Installing and Importing the dragonJSON Client Library

dragonJSON is a single-file ES module (client/web-client.js) with zero external dependencies — there is nothing to install and no build step required. Drop the file into your project, point an import at it, and you are ready to go. The Node.js server reference implementation is equally self-contained, relying only on Node’s built-in http and url modules.

Client setup

Copy file
Direct import (browser)

Copy client/web-client.js from the dragonJSON repository into your project, then import it as a standard ES module:

import dragonJSON from './web-client.js';

const [server, control] = dragonJSON("https://your-api.example.com", {
  enableBatching: true,
});

Because web-client.js uses only standard browser APIs (fetch and Proxy), no bundler or transpiler is needed. It works as-is in any modern browser or in Node.js 18+.

In a browser environment you can load the file without copying it first. Use a <script type="module"> tag pointing at the raw file URL:

<script type="module">
  import dragonJSON from 'https://example.com/path/to/web-client.js';

  const [server] = dragonJSON("https://your-api.example.com");
  const title = await server.posts.page1.title;
  console.log(title);
</script>

Or use a dynamic import() from inside an existing module:

const { default: dragonJSON } = await import('https://example.com/path/to/web-client.js');

Server setup

The repository ships a ready-to-run Node.js reference server at server/nodejs/server.js. Start the server:

node server/nodejs/server.js

Override the default port (3000):

PORT=8080 node server/nodejs/server.js

No npm install is required. The server uses only Node.js built-in modules (http and url). Swapping in your own data store: The reference server uses a simple in-memory object. To connect dragonJSON to a real database, replace the three data-access functions:

Function	Responsibility
`getByPath(pathArray)`	Read a value from the store at the given path.
`setByPath(pathArray, value)`	Write or update a value at the given path.
`deleteByPath(pathArray)`	Remove the entry at the given path.

These are the only functions that interact with storage; everything else — request routing, batch handling, invalidation responses — is handled by the server framework layer and does not need to change.

Browser compatibility

The dragonJSON client relies on two standard platform features:

fetch — the browser Fetch API for HTTP requests.
Proxy — the ES2015 Proxy object for path interception.

Both are supported in all modern evergreen browsers and in Node.js 18+. No polyfills are needed for current browser versions. If you need to support older environments, a fetch polyfill (such as whatwg-fetch) is sufficient for the network layer; Proxy cannot be polyfilled and requires a natively compliant runtime.

Writing your own server

The dragonJSON wire protocol is language-agnostic. Any HTTP server that speaks JSON can act as a dragonJSON backend as long as it honours the expected request shapes, response formats, batch protocol, and invalidate contract. See the Server Specification for the full contract, including the __more, __batch, and mutation response conventions.

Property names in your data must not contain ., cannot be then or exists, and cannot start with $. These names are reserved by the dragonJSON proxy for path separation, promise interception, and built-in methods respectively.

Get Started

Core Concepts

Guides

Installing and Importing the dragonJSON Client Library

Client setup

Server setup

Browser compatibility

Writing your own server

Build docs developers (and LLMs) love

Get Started

Core Concepts

Guides

Documentation Index

​Client setup

​Server setup

​Browser compatibility

​Writing your own server

Build docs developers (and LLMs) love

Client setup

Server setup

Browser compatibility

Writing your own server