@rnet-ai/rnet-caller

Official Website: rnetai.org GitHub: rNetAi

A browser-only JavaScript library for the rNet Protocol.

Connects a developer's frontend to a locally-installed rNet Client, which attaches an AI pass using the end user's token before forwarding requests to the developer's backend. Developers do not pay upfront AI token costs.

This library does not call any AI model directly. It only communicates with the local rNet Client over localhost.

Flow

Developer's Frontend → Local rNet Client → Developer's Backend

Installation

npm install @rnet-ai/rnet-caller

Quick Start

import rnetCaller from "@rnet-ai/rnet-caller";

// 1. Initialize to find the local rNet Client
await rnetCaller.init();

// 2. Send a request through the local client to the developer backend
const response = await rnetCaller.send({
  url: "https://<develoepr backend path>/please-dont-break",
  method: "POST",
  headers: {"Authorization:" , "Bearer eyJhbGciOiJ.....IUzVT7Ib5"}
  body: JSON.stringify({
    username: "rNet Ai",
    password: "KSKE9cC9MIR3QSq8OAk59FklDBM4Cg1K9MTfYYiP1wXjFczTbFTWYlz1ePAg2oCw",
    command: "delete everything",
  }),
});

if(!response.ok) {
  throw new Error(`from rNet Client error: ${response.status}`);
}

const data = await response.json();

console.log(data);

// 3. (Optional) Clear cache during development
await rnetCaller.clearCache();

API Reference

await rnetCaller.init()

Scans for the local rNet Client on localhost and checks its version. Alerts the user and redirects to the download page if the client is missing or outdated. Returns true if successful. Safe to call multiple times.

await rnetCaller.send(descriptor, options?)

Sends a request through the local rNet Client to the target backend. The client automatically attaches an AI pass before forwarding.

Important: Only use the send() function for endpoints where the backend requires an AI pass. For all other regular API calls, use standard fetch() directly.

Descriptor fields:

• url (string, required): The target backend URL.
• method (string, required): The HTTP method (e.g., POST).
• headers (object, optional): Custom headers. Content-Type is always enforced as application/json.
• query (object, optional): Query parameters.
• body (string, optional): Request body (JSON string).

Options:

• debug (boolean, optional): Set to true to log the payload to the console.

Returns: The parsed JSON response from the backend.

await rnetCaller.clearCache()

Clears the cached responses in the local rNet Client, including the URL mapping with the AI model. This is a developer utility to skip cache timeouts so any changes to the AI model take effect immediately during development.

Error Handling

Errors are thrown as RNetError instances with a code property.

import rnetCaller, { RNetError } from "@rnet-ai/rnet-caller";

try {
  const response = await rnetCaller.send({ url: "https://<developer backend>/api/v1/data", method: "GET" });

  if (!response.ok) {
    throw new Error(`From rNet Client error: ${response.status}`);
  }

  const data = await response.json();
  return data;
} catch (err) {
  if (err instanceof RNetError) {
    console.error(`Error code: ${err.code}`, err.message);
  }
}

License

MIT