azd-copilot

name

azd-copilot

description

AI-powered Azure development assistant built on GitHub Copilot CLI. Provides interactive sessions, automated app building, code review, optimization, diagnostics, and 16 specialized Azure agents. USE FOR: build Azure apps from descriptions, AI code review, cost optimization, deployment troubleshooting, security review, interactive Azure development, copilot sessions, checkpoint resume. DO NOT USE FOR: non-Azure projects, manual CLI operations (use azd directly), simple git operations.

azd-copilot Extension

AI-powered Azure development assistant that extends the Azure Developer CLI (azd) with GitHub Copilot CLI integration, 16 specialized Azure agents, and 29+ focused skills.

When to Use

• Building Azure applications from natural language descriptions
• AI-assisted code review, security audits, and optimization
• Diagnosing deployment failures and production issues
• Interactive Azure development sessions with checkpoint/resume
• Generating infrastructure-as-code (Bicep/Terraform) with best practices

Commands

Interactive Session (default)

azd copilot                          # Start interactive session
azd copilot -p "deploy a Python API" # Start with a prompt
azd copilot --resume                 # Resume last session
azd copilot --agent azure-architect  # Use a specific agent
azd copilot --yolo                   # Auto-approve all actions
azd copilot -m claude-sonnet-4       # Use a specific model

Build (2-Phase App Creation)

azd copilot build -p "React + FastAPI on Azure Container Apps"
azd copilot build --resume           # Resume from checkpoint

The build command runs two phases:

1. Spec phase: Generates a detailed application specification for review/approval
2. Build phase: Implements the spec with infrastructure, application code, and deployment config

Quick Actions

Command	Description
azd copilot init	Initialize a new Azure project with AI guidance
azd copilot review	AI code review of current changes
azd copilot fix	Diagnose and fix issues in the codebase
azd copilot optimize	Suggest performance and cost optimizations
azd copilot diagnose	Troubleshoot deployment or runtime issues

Management Commands

Command	Description
azd copilot agents	List available agents
azd copilot skills	List available skills
azd copilot sessions	Manage copilot sessions
azd copilot checkpoints	List and manage build checkpoints
azd copilot spec	View or edit the current build spec
azd copilot context	Show project context information
azd copilot mcp	Manage MCP server configuration
azd copilot version	Show version information

16 Specialized Agents

Agent	Specialization
azure-manager	Orchestrates design, build, and deployment (default)
azure-architect	Infrastructure design and architecture decisions
azure-dev	Application code — backend, frontend, data layer
azure-devops	CI/CD, deployment, reliability, observability
azure-data	Database selection, schema design, query optimization
azure-ai	AI services, agent frameworks, RAG, model deployment
azure-security	Code security, infrastructure security, identity & auth
azure-finance	Cost estimation, optimization, waste identification
azure-quality	Testing, code review, refactoring, package evaluation
azure-docs	README, API docs, ADRs, runbooks
azure-design	WCAG compliance, accessibility audits, UI review
azure-product	Requirements, user stories, acceptance criteria
azure-marketing	Positioning, landing pages, competitive analysis
azure-analytics	Usage analytics, dashboards, metrics design
azure-compliance	Framework assessment (GDPR, SOC2, HIPAA), gap analysis
azure-support	Troubleshooting, FAQ generation, error messages

Build Flow

1. Run azd copilot build -p "description"
2. Spec is generated and presented for review
3. Approve or edit the spec
4. Build proceeds through phases with automatic checkpoints
5. Resume anytime with azd copilot build --resume

Checkpoints save progress at key milestones so builds can survive interruptions.

MCP Servers

azd-copilot configures these MCP servers for Copilot CLI:

• azure — Azure resource management and queries
• azd — Azure Developer CLI operations
• azd-app — Application-level project context
• microsoft-learn — Official Microsoft documentation search
• context7 — Library documentation lookup
• playwright — Browser automation for testing

Key Flags

Flag	Short	Description
--prompt	-p	Run with a specific prompt
--resume	-r	Resume the last session
--yolo	-y	Auto-approve all actions
--agent	-a	Use a specific agent
--model	-m	Use a specific AI model
--add-dir		Additional directories to include
--verbose	-v	Verbose output
--debug		Enable debug logging
--cwd	-C	Set working directory

Common Workflows

New Azure App from Scratch

azd copilot build -p "Node.js API with Cosmos DB on Container Apps"

Review and Harden Existing App

azd copilot review
azd copilot --agent azure-security -p "audit this app for vulnerabilities"
azd copilot optimize

Troubleshoot a Failed Deployment

azd copilot diagnose
azd copilot --agent azure-devops -p "fix the CI/CD pipeline"

Cost Optimization

azd copilot --agent azure-finance -p "analyze costs and suggest savings"