microsoft/azure-skills
29 skills10M installs38.4k starsGitHub
Install
npx skills add https://github.com/microsoft/azure-skillsSkills in this repo
1Microsoft FoundryThe microsoft-foundry skill guides fine-tuning on Azure AI Foundry with supervised fine-tuning, direct preference optimization, and reinforcement fine-tuning with graders. It spans dataset preparation, validation, training job submission, monitoring, grader calibration, deployment, evaluation, and cleanup across quickstart, full pipeline, dataset creation, iterative training, and diagnose-poor-results workflows. Bundled Python scripts cover submit_training, monitor_training, calibrate_grader, check_training curves, deploy_model via ARM REST, evaluate_model with an LLM judge, convert_dataset formats, generate distillation data, score_dataset quality, and cleanup resources. Five rules insist on baseline evaluation first, validated data before submit, RFT grader calibration targeting 25-50% base-model failure, checkpoint evaluation before deploy, and token cost measurement alongside accuracy. The skill explicitly excludes general deployment without fine-tuning, agent creation, and prompt-only optimization, routing those to sibling azure-skills modules. Error handling documents API version issues, globalStandard training types, RFT endpoint scaling, deployment race conditions, and con.487kinstalls2Azure AiThe azure-ai skill is a Microsoft agent skill for Azure AI services spanning AI Search, Speech, Azure OpenAI, and Document Intelligence. It routes developers to the right service for full-text, vector, hybrid, and semantic search, plus speech-to-text, text-to-speech, transcription, and OCR form extraction. When Azure MCP is enabled, preferred commands include azure__search for index list, get, and query, and azure__speech for transcribe and synthesize. If MCP is unavailable, the skill points to /azure:setup or /mcp enablement and documents az search and az cognitiveservices CLI paths. Capability tables cover AI enrichment, speaker diarization, neural voices with SSML, and custom speech models. Condensed SDK guides link Python, TypeScript, .NET, and Java references for search documents, vision, transcription, translation, document intelligence, and content safety. Version 1.1.1 MIT-licensed metadata anchors the skill to official Microsoft Azure AI documentation patterns.482kinstalls3Azure Diagnosticsazure-diagnostics is a Claude Code skill for debugging production issues on Microsoft Azure. It provides systematic triage across App Service, Container Apps, Azure Functions, AKS, Event Hubs, and Service Bus, using AppLens, Azure Monitor, resource-health checks, and KQL log analysis. It follows a fixed diagnosis flow (identify symptoms, check resource health, review logs, analyze metrics, investigate recent changes) and routes AKS and messaging incidents to dedicated troubleshooting guides. Developers reach for it when a live Azure workload fails and root cause must follow official Microsoft guidance instead of ad-hoc log grepping.482kinstalls4Azure Deployazure-deploy is a Claude Code skill that executes deployments of already-prepared applications to Microsoft Azure. It runs azd up, azd deploy, terraform apply, and az deployment commands with built-in error recovery, then verifies endpoints and RBAC roles. It requires a deployment plan from azure-prepare and a Validated status from azure-validate, and it explicitly does not scaffold new apps or generate infrastructure. Developers invoke it for 'run azd up', 'push to production', 'terraform apply', or 'publish to Azure' once infrastructure is prepared and validated.482kinstalls5Azure Prepareazure-prepare is a Microsoft agent skill that generates the Azure infrastructure files an application needs before deployment: Bicep or Terraform modules, azure.yaml, and Dockerfiles. It runs a mandatory plan-first workflow that writes an .azure/deployment-plan.md skeleton, gathers requirements, scans the codebase, selects a recipe (AZD, AZCLI, Bicep, or Terraform), and maps components to Azure services like App Service, Container Apps, and Functions. A developer uses it when creating, modernizing, or preparing an app to deploy to Azure. It only prepares files; azure-validate and azure-deploy handle checking and execution.482kinstalls6Azure Validateazure-validate is a Microsoft agent skill that runs pre-deployment checks on Azure configuration and infrastructure before a production push. It reads the .azure/deployment-plan.md produced by azure-prepare, then runs recipe-specific validation commands (such as azd provision --preview, bicep build, terraform validate), verifies RBAC role assignments and managed identity permissions, and records proof in the plan. A developer uses it to confirm an app is ready to deploy and to troubleshoot deployment errors. It sits between azure-prepare and azure-deploy in the workflow and blocks deploy until every check passes.482kinstalls7Azure Storageazure-storage is a Microsoft agent skill covering Azure Storage services: Blob Storage, File Shares, Queue Storage, Table Storage, and Data Lake. It explains blob access tiers (hot, cool, cold, archive), storage account tiers, and redundancy options (LRS, ZRS, GRS, GZRS), and points to per-language SDK quick references. A developer uses it when integrating object storage, SMB file shares, async queue messaging, NoSQL key-value tables, or big-data analytics into an app, and provides both Azure MCP commands and az CLI fallbacks. It excludes SQL and Cosmos DB, which are covered by other skills.482kinstalls8Entra App Registrationentra-app-registration is a Microsoft agent skill that guides registering an application in Microsoft Entra ID (formerly Azure AD), configuring OAuth 2.0 flows, requesting API permissions, and integrating MSAL. It walks through registering the app, setting redirect URIs, choosing delegated or application permissions, creating client secrets or certificates, and producing working MSAL code for console, web, SPA, and service apps. A developer uses it when adding Azure AD / Entra authentication or generating a service principal. It defers Azure RBAC role assignments and Key Vault secret audits to other skills.481kinstalls9Appinsights Instrumentationappinsights-instrumentation is a Microsoft-authored reference skill that tells an agent how to add Azure Application Insights telemetry to a web app already hosted on Azure. It covers collecting the app's language and hosting context, preferring auto-instrumentation for ASP.NET Core on App Service, and falling back to manual SDK setup for Node.js and Python. It provides a Bicep example, an Azure CLI PowerShell script, and OpenTelemetry SDK references for creating the App Insights resource and modifying code. A developer reaches for it when they need APM and observability wiring rather than a request to add components, which it defers to the azure-prepare skill.481kinstalls10Azure Resource Lookupazure-resource-lookup lists, finds and inventories Azure resources of any type across subscriptions and resource groups. A developer uses it to answer 'what resources do I have', list web apps or VMs, audit tags, or find orphaned resources like unattached disks, routing through Azure Resource Graph when no dedicated MCP tool covers the type. It is read-only and defers deployment to azure-deploy and cost work to azure-cost. It requires the resource-graph az extension and Reader role on the target subscriptions.481kinstalls11Azure Complianceazure-compliance runs Azure compliance and security audits using the azqr CLI and checks Key Vault keys, secrets and certificates for expiration. A developer invokes it to validate resource configuration against best practices, find orphaned or misconfigured resources, and see what is expiring in the next 30 days before a deployment. It classifies each finding by priority and proposes remediation while keeping reporting separate from fixes. It requires an authenticated Azure session via az login and read permissions on resource and Key Vault metadata.481kinstalls12Azure Aigatewayazure-aigateway configures Azure API Management as an AI gateway that routes and governs traffic to LLMs, MCP servers and agent tools. A developer uses it to add Azure OpenAI or AI Foundry backends, apply policies for semantic caching, token limits, token metrics and content safety, and test the gateway endpoint. It documents an az CLI workflow plus a .NET SDK quick reference using Azure.ResourceManager.ApiManagement and Azure.Identity. It requires the Azure CLI for configuration and testing.481kinstalls13Azure Kustoazure-kusto executes KQL queries against Azure Data Explorer (Kusto/ADX) for log analytics, telemetry and time-series analysis. A developer uses it to query logs in an ADX cluster, aggregate telemetry by dimension, build time-series charts, detect anomalies, or run SIEM-style security analytics without leaving agent chat. It documents a discover-schema-query-analyze workflow, five reusable KQL patterns, and performance best practices like filtering early and always bounding time ranges. It manages Kusto resources through MCP tools for listing clusters, databases and table schemas.481kinstalls14Azure Resource VisualizerThis skill inspects an Azure resource group, analyzes each resource and its configuration, and maps the network, identity, data-flow and dependency relationships between them. It then generates a detailed Mermaid architecture diagram and a markdown file documenting the architecture. A developer uses it to understand or document how deployed Azure resources fit together. It performs read-only analysis and never modifies resources.481kinstalls15Azure MessagingThis skill helps diagnose and resolve problems with the Azure Messaging SDKs for Event Hubs and Service Bus. It covers connection failures, authentication and AMQP link errors, message lock and session lock issues, checkpoint and offset resets, and SDK configuration. A developer uses it when a messaging SDK throws errors or an event processor stops handling messages. It works through resource health, diagnostic logs and documentation to recommend a remediation.470kinstalls16Azure ComputeThis skill is a router for Azure Virtual Machine and VM Scale Set requests. It directs the user to workflows for VM recommendation and pricing, connectivity troubleshooting, capacity reservation groups, and Essential Machine Management. It reads the matched workflow file before touching reference files. A developer or operator uses it to choose, size, price, autoscale, or troubleshoot Azure compute.425kinstalls17Azure Cloud MigrateThis skill assesses existing cloud workloads and migrates them to Azure, generating an assessment report before any code conversion. It supports eight source-to-target scenarios such as AWS Lambda to Azure Functions and Kubernetes/Fargate/Cloud Run to Azure Container Apps. It runs phases sequentially, flags hardcoded service-discovery hostnames, and hands off to azure-prepare for infrastructure and deployment. A developer uses it to plan and execute a cross-cloud move onto Azure.414kinstalls18Azure QuotasAzure Quotas is a Microsoft agent skill for subscription service limits and capacity validation across Azure regions. It mandates az quota CLI as the primary tool after installing the quota extension, warning that REST API, Portal, and the azure-quota MCP server can show misleading No Limit values. The skill explains adjustable vs non-adjustable quotas, maps ARM resource types to quota resource names via localizedValue discovery, and ships check-quota.sh and check-quota.ps1 scripts returning limit, usage, and available capacity in one call. Core workflows cover single-resource checks, multi-region comparison loops, quota increase requests via az quota update, and full provider listings for planning. Troubleshooting tables address ExtensionNotFound, BadRequest unsupported providers, QuotaExceeded, and InvalidScope formatting. Best practices include checking before deployment, requesting twenty percent headroom, and using table output for scanning. Use when planning VM fleets, Container Apps, networking resources, or any deployment that might hit regional provisioning limits.351kinstalls19Azure UpgradeAssess and upgrade Azure workloads between plans, tiers, or SKUs. Handles Consumption to Flex Consumption, plan changes, Container Apps migration, and Azure SDK Java modernization. Generates assessments before upgrades and follows scenarios sequentially.343kinstalls20Azure Enterprise Infra PlannerArchitects enterprise Azure infrastructure from workload descriptions. Generates Bicep or Terraform for subscription-scope deployments, landing zones, hub-spoke networks, and multi-region topologies while validating against WAF guidelines.317kinstalls21Azure KubernetesPlan, provision, and configure production-ready Azure Kubernetes Service (AKS) clusters. Distinguish Day-0 decisions (networking, API server) from Day-1 features, choose cluster SKU, and implement security, observability, and upgrade strategies.310kinstalls22Azure CostAzure Cost is a unified Microsoft agent skill for querying historical Azure spend, forecasting future bills, and finding optimization opportunities. It routes user intent to three workflows: cost query for breakdowns by service, resource, location, or tag; cost optimization for orphaned resources, rightsizing, and waste reduction; and cost forecast for end-of-month projections. It uses the Cost Management REST API via az rest with api-version 2023-11-01, prefers MCP tools like azure__documentation and azure__extension_cli_generate over raw CLI guessing, and requires Cost Management Reader plus Monitoring Reader roles. Guardrails mandate presenting actual queried data labeled ACTUAL DATA, validating pricing from official sources, including Azure Portal links, and never executing destructive deletes without explicit approval. Optimization workflows must show the total bill alongside savings recommendations, and sub-ten-dollar monthly costs should emphasize operational improvements over financial savings. Rate-limit handling requires honoring x-ms-ratelimit retry-after headers. Do not use for deploying resources, security audits, or estimating costs for not-yet-deployed resources.286kinstalls23Airunway Aks SetupAI Runway setup skill for Azure Kubernetes Service. Walks through cluster verification, controller installation, GPU hardware assessment, inference provider setup (KAITO/Dynamo/KubeRay), and first model deployment. Includes cost awareness for GPU resources and detailed error handling.230kinstalls24Entra Agent IdAzure identity management skill covering Entra ID (formerly Azure AD) app registration, authentication flows, and RBAC configuration. Part of Azure Skills module with support for managed identities and service principals.204kinstalls25Azure ReliabilityAssess and improve reliability of PaaS applications on Azure Functions and App Service. The skill scans deployed resources for zone redundancy, ZRS storage, health probes, and multi-region failover, presenting findings as a feature-pivoted checklist. It then drives staged remediation via Azure CLI or IaC patches, with user confirmation at each step.164kinstalls26Python Appservice DeployDeploy Python applications (Flask, Django, FastAPI) to Azure App Service Linux. Handles resource group, plan, and web app provisioning with framework detection. Developers use this to quickly move Python services to cloud hosting.87.9kinstalls27Skill ReviewerA skill review tool for evaluating quality and best practices in skill implementations. Use this when you need code review feedback, quality assessment, or verification that your skill meets standards before shipping.124installs28Analyze Skill IssuesAn issue analysis skill for diagnosing problems in Azure skills and deployments. Use this when troubleshooting skill failures, analyzing error logs, or debugging Azure deployments to maintain system health.122installs29AzureAzure. Covers setup, configuration, and when-to-use guidance from upstream documentation for .github/plugins/azure-skills.0installs