Open source · desktop app & VS Code extension

DataWeave Studio.

Write, run and debug DataWeave 2.0 transforms locally — no 2 GB Anypoint Studio, no Docker container, and no pasting company data into a web playground. Context-aware autocomplete, visual message flows, real secure-property decryption and a 309-function reference. Run it from the Microsoft Store, as a standalone desktop app, or right inside VS Code — no installer, no admin rights.

Presets
Input payload
DataWeave script
Output
Forging output…
WINDOWS · MACOS · LINUX · VS CODE LATEST DESKTOP:
0+
DataWeave functions indexed offline
0 OS
Native desktop builds + a VS Code extension
0%
Offline — payloads never leave your machine
MIT
Free & open source, forever
01 — Features

Everything the playground
should have been.

A compact, correct DataWeave workbench built to sidestep Anypoint bloat — not a watered-down web toy.

// new · vs code
Now on the Marketplace

The whole engine, right inside VS Code

No desktop installer fighting your company's IT policy, no admin rights, no "unknown publisher" prompt. The full DataWeave 2.11 engine ships as a VS Code extension — install it from the Marketplace and get the same autocomplete, secure tools and live playground in the editor you already work in.

Install from the VS Code Marketplace →
// share
New

Send a whole setup in one link

Copy a link that carries the script, payload, variables, headers and query params — one request or the entire workspace. Whoever opens it gets an identical setup and can press Run, instead of a snippet they have to rebuild. The data rides in the part of the URL browsers never send to a server, so nothing is uploaded to create one.

Editor

Monaco script engine

Suggestions drawn from your own payload, variables and config — plus every reader and writer option for 16 formats, so output application/json  completes instead of leaving you guessing. Syntax colouring, bracket pairing and inline errors as you type.

// canvas
Visual

Interactive message-flow designer

Drag connectors onto a canvas, set payload contexts, map HTTP, Salesforce and database steps, then step through node by node and inspect the payload, variables and attributes at each stage — the structure of a real flow without spinning up a runtime.

Specs

OpenAPI & Swagger reader

Open a 3.x or 2.0 spec, browse its operations and types, and drop a sample payload plus a matching DataWeave skeleton straight into the workspace. Resolves $refs, enums and allOf/oneOf. Entirely offline.

// mcp
AI

Your AI assistant can run DataWeave

A local MCP server lets Claude, Cursor or any MCP client compile and run transforms against the real engine — so it checks its DataWeave instead of confidently inventing it. Bound to loopback, started only when you start it, and nothing ever leaves the machine.

Reference

309 functions and the official cookbook

The complete function reference, searchable and offline, docked in the workspace. Plus MuleSoft’s cookbook examples — each one executed against the bundled engine and kept only if it actually runs, so nothing you open is a broken snippet.

// secrets
Offline by design

Encrypt secure properties without a website

The hosted encryption tools are websites — they go down, and plenty of corporate networks block them outright. This one is a dialog in the app: AES, Blowfish, DES, DESede and RC2 across CBC/CFB/ECB/OFB, byte-for-byte compatible with Mule’s secure-properties-tool. Paste a real secure-config.yaml with ![…] values and run scripts against the decrypted result. The key is held for the session and never written to disk — which is rather the point, because a production secret should never go into someone else’s web form.

Import

cURL importer

Paste a raw curl string from Postman or devtools and get method, headers, query params and body filled in, with a matching DataWeave scaffold for whatever format the body turned out to be.

// correctness
Correctness

Tests, and a diff that ignores noise

Snapshot a transform’s output as its expected result and re-run it as a test suite. Compare two payloads side by side — with an Ignore IDs toggle that blanks doc:id and UUID churn so you see the change that actually matters.

// runtime
Self-contained

Bundled DataWeave engine & JRE 17

Works the moment the installer finishes. Ships with JRE 17 and the native Mule runtime engine — no system PATH, no JDK setup, no global config to break.

Workspaces

Postman-style collections

Group related transforms in one workspace — each request keeps its own script, payload, context, named inputs and tests. Search inside them, pin the ones you live in, and resume exactly where you left off.

Themes

Dark & paper themes

Switch between a high-contrast dark space and a clean paper-light layout to match the rest of your environment.

03 — Workflow

Two ways to build.

Write transforms directly, or model them visually — then step through and ship.

01

Script mode

Draft transforms directly with compiler-backed autocomplete and compile instantly against real input.

02

Flow canvas

Chain elements in sequence, modelling input/output properties and payload routing with drag-and-drop connectors.

03

Step debugger

Walk the pipeline node by node — inspect variables, header metadata and intermediate output states in order.

04

Workspace sync

Save configurations as lightweight project packages. Resume later or hand them to a teammate instantly.

04 — Download

Get DataWeave Studio

Open source and fully local — payloads never leave your machine. Three ways to run it: from the Microsoft Store, as a VS Code extension, or as a standalone desktop app.

Microsoft Store windows

Signed by Microsoft, so there’s no SmartScreen warning and no “unknown publisher” prompt. Updates arrive through the Store.


VS Code extension recommended

No installer, no admin rights, no code-signing prompt — the easiest way onto a locked-down work machine. Always the latest on the Marketplace.


Desktop app

Standalone build for Windows, macOS & Linux. The installers aren't code-signed yet, so your OS may warn about an "unknown publisher" on first launch — that's expected: choose More info → Run anyway on Windows, or right-click → Open on macOS.