Manage Azure Queue Storage with Python

This skill equips Claude with the Azure Queue Storage Python SDK (azure-storage-queue) for simple, cost-effective message queuing: creating and listing queues, sending messages (including delayed and expiring ones), receiving and deleting them, peeking without affecting visibility, updating a message's content or visibility timeout, and clearing a queue entirely.

Use it when writing Python code for asynchronous task distribution or decoupled processing on Azure - sending a JSON task payload, receiving messages with a visibility timeout so other consumers don't grab the same one, deleting a message only after it's successfully processed, checking dequeue_count to detect a poison message that keeps failing, extending a message's visibility timeout with update_message when processing needs more time, or clearing an entire queue with clear_messages. It handles binary payloads via BinaryBase64EncodePolicy/BinaryBase64DecodePolicy. Its own guidance says to reach for Azure Service Bus instead when the workload needs advanced features such as sessions or topics - Queue Storage is the simple, low-cost option, not the full-featured enterprise messaging one.

Inputs: an account URL plus DefaultAzureCredential, a queue name (via QueueServiceClient.get_queue_client() or a direct QueueClient), message content as a string, JSON-dumped payload, or (with the base64 policies) raw bytes, and per-call options like visibility_timeout and time_to_live. Outputs: received message objects with .id, .content, and .dequeue_count; queue properties with approximate_message_count and custom metadata; and peeked messages (via peek_messages(max_messages=...)) that leave the queue's visibility state untouched, unlike receive_messages, which hides each returned message for its visibility_timeout window. update_message can also rewrite a message's content in the same call that extends its timeout.

Built on azure-identity for DefaultAzureCredential authentication, with an async mirror (azure.storage.queue.aio, azure.identity.aio) usable in an async with block for high-throughput scenarios. Queue-level metadata is set and read via set_queue_metadata/get_queue_properties, which also reports the queue's approximate_message_count for monitoring backlog size. No separate error-handling module is documented beyond standard exception handling around delete_message - a failed delete simply lets the message become visible again after its timeout expires, which is the built-in poison-message safety net.

pip install azure-storage-queue azure-identity

Python developers building simple asynchronous task queues or decoupled worker pipelines on Azure who need low operational overhead rather than Service Bus's sessions, topics, and transactional messaging - and who follow the skill's guidance to delete messages promptly after processing and to set a time_to_live that prevents stale messages from lingering - a good fit for a background job queue that only needs at-least-once delivery, not the ordered, session-scoped guarantees Service Bus provides.