Manage Azure Blob Storage Data

This skill equips Claude with @azure/storage-blob (v12.x, Node.js >= 18) for Azure Blob Storage in TypeScript/JavaScript: container CRUD, blob CRUD across three blob types (block, append, page), copying blobs server-side, generating blob/container/account-scoped SAS tokens, and reading or setting blob metadata and HTTP headers.

Use it when writing frontend or Node.js code that stores or serves files on Azure Blob Storage - creating a container (create() or the idempotent createIfNotExists()), listing containers with an optional name prefix, deleting a container (delete()/deleteIfExists()), uploading from a Node stream/file/buffer or a browser File/ArrayBuffer, downloading to a file, buffer, or string, listing blobs flat or by virtual-directory hierarchy, deleting a blob (optionally including its snapshots with deleteSnapshots: "include"), copying a blob server-side with beginCopyFromURL and polling until done, appending log lines to an append blob with appendBlock, or working with page blobs for VHD-style random read/write, where pages must be created and written in 512-byte-aligned chunks. It distinguishes platform capability sharply: uploadFile, uploadStream, downloadToFile, downloadToBuffer, SAS generation, and DefaultAzureCredential are Node.js-only, while uploadData and anonymous/SAS-URL access work in the browser too. Its own best-practice guidance favors AAD (DefaultAzureCredential) over connection strings/keys, streaming for files over 256MB, short-lived SAS tokens for browser uploads, and *IfNotExists methods for idempotent creation. It is specific to blob storage - not Table, Queue, or File Storage.

Inputs: an account URL with DefaultAzureCredential, a connection string (BlobServiceClient.fromConnectionString), a StorageSharedKeyCredential, or a SAS token appended to the account URL, through the client hierarchy BlobServiceClient -> ContainerClient -> BlobClient/BlockBlobClient/AppendBlobClient/PageBlobClient; upload data as a string, Buffer, file path, stream, or browser File/ArrayBuffer. Outputs: a download response with a readable stream (read to a string via a manual chunk-concatenation helper, or straight to a file/buffer on Node), BlobProperties (contentType, contentLength, lastModified, etag), async iterables of BlobItem/ContainerItem from list calls (listContainers, listBlobsFlat, listBlobsByHierarchy), and SAS query-parameter strings appended to a blob/container/account URL.

Built on @azure/identity for AAD auth. SAS generation covers three scopes - BlobSASPermissions/generateBlobSASQueryParameters for a single blob, ContainerSASPermissions for a container (e.g. read/add/create/write/delete/list), and AccountSASPermissions/AccountSASServices/AccountSASResourceTypes for account-wide access - each keyed by a StorageSharedKeyCredential and an expiry. Errors surface as RestError with a statusCode (404 not found, 409 already exists, 403 access denied). setMetadata attaches arbitrary key-value pairs to a blob, and setHTTPHeaders sets blobContentType, blobCacheControl, and blobContentDisposition for browser-served content. Configuration reads AZURE_STORAGE_ACCOUNT_NAME/AZURE_STORAGE_ACCOUNT_KEY or AZURE_STORAGE_CONNECTION_STRING from the environment.

npm install @azure/storage-blob @azure/identity

Frontend and Node.js developers building file upload/download, static asset serving, append-only logging, or VHD-style random-access storage on Azure Blob Storage, in both server and browser contexts - and who follow the skill's own guidance to check RestError.statusCode for specific handling and to close long-running clients as good practice, even though it is not strictly required.