Azure Storage Skill
Gives coding agents practical guidance and operations for Azure objects, files, queues, tables, and data lakes.
The skill documents managed identities, RBAC, secret avoidance, and production authentication guidance, with understandable permission concepts. However, upload, download, and listing actions lack explicit user confirmation, and sensitive-data handling, external side effects, auditing, and rollback are not specified, so points were deducted.
The main instructions, CLI fallback, and SDK examples are broadly consistent, and the authentication reference is useful. However, there are no skill-specific tests for key paths, parameter validation, abnormal-input handling, or diagnostic failure guidance; the static-review cap also prevents a score above 10.
Service coverage, scenarios, and exclusions for SQL, Cosmos DB, Event Hubs, and Service Bus are relatively clear, with multi-language support. Trigger priority, input/output contracts, non-Azure boundaries, and mainland-China network reachability are not established, so points were deducted.
The skill has front matter, an MIT license, a version, service tables, MCP/CLI guidance, reference documentation, and authentication practices with reasonable information architecture. It lacks a changelog, explicit maintenance owner and update path, compatibility matrix, FAQ, and systematic troubleshooting, so points were deducted.
It can directly support Azure Storage concepts, CLI usage, authentication, and introductory multi-language SDK work, so the core task is plausible. The examples are mostly minimal client initialization or upload snippets and omit complete download flows, recovery, lifecycle configuration, and verified outputs; additional manual work remains, and the static cap prevents a score above 7.
Microsoft Learn references and auditable code are provided, alongside repository CI/test context. However, there is no skill-specific acceptance suite, test result, or third-party execution evidence, so conclusions rely primarily on static inspection and the static cap prevents a score above 5.
- Upload and download commands create external data read/write side effects, but the skill does not require confirmation, target-scope confirmation, overwrite warnings, or recovery planning.
- The examples use placeholder accounts and credential chains; actual SDK versions, CLI parameters, permissions, and abnormal behavior are not statically verified.
- Coverage spans Blob, File Share, Queue, Table, and Data Lake services but is uneven in depth; complex production use still requires official documentation and manual security review.
- No guidance is provided for mainland-China reachability, proxies, private endpoints, or offline operation.
What it does & when to use it
The Azure Storage skill covers Blob Storage, File Shares, Queue Storage, Table Storage, and Data Lake. It explains storage-account and blob access tiers, lifecycle management, redundancy choices, and common upload, download, and listing operations. Azure MCP is the preferred integration, with Azure CLI as a fallback. SQL databases, Cosmos DB, Event Hubs, and Service Bus messaging are explicitly outside its scope.
It guides or performs listing of storage accounts, containers, and blobs, plus blob uploads and downloads. It explains when to use file shares, queues, tables, and Data Lake, and compares Standard versus Premium account tiers, Hot, Cool, Cold, and Archive blob tiers, and LRS, ZRS, GRS, and GZRS redundancy. With Azure MCP enabled, it uses azure__storage commands including storage_account_list, storage_container_list, storage_blob_list, storage_blob_get, and storage_blob_put. Without MCP, it provides az storage account, container, blob, file, queue, table, and fs CLI examples.
- An Azure developer needs to upload, download, or list blobs and manage containers.
- An engineer needs to choose blob access tiers for backups, static content, or object data.
- A team is migrating a file server and needs SMB file shares.
- An application developer needs asynchronous task queues or simple NoSQL key-value storage.
- An Azure user needs hierarchical-namespace storage for big-data analytics.
- A storage administrator wants lifecycle management guidance for long-term data.
Pros & cons
- Covers five major Azure Storage services: blobs, file shares, queues, tables, and Data Lake.
- Offers both a preferred Azure MCP path and an Azure CLI fallback.
- Includes practical guidance on access tiers, lifecycle management, and redundancy.
- Links to condensed SDK references across .NET, Java, JavaScript, Python, Go, and Rust.
- Requires an Azure environment; the MCP path requires Azure MCP and the fallback requires Azure CLI.
- The SKILL.md does not document specific authentication, RBAC, or cost-estimation procedures.
- It explicitly excludes SQL databases, Cosmos DB, Event Hubs, and Service Bus messaging.
- The repository README describes the collection as work in progress, so collection-wide counts and coverage can change.
How to install
Install the collection with: npx skills add microsoft/skills. Use the wizard to select the skills you need; installation goes to the chosen agent directory, such as .github/skills/ for GitHub Copilot. The skill is located at .github/plugins/azure-skills/skills/azure-storage/SKILL.md. The README does not provide a separate command for installing only azure-storage.
How to use
Give the coding agent a concrete request such as “list the containers in my storage account and download a specified blob” or “compare Hot, Cool, Cold, and Archive tiers for a backup workload.” When Azure MCP is enabled, it can use azure__storage. If MCP is unavailable, run /azure:setup or enable it through /mcp, then use the CLI fallback, for example: az storage blob list --account-name ACCOUNT --container-name CONTAINER --output table.
Compared to similar skills
For NoSQL key-value storage, the skill covers Table Storage but says to consider Cosmos DB. Event Hubs and Service Bus messaging should be handled by a dedicated Azure messaging skill rather than this one.