Cloudflare
R2.

Declare an R2 bucket in wrangler.toml and access it via env.BUCKET. The API supports put, get, delete, list, and head.

# wrangler.toml
[[r2_buckets]]
binding = "BUCKET"
bucket_name = "uploads"

// Worker code
export default {
  async fetch(req, env) {
    // Upload
    await env.BUCKET.put("images/photo.jpg", req.body, {
      httpMetadata: { contentType: "image/jpeg" },
    });

    // Download
    const obj = await env.BUCKET.get("images/photo.jpg");
    return new Response(obj.body, {
      headers: { "Content-Type": obj.httpMetadata.contentType },
    });
  }
};

Use any S3 client with R2's endpoint. Useful for migrations from AWS, or for tools that don't know about Cloudflare.

# AWS CLI with R2 endpoint
aws s3 cp ./backup.tar.gz \
  s3://my-bucket/backups/backup.tar.gz \
  --endpoint-url https://<account-id>.r2.cloudflarestorage.com

# rclone sync
rclone sync ./dist r2:my-bucket/site/ --progress

Generate a presigned URL from your Worker, hand it to the client, and the client uploads directly to R2 without routing through your Worker. Saves compute and keeps large files off your Worker's memory.

import { AwsClient } from "aws4fetch";

const r2 = new AwsClient({
  accessKeyId: env.R2_ACCESS_KEY,
  secretAccessKey: env.R2_SECRET_KEY,
});

const signed = await r2.sign(
  new Request(`https://${ACCOUNT}.r2.cloudflarestorage.com/bucket/key`, {
    method: "PUT",
  }),
  { aws: { signQuery: true } }
);