JSPM

@ai2in/sdk

0.3.0
  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 296
  • Score
    100M100P100Q5701F
  • License MIT

JavaScript / TypeScript SDK for AI2IN — secure AI code sandboxes, hosted in India.

Package Exports

  • @ai2in/sdk
  • @ai2in/sdk/ai
  • @ai2in/sdk/langchain

Readme

@ai2in/sdk — JavaScript / TypeScript SDK

The developer client for AI2IN.dev — secure AI code sandboxes, hosted in India. Spin up an isolated Linux sandbox, run untrusted / AI‑generated Python in a stateful kernel, and get back rich, typed results (text, HTML tables, charts, JSON) — the E2B code‑interpreter model, in‑region.

Zero runtime dependencies (global fetch, Node 18+).

Install

npm install @ai2in/sdk          # once published
# or, from this repo:
cd sdk/js && npm install && npm run build

Quickstart

import { Sandbox } from "@ai2in/sdk";

const sbx = await Sandbox.create({
  apiKey: "ai2in_live_…",
  baseUrl: "https://api.ai2in.dev",
});

const execution = await sbx.runCode(
  "import pandas as pd; pd.DataFrame({'gst_lakh_cr': [1.87, 1.73, 1.95]})"
);

console.log(execution.text);          // the main result's plain text
console.log(execution.logs.stdout);   // ['…']

for (const r of execution.results) {  // rich, typed outputs
  if (r.html) save(r.html);           // e.g. a DataFrame → HTML table
  if (r.png) saveImage(r.png);        // e.g. a matplotlib figure → base64 PNG
}

if (execution.error) {                // structured error for self-correction
  console.error(execution.error.name, execution.error.value);
}

await sbx.kill();

State persists across runCode calls in the same sandbox (like a notebook):

await sbx.runCode("x = 41");
(await sbx.runCode("x + 1")).text;    // "42"

Streaming

Pass callbacks to stream output live as the kernel produces it (long loops, training logs, incremental prints). The events are still accumulated into the resolved Execution, so the return value is identical whether or not you stream:

const execution = await sbx.runCode(
  "for i in range(5):\n    print('step', i); import time; time.sleep(1)",
  {
    onStdout: (t) => process.stdout.write(t),  // arrives chunk-by-chunk
    onResult: (r) => render(r),                // display() calls, figures, the value
    onError:  (e) => console.error(e.value),
  }
);
console.log(execution.executionMs);            // still get the full Execution

Result model

runCode() returns an Execution:

Field Type Notes
results Result[] rich outputs; the cell's value has isMainResult: true
logs Logs .stdout / .stderr (string arrays)
error ExecutionError | null .name, .value, .traceback
text string | null convenience: the main result's text

Each Result may carry any of: text, html, markdown, svg, png, jpeg, pdf, latex, json, javascript.

Local dev

Point at a local engine (default http://localhost:4000):

const sbx = await Sandbox.create();            // talks to localhost:4000
console.log((await sbx.runCode("print(1 + 1)")).logs.stdout);

© 2026 AI2IN.dev — see the repository LICENSE.