Azure Blob

Overview

The Azure Blob adapter maps each storagesdk bucket to an Azure container. Snapshots and forks create sibling containers populated by server-side Copy Blob operations. Same-account copies don’t require SAS tokens; the manifest lives at .storagesdk.metadata.json inside each container.

Configuration

import { Storage } from '@storagesdk/core';
import { azure } from '@storagesdk/adapters/azure';

const storage = new Storage({
  adapter: azure({
    bucket: 'agent-runs',
    accountName: process.env.AZURE_ACCOUNT_NAME,
    accountKey: process.env.AZURE_ACCOUNT_KEY,
  }),
});

azure({
  bucket: string;
  accountName: string;
  accountKey: string;
  endpoint?: string;       // override for Azure Gov, China, Azurite, etc.
});

Notes

• Peer dependency: @azure/storage-blob.
• For Entra ID, SAS tokens, or connection-string auth, construct a BlobServiceClient yourself and use storage.raw directly.
• maxSize / minSize on uploadUrl() are accepted but not enforced — Azure SAS URLs have no content-length-range equivalent. Cap server-side if you need a hard limit.
• uploadUrl() returns x-ms-blob-type: BlockBlob in headers so callers don’t need Azure-specific knowledge.
• Cross-container copies for snapshots/forks use syncCopyFromURL with a short-lived read SAS (real Azure requires source URL to self-authenticate; Azurite is lax).

Compatibility