Zip Files in S3 with Node.js and the S3zipper API

Node.js is a natural fit for a zip workflow: you trigger it from a web request and want the work to happen elsewhere. The S3zipper API lets you do exactly that — start a background job, poll it, and hand the signed download URL back to the browser.

1. Exchange your key for a token

const BASE = "https://api.s3zipper.com";

const tokRes = await fetch(`${BASE}/gentoken`, {
  method: "POST",
  body: new URLSearchParams({ userKey: process.env.S3Z_KEY, userSecret: process.env.S3Z_SECRET }),
});
const { token } = await tokRes.json();
const headers = { Authorization: `Bearer ${token}`, "Content-Type": "application/json" };

2. Start the zip

const start = await fetch(`${BASE}/v1/zipstart`, {
  method: "POST",
  headers,
  body: JSON.stringify({
    awsKey: process.env.AWS_ACCESS_KEY_ID,
    awsSecret: process.env.AWS_SECRET_ACCESS_KEY,
    awsBucket: "my-bucket",
    awsRegion: "us-east-1",
    filePaths: ["my-bucket/exports/2026-10"],
    zipFileName: "exports-2026-10.zip",
  }),
}).then((r) => r.json());

filePaths are full S3 keys — a folder prefix zips everything below it. The response contains a taskUUID and reports the total size.

3. Poll for completion

const sleep = (ms) => new Promise((r) => setTimeout(r, ms));

let state;
for (let i = 0; i < 120; i++) {
  state = await fetch(`${BASE}/v1/zipstate`, {
    method: "POST",
    headers,
    body: JSON.stringify(start),
  }).then((r) => r.json());

  if (state.State === "SUCCESS" || state.State === "FAILURE") break;
  await sleep(2000);
}

if (state.State !== "SUCCESS") throw new Error(state.Error || "zip failed");

4. Return the download URL

const result = await fetch(`${BASE}/v1/zipresult`, {
  method: "POST",
  headers,
  body: JSON.stringify(start),
}).then((r) => r.json());

const url = result.results[0]; // presigned S3 URL
res.redirect(url); // or return it as JSON

The URL is signed and expires (seven days by default). It is safe to hand to the user and safe to request again.

Tips

Keep the token for the process lifetime — it lasts 24 hours.
Do not run the loop on the request path for huge archives; return the task id to the client and poll from there, or use resultsEmail.
Store credentials in environment variables, never in code.