EdgazeDocsEdgaze Docs
DocumentationAPI Reference
Home
Core
Workflow InputWorkflow FormWorkflow OutputMergeMerge ObjectsTemplateMap
AI
LLM ChatLLM ImageLLM Embeddings
Logic & integration
HTTP RequestWeb ScrapeYouTube TranscriptJSON ParseConditionDelayLoopRun Another Workflow (Remix)

Platform status

Checking platform status
Block reference

Workflow Input

Ask the customer one clear question and turn their answer into data your workflow can use.

Audience

Creators building and publishing workflows.

Overview#

Workflow Input is where customer data enters your product. It shows a question in the run modal and passes the answer downstream as structured data.

Use it whenever your workflow starts with something the customer must provide: a prompt, a URL, a file, or a choice from a list.

Workflow Input

Press enter or space to select a node. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Ports#

  • Output: Data: The customer's answer, ready for the next block. No inputs; this block starts the flow.

Inspector#

  • Question: What you ask the customer. Keep it direct and outcome-focused.
  • Input Type: How the answer is collected: text, long paragraph, number, URL, dropdown, file upload, or JSON.
  • Description: Optional help text shown under the question, such as examples, limits, or format hints.
  • Dropdown options: When Input Type is Dropdown, define the choices the customer sees.

File uploads#

A file input accepts documents, images, and any text-based file:

KindAcceptedMax size
PDF.pdf25 MB
Office.docx, .xlsx, .pptx25 MB
Text and code.txt, .md, .csv, .json, .yaml, .ts, .py, .sql, etc.10 MB
Image.png, .jpg, .webp, .gif5 MB

Every upload is identified by its actual contents, not its extension, so a renamed file is rejected rather than silently mishandled. Images larger than the limit are compressed in the browser before upload.

Downstream AI blocks receive each file in the best form the chosen model supports. Images go to the model natively. PDFs go natively to models that read PDFs (Claude, Gemini, and OpenAI's document-capable models) and are otherwise text-extracted. Office, text, and code files are always extracted to text and appended to the prompt under a heading naming the file. Extraction is capped at 200,000 characters per file, and the model is told when a file was truncated, had no text layer (a scanned PDF, for example), or could not be read.

When you publish, Max file size in pricing bounds what customers may upload for this workflow. Lower it if your workflow only needs small files.

Tips#

  • Ask for only what the workflow actually needs.
  • Match the input type to the answer shape; do not use JSON when a short text field is clearer.
  • Use the description for examples or constraints, not to repeat the question.
  • Place this block at the start of any branch that depends on customer input.

Related#

  • Workflow Studio

Was this useful?

Your response helps us improve the documentation.

More blocks

Workflow Form

Collect several run-modal fields in one grouped form, each with a stable key for downstream steps.

Workflow Output

Define what the customer receives when the workflow finishes.

Merge

Combine several upstream values into one payload for the next step.

On this page
OverviewPortsInspectorFile uploadsTipsRelated
© 2026 Edge Platforms, Inc. All rights reserved.