Add Azure services to an application
When implementing authentication, storage, messaging, or AI features, choose the SDK plugin for your language. Its skills provide library-specific patterns for client setup, service operations, and integration.
A marketplace of agent plugins for Azure development, cloud operations, and repository documentation.
Microsoft’s Agent Skills marketplace adds Azure SDK guidance, cloud workflows, and repository documentation tools to coding agents. Choose the plugins relevant to your project: language plugins cover SDK patterns, Azure plugins guide infrastructure work, and Deep Wiki turns source code into documentation with diagrams and citations.
Agents can select relevant skills from your request. Deep Wiki also provides commands you invoke explicitly for tasks such as generating a wiki or investigating a component. Bundled MCP connections and telemetry hooks depend on the plugin and client you choose; adding the marketplace alone does not install its plugins.
When implementing authentication, storage, messaging, or AI features, choose the SDK plugin for your language. Its skills provide library-specific patterns for client setup, service operations, and integration.
Assess deployment blockers, plan the Azure services an application needs, and prepare its infrastructure. The Azure workflows distinguish app onboarding from projects that already use Azure Developer CLI.
Use the AKS skills to investigate pod, node, networking, or GPU inference problems. General incident triage starts with read-only evidence; packet capture is a separate workflow for incidents that need wire-level investigation.
Use cost analysis to investigate actual spending and bill changes. Separate skills handle forecasts, resource optimization, and budgets, so a billing investigation does not become an unrelated provisioning or governance task.
Use Deep Wiki to map repository architecture, write component explanations, and produce onboarding guides with source citations. Generate a complete wiki or focus on a single topic.
Use Foundry IQ to create or query a knowledge base from supported document sources and connect it to an existing agent. SDK search skills cover application-level indexing and vector or hybrid search.
Skills are grouped by the plugin to install. Use each description to choose a task; Deep Wiki’s commands offer explicit entry points for documentation work.
The AKS, Advisor, Cost, Kusto, Azure Local, Azure, and Foundry IQ plugins include telemetry hooks. Their client-specific configurations load with the plugin. Session hooks report session metadata; tool hooks filter for matching skill, reference-file, or Azure MCP activity. Reporting depends on the installed plugin’s metadata and your telemetry setting.
To stop these scripts from collecting telemetry, set the environment variable AZURE_MCP_COLLECT_TELEMETRY to false for the process running your agent. The hook checks this before writing diagnostic logs or starting its telemetry publisher.
For optional script diagnostics, set AZURE_SKILLS_TELEMETRY_LOG_DIR to a local folder. It receives raw hook input and a telemetry log, which can contain tool arguments and paths. Leave it unset unless you need those diagnostics.
Showing all 184
Convert a VitePress wiki into Azure DevOps Wiki-compatible Markdown.
Create coding-agent instructions in repository folders where they are missing.
Analyze repository structure and plan a hierarchical documentation catalogue.
Turn recent Git commits into a categorized changelog.
Generate concise and expanded documentation files for language-model access.
Write onboarding guides for contributors, engineers, executives, and product managers.
Write technical wiki pages with source citations and Mermaid diagrams.
Answer repository questions using documentation context and source references.
Investigate a topic across files by tracing implementation and evidence.
Package generated documentation as a VitePress site.
Set up AI Runway on AKS through the first model deployment.
Choose Application Insights instrumentation patterns and SDK configuration for web applications.
Find guidance for Azure search, speech, OpenAI, and document processing.
Configure API Management policies for AI models, MCP tools, and agents.
Assess cross-cloud workloads and convert supported applications for Azure.
Audit Azure resources for compliance, security posture, and expiring credentials.
Plan and manage virtual machines, scale sets, sizing, and capacity.
Execute a prepared, validated Azure deployment using its existing deployment plan.
Investigate Azure production failures using logs, health data, and diagnostic evidence.
Design enterprise infrastructure and generate Bicep or Terraform without azd.
Plan and manage AKS clusters and Kubernetes workloads.
Generate deployment infrastructure and configuration for projects using Azure Developer CLI.
Check provider quotas, current usage, regional capacity, and increase requests.
Assess App Service and Functions resilience and guide confirmed remediation.
Find Azure resources across subscriptions, resource groups, types, and tags.
Turn resource-group relationships into Mermaid architecture diagrams.
Choose and use Azure storage services, access tiers, and lifecycle policies.
Assess hosting-plan upgrades, service migrations, and Azure SDK modernization.
Check deployment configuration, infrastructure, permissions, and prerequisites before release.
Provision Entra agent identities and configure agent-specific OAuth token exchange.
Register Entra applications and configure OAuth permissions and MSAL integration.
Route Foundry tasks toward platform guidance and API discovery tools.
Deploy Flask, Django, or FastAPI code to Linux App Service.
Plan and guide application onboarding through infrastructure preparation and approved deployment.
Assess source-code readiness and identify blockers before infrastructure work.
Write and use Kusto queries for Azure data analysis.
Diagnose Event Hubs and Service Bus SDK connection or processing failures.
Find relevant Azure skills and their plugin installation guidance.
Build persistent Foundry agents with tools, conversation threads, and streaming responses.
Detect harmful content in text and images with severity classifications.
Extract structured information from documents, images, audio, and video.
Identify conversational intent and entities with Conversational Language Understanding.
Manage machine learning workspaces, training jobs, models, compute, and pipelines.
Build Foundry applications with project clients, versioned agents, and evaluations.
Analyze sentiment, entities, key phrases, language, and sensitive text.
Transcribe speech with timestamps and speaker diarization.
Translate documents in batches while preserving their format.
Translate text, detect languages, transliterate, and look up dictionary entries.
Analyze images for captions, objects, text, people, and cropping suggestions.
Build streaming voice applications with conversation, tools, and avatar integration.
Centralize application settings and feature flags.
Manage container registry repositories, images, and artifacts.
Build Cosmos DB service layers for FastAPI with authentication and partitioning.
Read, write, and query documents with the Cosmos DB NoSQL SDK.
Manage table entities and batch operations in Storage or Cosmos DB.
Publish Event Grid events and handle CloudEvents.
Build event producers and consumers with checkpointing.
Configure Azure authentication, credential chains, managed identity, and token caching.
Manage secrets, cryptographic keys, and certificates in Key Vault.
Manage WebSocket connections and publish real-time messages.
Manage API inventory, metadata, and governance.
Manage API Management services, APIs, subscriptions, products, and policies.
Manage Microsoft Fabric capacities and Azure resources.
Send custom logs to Log Analytics through the Logs Ingestion API.
Configure low-level OpenTelemetry exporters for Application Insights.
Add Application Insights monitoring with the Azure Monitor OpenTelemetry distribution.
Query Log Analytics logs and Azure Monitor metrics.
Build vector or hybrid search and manage indexes, indexers, and skillsets.
Send and receive messages through queues, topics, and subscriptions.
Transcribe audio up to 60 seconds through the Speech REST API.
Upload, download, list, and manage blobs, containers, and lifecycle settings.
Manage hierarchical Data Lake Storage Gen2 files and directories.
Work with cloud SMB file shares, directories, and files.
Use Storage queues for asynchronous processing and task distribution.
Create FastAPI endpoints with authentication dependencies and typed response models.
Build Microsoft 365 agents with aiohttp hosting, routing, and streaming.
Define separate Pydantic models for creation, updates, responses, and database records.
Create and manage Azure Bot Service resources, channels, and connections.
Manage persistent agents, threads, messages, runs, and tools.
Extract text, tables, and structured fields with document analysis models.
Implement chat, embeddings, image generation, and audio features with Azure OpenAI.
Work with Foundry projects, agents, connections, datasets, and evaluations.
Build voice assistants with bidirectional audio over WebSocket.
Publish and consume Event Grid events, including CloudEvents.
Send and process event streams with partitioning and checkpointing.
Authenticate Azure SDK clients with developer, application, or managed credentials.
Add geocoding, directions, map rendering, geolocation, and weather data.
Manage an API inventory with versions, definitions, environments, and metadata.
Manage API Management services, APIs, products, policies, and gateways.
Manage Application Insights resources, availability tests, and workbooks.
Manage Azure-side Arize AI organization resources through Azure Marketplace.
Manage Azure Bot resources, channels, and connection settings.
Provision, scale, pause, and resume Microsoft Fabric capacities.
Manage Azure-side MongoDB Atlas organizations, rather than Atlas databases or clusters.
Provision Weights & Biases instances and configure Azure Marketplace integration.
Provision Cosmos DB accounts, databases, containers, and throughput; excludes document operations.
Manage Durable Task schedulers, task hubs, and retention policies.
Manage MySQL Flexible Servers, databases, firewall rules, and backups.
Provision Playwright Testing workspaces; running browser tests uses a separate library.
Manage PostgreSQL Flexible Servers, databases, firewall rules, and backups.
Manage Redis resources and networking; excludes reading or writing cached values.
Manage Azure SQL servers, databases, pools, and failover; excludes SQL queries.
Build full-text, vector, semantic, and hybrid search with indexing workflows.
Create, rotate, and use cryptographic keys in Key Vault or Managed HSM.
Build queue and publish-subscribe messaging with sessions and dead-letter handling.
Build multichannel Microsoft 365 agents with ASP.NET Core hosting.
Customize Entra authentication events with Azure Functions triggers.
Capture browser activity, errors, dependencies, and agent traces with Application Insights.
Moderate text and images and manage custom content blocklists.
Process invoices, receipts, forms, and IDs with document analysis models.
Use Foundry project clients for agents, connections, deployments, and evaluations.
Translate text or documents and add language detection or transliteration.
Build browser or Node.js voice applications with bidirectional audio.
Manage shared configuration, feature flags, and dynamic settings refresh.
Implement Cosmos DB document operations, queries, and bulk processing.
Ingest event streams and process them with partitioned consumers.
Authenticate Azure clients with managed identity, service principals, or interactive login.
Create, rotate, and use Key Vault cryptographic keys.
Store and retrieve application secrets in Key Vault.
Run Playwright tests across cloud-hosted browsers and publish test results.
Instrument Node.js applications with distributed traces, metrics, and logs.
Query PostgreSQL with connection pooling, transactions, and optional Entra authentication.
Build search indexes, vector or hybrid queries, and agentic retrieval.
Implement queues, subscriptions, sessions, and dead-letter handling.
Manage blobs and containers with streaming and shared-access support.
Create SMB shares and manage directories, files, and metadata.
Send, receive, inspect, and delete Storage queue messages.
Build WebSocket messaging, group chat, and live notifications.
Build dark React interfaces with Tailwind styling and motion.
Build Microsoft 365 agents with Express hosting and streaming responses.
Build typed React Flow nodes with handles and Zustand integration.
Create typed Zustand stores with separate state and actions.
Manage persistent agents and their conversation threads, runs, and tools.
Detect anomalies in single-variable or multivariable time-series data.
Analyze text and images for harmful content and manage blocklists.
Extract document text, tables, and fields with Document Intelligence.
Manage Foundry project connections, datasets, indexes, and evaluations.
Extract image captions, text, tags, and detected objects.
Stream bidirectional voice conversations with AI assistants over WebSocket.
Manage shared settings, feature flags, and configuration snapshots.
Build automated call flows with routing, recording, and speech interaction.
Maintain legacy CallingServer integrations; use Call Automation for new projects.
Build chat threads with participants, messages, receipts, and typing notifications.
Handle Communication Services credentials, user identifiers, and token refresh.
Send SMS notifications and track message delivery.
Run parallel or HPC workloads with Batch pools, jobs, and tasks.
Implement Cosmos DB document operations, queries, and reactive access.
Store and query structured entities in Azure Tables or Cosmos DB.
Publish events and integrate services through Event Grid.
Build high-throughput event ingestion and streaming applications.
Configure Azure SDK authentication with Entra ID credentials.
Add WebSocket messaging, live updates, and server-to-client notifications.
Send custom logs through Data Collection Rules and Endpoints.
Export OpenTelemetry traces, metrics, and logs to Azure Monitor.
Query Log Analytics workspaces and Azure resource metrics.
Manage RSA or EC keys and perform encryption or signing operations.
Store, retrieve, and manage passwords, API keys, and connection strings.
Upload, download, stream, and manage blobs and containers.
Work with Cosmos DB documents, containers, and partitioned data.
Send and receive Event Hubs streams for ingestion and processing.
Authenticate Azure SDK clients with Microsoft Entra ID.
Create, import, and manage X.509 certificates in Key Vault.
Manage cryptographic keys and perform encryption, signing, or key wrapping.
Store and retrieve passwords, API keys, and other secrets.
Upload, download, and manage blobs and containers.
Send, receive, and manage Azure Storage queue messages.
Investigate advanced AKS efficiency, bin packing, scaling bottlenecks, and storage costs.
Investigate existing AKS GPU and KAITO incidents using read-only evidence.
Find documented AKS fixes for exact error signatures with their required qualifiers.
Collect bounded AKS packet captures after read-only network diagnosis.
Investigate live AKS failures through read-only cluster evidence.
Find supported Azure portal blades and generate resource-specific navigation links.
Review Advisor recommendations and service retirements, or update recommendation status.
Plan and carry out remediation for Azure Advisor recommendations.
Investigate actual Azure spending, bill changes, and AKS or AI costs.
Forecast spending and estimate prices for planned resources or workloads.
Manage budgets, alerts, cost-allocation tags, and policy restrictions.
Find savings through rightsizing, idle-resource analysis, reservations, and Savings Plans.
Build and query Kusto graphs for relationships, paths, and connected components.
Compose incident-response hunting pipelines with IRQL functions.
Map query results into IRQL graph functions for Kusto Explorer visualization.
Plan, deploy, operate, and troubleshoot standard Azure Local environments.
Manage rack-scale Azure Local infrastructure and its network control plane.
Create or query Foundry IQ knowledge bases and connect existing agents.
No matching results.
/deep-wiki:adoGenerate a script that converts the wiki for Azure DevOps Wiki.
In Copilot CLI, invoke Deep Wiki’s ado command from the repository containing the generated VitePress wiki. It produces a conversion script for Azure DevOps Wiki output.
/deep-wiki:agentsCreate AGENTS.md files in pertinent folders where they are missing.
Invoke Deep Wiki’s agents command in the repository you want to document. It analyzes relevant folders and creates instructions only where an AGENTS.md file is missing.
/deep-wiki:askAnswer a repository question using wiki context and source references.
Invoke Deep Wiki’s ask command in Copilot CLI and provide your question. Keep the relevant repository open so it can inspect source files and existing wiki context.
/deep-wiki:buildPackage generated wiki pages into a themed VitePress site.
Invoke Deep Wiki’s build command after generating wiki pages. It prepares the VitePress project, theme, navigation, and site build.
/deep-wiki:catalogueGenerate the repository wiki’s hierarchical table of contents as JSON.
Invoke Deep Wiki’s catalogue command in the target repository to request its documentation structure without generating every page.
/deep-wiki:changelogCreate a categorized changelog from recent Git commits.
Invoke Deep Wiki’s changelog command in a repository with Git history. It groups recent commits by change type.
/deep-wiki:crispGenerate a concise wiki without running its site build.
Invoke Deep Wiki’s crisp command in the target repository when you want a concise documentation site. This flow generates the site files without running a build.
/deep-wiki:deployGenerate GitHub Actions configuration for publishing the wiki to GitHub Pages.
Invoke Deep Wiki’s deploy command for an existing VitePress wiki in a GitHub repository. It prepares the publishing workflow and repository-specific site configuration.
/deep-wiki:generateGenerate a full wiki, onboarding guides, and VitePress site.
Invoke Deep Wiki’s generate command in Copilot CLI with the target repository open. The workflow covers the catalogue, documentation pages, onboarding guides, and VitePress site.
/deep-wiki:llmsGenerate llms.txt and an expanded documentation file for language-model access.
Invoke Deep Wiki’s llms command with the repository and generated wiki available. It creates a linked documentation summary and a fuller text version.
/deep-wiki:onboardGenerate onboarding guides for four audiences.
Invoke Deep Wiki’s onboard command in the target repository. It produces guides for contributors, staff engineers, executives, and product managers.
/deep-wiki:pageWrite one technical wiki page with diagrams and source citations.
Invoke Deep Wiki’s page command and supply the topic or title. Include specific file paths when you want to focus the investigation.
/deep-wiki:researchInvestigate a repository topic by tracing code across files.
Invoke Deep Wiki’s research command and describe the topic to investigate. It follows code paths across the open repository and records supporting evidence.
Analyze repository architecture and plan documentation catalogues and onboarding.
Install Deep Wiki in Copilot CLI to make this agent definition available in its agents list. Begin documentation planning through Deep Wiki’s catalogue or generate command, with the repository open and its source location identified.
Trace implementations across source files and assemble evidence-based findings.
This definition is bundled with Deep Wiki for Copilot CLI. Start through its research command and provide the implementation question and repository context.
Write technical documentation with source citations and Mermaid diagrams.
This definition is bundled with Deep Wiki for Copilot CLI. Start through its page or generate command and provide the topic and relevant repository context.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
After a Cursor MCP call, reports matching Azure server activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, reports matching Azure server activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill or reference-file activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill or reference-file activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill or reference-file activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, checks for matching Azure server telemetry.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, reports matching Azure server activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, reports matching Azure server activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, reports matching Azure server activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Claude Code session start or resume, attempts to report session metadata.
After Claude Code tool use, reports matching skill, reference, or MCP activity.
On a Copilot session start, attempts to report session metadata.
After Copilot tool use, reports matching skill, reference, or MCP activity.
On Cursor session start or resume, attempts to report session metadata.
After Cursor tool use, reports matching skill and reference-file activity.
After a Cursor MCP call, reports matching Azure server activity.
Connect the Cost plugin to Azure Resource Manager’s Cost Management tools.
Give the agent tools for Azure resources, deployments, and service operations.
Give the agent tools for Azure resources, deployments, and service operations.
Give the agent tools for Azure resources, deployments, and service operations.
Give the agent tools for Azure resources, deployments, and service operations.
Connect the Foundry plugin to Foundry platform operations.
Search Microsoft Learn for product documentation and implementation guidance.
Add the marketplace, then install the plugins you need. Each selected plugin installs as a package; you do not need to install its commands, agents, or hooks separately.
Choose complete plugins from the skills marketplace. For plugins that start Azure MCP, have Node.js 22 or later available and use a trusted device with trusted npm configuration. Azure operations also need an Azure identity with access to the intended resources.
The Azure operations plugins include telemetry hooks. To opt out before they run, set AZURE_MCP_COLLECT_TELEMETRY to false in Claude Code’s process environment.
In your Claude Code session, add Microsoft’s marketplace.
/plugin marketplace add microsoft/skillsOpen the plugin panel. In Discover, choose a plugin from skills, review it, select its installation scope, and install it.
/pluginChoose Deep Wiki for repository documentation, azure for Azure workflows, or the azure-sdk plugin matching your language. The marketplace also offers AKS, Advisor, Cost, Kusto, Azure Local, and Foundry IQ plugins.
If the panel reports Active now, continue to the check below. If it reports Reload needed, it closes and runs a reload automatically. If Claude Code warns that the plugin remains pending, restart Claude Code or run the following command in your session to activate it; forcing the reload costs one uncached request.
/reload-plugins --forceConfirm the plugin is enabled in Installed and that its skills appear in the slash menu. For a plugin with MCP servers, also inspect the MCP connection view and complete any requested authentication. Confirm the intended Azure account before using resource tools.
For Azure MCP authentication, use the Azure MCP setup guidance. Plugin loading and installation scopes are covered in Claude Code’s install guide.
Use VS Code’s Preview marketplace controls to add Microsoft’s marketplace and install the plugins you need.
For plugins that start Azure MCP, have Node.js 22 or later and trusted npm configuration. To opt out of Azure plugin hook telemetry, set AZURE_MCP_COLLECT_TELEMETRY to false in the agent process environment before starting the session.
Run Chat: Open Customizations from the Command Palette. In the Agent Customizations editor, select Browse, open the source menu, and choose Configure Marketplaces.
Select Add Marketplace and enter microsoft/skills. Only add a marketplace you trust. Adding it makes its plugins browsable; it does not install or enable them.
Return to Browse and select the added marketplace from the source menu. Choose the plugin for your task and select Install. On the first installation from this marketplace, review the source before confirming the trust prompt.
Choose Deep Wiki for repository documentation, Azure for Azure workflows, or the Azure SDK plugin matching your language. Specialist plugins cover AKS, Advisor, Cost, Kusto, Azure Local, and Foundry IQ; Microsoft Foundry is also listed.
In the Agent Customizations editor, open Plugins and confirm that the plugin is listed and enabled for the intended harness and workspace.
Start a fresh session with that harness and workspace. Check that the expected skills appear in the slash menu. For bundled MCP servers, inspect their connection status and complete any requested authentication before using them.
See VS Code agent plugins for marketplace controls, enablement, and component availability.
Run the following commands in your terminal. Each choice installs one complete plugin; repeat the plugin selection for any additional packages you need.
Plugins that start Azure MCP need Node.js 22 or later and trusted npm configuration. Azure operations need an authenticated Azure identity with access to the intended resources. To opt out of the Azure plugins’ hook telemetry, set AZURE_MCP_COLLECT_TELEMETRY to false in the agent’s process environment before starting it.
Add the skills marketplace.
copilot plugin marketplace add microsoft/skillsChoose and install the plugin for your task.
Plugin
For repository documentation, source-based investigation, and wiki publishing.
copilot plugin install deep-wiki@skillsFor application onboarding, deployments, diagnostics, and infrastructure workflows.
copilot plugin install azure@skillscopilot plugin install azure-sdk-python@skillscopilot plugin install azure-sdk-dotnet@skillscopilot plugin install azure-sdk-typescript@skillscopilot plugin install azure-sdk-java@skillscopilot plugin install azure-sdk-rust@skillscopilot plugin install azure-kusto-graph-skills@skillscopilot plugin install azure-local-skills@skillscopilot plugin install foundry-iq-skills@skillscopilot plugin install azure-cost@skillscopilot plugin install aks-skills@skillscopilot plugin install azure-advisor@skillsConfirm the plugin appears in the installed list.
copilot plugin listStart a Copilot CLI session and list the available skills. For bundled MCP connections, inspect the MCP dashboard and complete any requested authentication. Confirm the intended Azure account before using resource tools.
/skills listFor Azure MCP authentication, use the Azure MCP setup guidance. The Copilot plugin reference covers plugin management.
Choose this plugin for its Foundry routing skill and bundled Azure, Foundry, and Microsoft Docs connections. You need Node.js 22 or later for Azure MCP. Foundry work also needs an Azure account and access to a Foundry project.
In your Copilot CLI session, add the marketplace and install Microsoft Foundry.
/plugin marketplace add microsoft/skills/plugin install microsoft-foundry@skillsIf you will use the Azure connection, sign in to the intended Azure account from your terminal.
az login # for az + Azure MCP + DefaultAzureCredential locallyList the available skills and confirm microsoft-foundry appears. Open the MCP dashboard to inspect the Azure, Foundry, and Microsoft Docs connections and complete any requested authorization.
/skills listThe plugin’s Foundry guide covers project endpoints and additional setup for hosted-agent work.
Use Codex CLI for this route. Plugins are not available in the IDE extension. The marketplace’s legacy manifest supplies the plugin catalogue.
For a plugin that starts Azure MCP, have Node.js 22 or later and trusted npm configuration. Azure resource operations need an identity with access to the intended account and resources.
Add Microsoft’s marketplace from your terminal.
codex plugin marketplace add microsoft/skillsOpen Codex CLI’s plugin browser, choose a plugin from skills, and install it.
/pluginsChoose Deep Wiki, Azure, a language-specific Azure SDK plugin, or a specialist plugin for AKS, Advisor, Cost, Kusto, Azure Local, or Foundry IQ.
Start a new Codex session. Confirm installation in the plugin browser and inspect the skills list. For a plugin with MCP servers, inspect its active connections and complete any requested authorization.
/plugins/mcpSee Codex plugins for the browser and session-loading behavior.