Skip to main content

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.

Every workflow in Knouds has at least one way to receive data and at least one way to return results. The nodes on this page are the building blocks for that boundary: what your callers send in, what your pipeline returns, and how you embed static files directly on the canvas.

​
Request Input node

The Request Input node defines the fields that an API caller (or the Playground) must provide when running your workflow. Each field you configure here becomes a key in the JSON request body when your workflow is called at /api/workflows/:slug/run. For example, if you add a field named prompt, callers send: 
{
  "prompt": "A futuristic cityscape at night"
}
You can add as many fields as your workflow needs and drag them to reorder.

​
Field types

Click Add field on the node to open the field picker. You can choose from catalog fields (common parameters like prompt, aspect_ratio, resolution, num_images, duration) or create a Custom field with any name. When you create a custom field, you select its UI type:
TypeWhat it renders
TextSingle-line input
TextareaMulti-line text area
NumberNumeric input
SliderRange slider with min, max, and step
PillsButton group (you define the options)
ToggleOn/off switch
For image or video upload fields, the node renders a media uploader that accepts files and stores an HTTPS URL.

​
Output sockets

Each text or media field gets its own output socket on the right side of the node. Parameter fields (aspect ratio, resolution, count, duration, and similar) are grouped into a single params socket to keep the canvas tidy. Connect these sockets to the matching inputs on downstream nodes — for example, wire the prompt socket to the prompt-in socket of an Image Generator.

​
Field settings

Click the gear icon on any field to open its constraint panel. You can:
  • Restrict allowed options on pill fields (or let them auto-sync from the connected downstream node)
  • Set min, max, and step for number and slider fields
  • Write helper text that appears below the field in the Playground
  • Override the playground label shown to end users
  • Toggle whether the field shows on Playground
When you connect a Request Input field to an Image or Video Generator, the node automatically reads the generator’s allowed values and shows an AUTO badge on the field. Options stay in sync as you switch models.

​
Response Output node

The Response Output node collects the final result of your workflow and returns it to the API caller. Connect the output socket of any generator or assistant to this node. The node accepts three input types on its left side:
  • image — renders a preview thumbnail after a run and returns the image URL
  • video — returns the video URL
  • text — returns text (for example, from an Assistant node)

​
Property name

The Property name field sets the JSON key under which the result appears in the API response. The default is output. 
{
  "output": "https://..."
}
After a successful run, the node previews the result directly on the canvas — image results show a thumbnail, text results show the first 400 characters.

​
Media Upload node

The Media Upload node lets you embed a static file — an image, video, or audio clip — directly in your workflow. You upload the file once and it becomes part of the saved workflow. This is useful for fixed references that every run should use: a brand logo for style transfer, a background audio track, or a sample image you always want to include.
Media Upload is not exposed to API callers. The file is baked into the workflow. If callers need to supply their own file on each run, use a Request Input node with an image or video upload field instead.

​
Output sockets

The sockets that appear depend on what you upload:
Uploaded fileAvailable outputs
Imageimage-out
Videovideo-out, start-out (first frame), end-out (last frame), audio-out
Audioaudio-out
You can replace the file at any time with the Replace button. The node auto-detects the file type from the MIME type — you never have to specify it manually.

​
Frame Extractor node

The Frame Extractor node takes a video and extracts individual frames as image outputs. Connect a video source to it and use the extracted frames as reference images for downstream nodes — for example, feeding the first frame of a generated video into an Image Generator as a style reference, or chaining a frame into a second Video Generator to extend a clip. Use Frame Extractor when you need more control over which frame is extracted, or when the video is produced dynamically during a run rather than uploaded ahead of time. For static uploaded videos, the Media Upload node already provides start-out and end-out sockets as a shortcut.