display a lock badge. Clicking the badge shows which plan you need to access that template.
</Note>

## Running a workflow

You have two ways to run a workflow:

<Steps>
  <Step title="Run on the canvas">
    Press the **Run** button in the canvas toolbar to execute the full workflow interactively. You can also right-click individual nodes to run a single node in isolation.
  </Step>

  <Step title="Call via API">
    Every workflow is available as a REST endpoint. Send a `POST` request to `/api/workflows/{slug}/run` with your API key and a JSON body matching the fields defined in your Request Input node.
  </Step>
</Steps>

## The Request Input node

The **Request Input** node defines the API surface of your workflow. Each field you add to a Request Input node becomes a key in the JSON body that callers send when hitting the API endpoint. When running on the canvas, you fill in these fields directly on the node.

## Playground

Each workflow can be shared as a **Playground** link — a hosted page where anyone can try the workflow without writing code. You control whether the playground is public or private from the workflow settings.

<Tip>
  The Playground is a great way to let non-technical stakeholders test a workflow before you integrate it into your application.
</Tip>