Circuitry Logo
Circuitry
Open Circuitry

Code Generation

Use AI to export the logic of your visual workflows as code in 13 programming languages.

Generation is AI-powered, so output quality depends on the model you choose. JavaScript, TypeScript, and Python are the most thoroughly tested — the other languages are generated best-effort, so review before you ship.

Visual to Code

Design complex workflows visually, export as clean, executable code

True Parallelism

Fork/join nodes generate native parallel execution patterns

Helper Libraries

Complete runtime libraries for HTTP, AI, UI interactions, and more

Supported Languages

JS

JavaScript

InlinePromise.all()

Native JavaScript execution with inline Code nodes

TS

TypeScript

InlinePromise.all()

Type-safe JavaScript with full TypeScript support

PY

Python

executeJavaScript()asyncio.gather()

Python with async/await and PyExecJS for Code nodes

GO

Go

ExecuteJavaScript()Goroutines + WaitGroup

High-performance Go with goroutines and WaitGroups

C

C

execute_javascript()pthread (optional)

System-level C with optional pthread parallelism

JV

Java

executeJavaScript()CompletableFuture.allOf()

Enterprise Java with CompletableFuture parallelism

Walkthrough coming soon.

RS

Rust

execute_javascript()tokio::join!

Memory-safe Rust with Tokio async runtime

Walkthrough coming soon.

C++

C++

executeJavaScript()std::async

Modern C++ with std::async parallelism

Walkthrough coming soon.

C#

C#

ExecuteJavaScript()Task.WhenAll

.NET C# with Task.WhenAll parallelism

Walkthrough coming soon.

PHP

PHP

executeJavaScript()Sequential

PHP with sequential execution and JavaScript support

Walkthrough coming soon.

RB

Ruby

execute_javascript()concurrent-ruby

Ruby with concurrent-ruby for parallel execution

Walkthrough coming soon.

KT

Kotlin

executeJavaScript()Coroutines

JVM Kotlin with coroutines for async operations

Walkthrough coming soon.

SW

Swift

executeJavaScript()async let

iOS/macOS Swift with async let parallelism

Walkthrough coming soon.

What gets turned into code

Code generation exports the runnable logic of your workflow — the same parts that run when a workflow is triggered headlessly by a webhook. Editor-only and interactive nodes are intentionally not generated: they exist to help you build and visualize the flow, not to run on a server or inside your project.

✅ Generated

Control flow, compute, and integrations:

  • • Start / Webhook trigger
  • • Condition, Loop, Fork, Join, Then
  • • Code (JavaScript / Python)
  • • Agent (cloud or your-own-key models)
  • • HTTP / Action requests
  • • Database, Message, Save / Files / Report / PDF
  • • Vector store, Plugin, Webhook Response

🟡 Skipped — editor-only

Visualization / annotation nodes. Data flows straight through them:

  • • Charts, Data viz, Stat plot
  • • Image, SVG, 3D scene, ToVec, Vision
  • • Web page, Chat, Sheet / Data grid
  • • Sticky note, Group, Flow, Mind map

🔴 Skipped — needs a person/device

Interactive or machine-bound nodes can't run in a standalone script:

  • • Confirm / Form / Message / Choice / File-upload dialogs
  • • Terminal
  • • Shell-language Code nodes
  • • On-device / localhost AI models

This is the same classification used for webhook/headless runs, so the generated code behaves like your deployed workflow. Tip: build the parts you want exported out of the “Generated” node types — the visual nodes are free to use for design and won’t affect output.

Quick Start Guide

1. Generate Code

  • • Open your workflow in Circuitry
  • • Click "Generate Code"
  • • Choose your target language and AI model
  • • Keep "Include Helper Libraries" on (some languages need them)
  • • Download the generated code + helpers

2. Setup & Run

  • • Install language dependencies
  • • Set environment variables (API keys)
  • • Import generated functions
  • • Call with your input data
  • • Deploy to your platform