Skip to main content
DocumentationNode Types

Node Types

This page has been cleared and is ready for updated Builder Studio documentation.

Every workflow in Builder Studio is built from nodes. Each node is a typed unit on the canvas with its own data, ports, and behavior. This page groups the node types by what they do and links to the reference page for each one.

Reference, not a glossary
Each node type has its own page with its real fields, ports, and limits. Use this page to find the right node, then open its reference for specifics.

Content nodes

Content nodes hold text you author directly. They are how a workflow starts with words instead of a model call.

  • Prompt — structured text input for reusable prompts and upstream instructions. Outputs its text so downstream generation and processing nodes can consume it.
  • Text — a static text layer for literals, notes, and labels placed directly on the canvas. It is a visual layer, not a runtime input.

AI generation and agent nodes

Generation nodes call a model provider to produce new content. They expose provider, model, and parameter controls and report status while they run.

  • Text Generation — general text generation for reasoning, chat, summarization, or code output, with streaming results.
  • Image Generation — image synthesis with provider-bound presets and generic image generation paths.
  • Video Generation — image-to-video and text-to-video flows.
  • Audio Generation — speech and voice generation for narrated output.
  • SFX Generation — non-voice sound effect generation.
  • 3D Generation — 3D asset generation, with provider-locked Tripo and Meshy variants of the same surface.
  • Agent — a longer-running agent runtime with tool use and traces.

Media nodes

Media nodes store or display media that already exists — uploads and generation outputs — and preserve a reference downstream nodes can reuse. They do not call a model themselves.

  • Image — upload, crop, export, and reuse image files.
  • Video — play, sync, and reuse video files with custom playback controls.
  • Audio — play, sync, and reuse audio files with custom playback controls.
  • 3D Model — display generated or uploaded 3D assets with interactive orbit controls.

Logic and control nodes

Logic nodes shape how a workflow flows — branching, combining, pausing, and repeating.

  • Conditional — branch workflow logic based on upstream values.
  • Merge — combine multiple upstream branches into one downstream flow.
  • Delay — pause execution for a specified duration.
  • Loop — iterate over a collection or repeat a sub-flow.

Utility and integration nodes

Utility nodes connect a workflow to the outside world or transform data as it passes through.

  • Webhook Call — make outbound HTTP requests to external APIs.
  • Webhook Trigger — create a dedicated HTTP entrypoint that starts a workflow.
  • OpenClaw Trigger — start a workflow from OpenClaw and Telegram-linked inbound events.
  • Transcribe — turn audio input into text output.
  • Browse Web — fetch, screenshot, and extract content from web pages.
  • HTML Viewer — render sandboxed HTML output directly on the canvas.

MCP Server Builder nodes

These node types are used inside the visual MCP server builder to define the tools, outputs, and credentials a published user MCP server exposes.

  • MCP Tool — define a workflow-backed tool surface a published MCP server can expose.
  • MCP Output — select the workflow value the published tool returns.
  • MCP Credential — configure an encrypted credential surface for tool handlers.
Availability varies by node
Some node types in this reference describe a node's intended purpose ahead of full canvas support. Each node page states what is available today rather than assuming every node is wired into the runtime. To see how nodes connect and run, read Nodes on the canvas, Edges, and Running workflows.

Was this page helpful?