Agents Overview

Agents Overview

Version 3.7.8 | Complete reference for all agents in the Agentic InfraOps workflow

This document provides a comprehensive overview of all agents in Agentic InfraOps, covering the complete 8-phase infrastructure development and operations workflow. Each agent is designed for a specific purpose in the lifecycle from requirements gathering through post-deployment diagnostics.

Contents

• Workflow Visualization
• Agent Comparison Table
• How to Use Agents
• Phase-by-Phase Workflow
• Agent Details
• Quick Reference Guide
• Related Documentation

Workflow Visualization

The Agentic InfraOps workflow spans 8 phases, from initial requirements through post-deployment health diagnostics. Agents work together in a coordinated workflow, with each specializing in a specific phase or cross-cutting concern.

%%{init: {'theme':'neutral'}}%%
graph TB
    subgraph "Phase 1: Requirements"
        P["Project Planner<br/>(custom agent)<br/>📋"]
    end

    subgraph "Phase 2: Architecture & Cost"
        A["architect<br/>🏛️"]
        MCP["💰 Azure Pricing MCP"]
    end

    subgraph "Phase 3: Design Artifacts"
        D["diagram<br/>📊 (-des)"]
        ADR1["adr<br/>📝 (-des)"]
        COST["Cost Estimate<br/>💰 (-des)"]
    end

    subgraph "Phase 4: Policy & Planning"
        POL["Policy Discovery<br/>🛡️ (bicep-plan)"]
        B["bicep-plan<br/>📐"]
    end

    subgraph "Phase 5: Code Generation"
        I["bicep-code<br/>⚙️"]
    end

    subgraph "Phase 6: Deployment"
        DEP["deploy<br/>🚀"]
    end

    subgraph "Phase 7: As-Built Documentation"
        D2["diagram<br/>📊 (-ab)"]
        ADR2["adr<br/>📝 (-ab)"]
        COST2["Cost Estimate<br/>💰 (-ab)"]
        WL["workload-documentation<br/>📚"]
    end

    subgraph "Phase 8: Health & Diagnostics"
        DIAG["azure-resource-health<br/>🏥"]
    end

    P -->|"01-requirements.md"| A
    MCP -.->|"real-time pricing"| A
    A -->|"02-assessment.md"| D
    A --> ADR1
    A --> COST
    D --> POL
    ADR1 --> POL
    COST --> POL
    POL -->|"04-governance-<br/>constraints.md"| B
    B -->|"04-implementation-<br/>plan.md"| I
    I -->|"infra/bicep/<br/>{project}/"| DEP
    DEP -->|"06-deployment-<br/>summary.md"| D2
    DEP --> ADR2
    DEP --> COST2
    DEP --> WL
    DEP -.->|"optional"| DIAG
    DIAG -->|"08-health-<br/>report.md"| WL

    style P fill:#e1f5fe
    style A fill:#fff3e0
    style MCP fill:#fff9c4
    style D fill:#f3e5f5
    style ADR1 fill:#e8eaf6
    style COST fill:#c8e6c9
    style POL fill:#ffe0b2
    style B fill:#e8f5e9
    style I fill:#fce4ec
    style DEP fill:#c8e6c9
    style D2 fill:#f3e5f5
    style ADR2 fill:#e8eaf6
    style COST2 fill:#c8e6c9
    style WL fill:#e3f2fd
    style DIAG fill:#ffccbc

Workflow Flow Legend

Agent Comparison Table

A comprehensive overview of all agents, their workflow phases, and outputs:

Key Capabilities by Function

How to Use Agents

Invocation Steps

1. Press Ctrl+Alt+I to open Copilot Chat
2. Select agent from the agent picker dropdown at the top
3. Type your prompt and submit
4. Wait for approval prompt before proceeding to next step

Agent Selection Guide

Phase-by-Phase Workflow

This section provides a detailed walkthrough of each phase in the 8-phase workflow.

Phase 1: Requirements Definition

Agent: plan (custom)

Objectives:

• Capture comprehensive project requirements
• Define non-functional requirements (NFRs): SLA, RTO, RPO, performance
• Document compliance and security requirements
• Establish cost constraints and budget

Key Outputs:

• 01-requirements.md - Structured requirements document

Next Phase: Architecture assessment with architect

Phase 2: Well-Architected Framework Assessment & Cost Estimation

Agent: architect

Objectives:

• Assess architecture against Azure WAF five pillars
• Recommend Azure services and patterns
• Provide real-time cost estimates via Azure Pricing MCP
• Validate NFRs are captured before proceeding

Key Outputs:

• 02-architecture-assessment.md - WAF pillar scores and recommendations

MCP Integration: Azure Pricing MCP provides real-time SKU pricing and cost comparisons

Next Phase: Design artifacts (diagrams, ADRs, cost estimates)

Phase 3: Design Artifacts & Documentation

Agents: diagram, adr

Objectives:

• Visualize proposed architecture with diagrams
• Document key architectural decisions
• Generate detailed cost estimates

Key Outputs:

• 03-des-diagram.py + .png - Architecture visualization
• 03-des-adr-NNN-{title}.md - Architecture Decision Records
• 03-des-cost-estimate.md - Detailed cost breakdown

Suffix Convention: -des indicates design-phase artifacts

Next Phase: Policy discovery and implementation planning

Phase 4: Azure Policy Discovery & Implementation Planning

Agent: bicep-plan

Objectives:

• Query Azure Resource Graph for policy assignments
• Discover governance constraints and compliance requirements
• Plan module structure and resource dependencies
• Validate cost estimates against implementation plan

Key Outputs:

• 04-implementation-plan.md - Detailed implementation roadmap
• 04-governance-constraints.md - Policy compliance documentation
• 04-governance-constraints.json - Machine-readable policy data

MCP Integration: Azure Pricing MCP for cost validation

Next Phase: Bicep code generation

Phase 5: Bicep Infrastructure-as-Code Generation

Agent: bicep-code

Objectives:

• Generate production-ready Bicep templates
• Create modular, reusable infrastructure code
• Follow AVM-first approach (Azure Verified Modules)
• Validate code with bicep build and lint

Key Outputs:

• infra/bicep/{project}/main.bicep - Main template
• infra/bicep/{project}/modules/*.bicep - Modular resources
• infra/bicep/{project}/deploy.ps1 - Deployment script
• 05-implementation-reference.md - Code documentation

Standards Applied:

• Unique suffix pattern for resource naming
• Security defaults (TLS 1.2, HTTPS-only, Azure AD auth)
• Diagnostic settings for all resources

Next Phase: Deployment to Azure

Phase 6: Azure Deployment

Agent: deploy

Objectives:

• Execute infrastructure deployment to Azure
• Perform what-if analysis before deployment
• Manage deployment lifecycle
• Generate deployment summary with resource IDs

Workflow Steps:

1. Pre-deployment validation (Azure CLI auth, Bicep build)
2. Run what-if analysis for impact assessment
3. Execute deployment after user approval
4. Generate deployment summary

Key Outputs:

• 06-deployment-summary.md - Deployment record with endpoints

Next Phase: As-built documentation and optional health diagnostics

Phase 7: As-Built Documentation

Agents: diagram, adr, docs

Objectives:

• Document deployed infrastructure state
• Create customer-deliverable documentation package
• Capture implementation decisions
• Generate operational runbooks

Key Outputs:

• 07-ab-diagram.py + .png - As-built architecture diagram
• 07-ab-adr-NNN-{title}.md - Implementation ADRs
• 07-ab-cost-estimate.md - Actual deployment costs
• 07-documentation-index.md - Master documentation index
• 07-design-document.md - Complete design document
• 07-operations-runbook.md - Day-2 operations procedures
• 07-resource-inventory.md - Resource listing
• 07-compliance-matrix.md - Security control mapping (optional)
• 07-backup-dr-plan.md - DR procedures (optional)

Suffix Convention: -ab indicates as-built artifacts

Next Phase: Optional health diagnostics

Phase 8: Resource Health Diagnostics (Optional)

Agent: diagnose

Objectives:

• Validate deployed resource health
• Identify and diagnose issues
• Recommend remediation actions
• Generate comprehensive health report

When to Use:

• Post-deployment validation before go-live
• Troubleshooting production issues
• Periodic health assessments
• Before major changes or scaling events

Key Outputs:

• 08-resource-health-report.md - Diagnostic findings and remediation actions

Workflow Phases:

1. Resource discovery and validation
2. Health status assessment
3. Log and telemetry analysis
4. Issue classification and root cause
5. Remediation planning and execution
6. Report generation

Interactive Features: Approval-first execution, clarifying questions at each phase

Agent Details

This section provides detailed information about each agent, including purpose, behaviors, example prompts, and outputs.

requirements

Phase 1 | Custom requirements gathering agent

Purpose: Gather comprehensive requirements including business context, NFRs (SLA/RTO/RPO), compliance needs, and cost constraints. This is a custom agent specifically designed for Azure infrastructure projects.

Note: VS Code includes a built-in "Plan" agent for general planning. This repository uses the custom Requirements agent (requirements.agent.md) with Azure-specific instructions and workflow handoffs.

📋 Template: Use .github/prompts/plan-requirements.prompt.md

Example Prompt:

Create deployment plan for HIPAA-compliant patient portal.
Business context: Healthcare startup serving 10,000 patients.
NFRs: 99.95% SLA, RTO=4hrs, RPO=1hr, <2s page load.
Compliance: HIPAA, SOC 2, data residency in EU.
Budget: $3,000/month maximum.

Output: agent-output/{project}/01-requirements.md

architect

Phase 2 | Azure Well-Architected Framework assessment

Purpose: Provide architectural guidance based on Azure WAF pillars. Does NOT create any code—only assessment and recommendations.

Key Behaviors:

• Validates that Step 1 captured NFRs before proceeding
• Scores each WAF pillar (Reliability, Security, Cost, Operations, Performance)
• Recommends Azure services and patterns
• Integrates with Azure Pricing MCP for cost estimates

Example Prompt:

@architect Assess architecture for patient portal from Step 1.
Focus on HIPAA compliance and high availability requirements.

Output: agent-output/{project}/02-architecture-assessment.md

📖 Definition: .github/agents/architect.agent.md

diagram

Phases 3 & 7 | Python architecture diagrams

Purpose: Generate architecture diagrams using the diagrams Python library. Creates both .py source files and rendered .png images.

When to Use:

• Step 3: Design-phase diagrams (-des suffix)
• Step 7: As-built diagrams (-ab suffix)

Example Prompts:

@diagram Create architecture diagram for the patient portal design.

@diagram Generate as-built diagram showing deployed infrastructure.

Output:

• agent-output/{project}/03-des-diagram.py + .png
• agent-output/{project}/07-ab-diagram.py + .png

📖 Definition: .github/agents/diagram.agent.md

adr

Phases 3 & 7 | Architecture Decision Records

Purpose: Create ADRs documenting significant architectural decisions with context, options considered, and rationale.

When to Use:

• Step 3: Design-phase decisions (-des suffix)
• Step 7: Implementation decisions (-ab suffix)

Example Prompts:

@adr Create ADR for database selection decision (SQL vs Cosmos DB).

@adr Document the decision to use App Service instead of AKS.

Output:

• agent-output/{project}/03-des-adr-NNN-{title}.md
• agent-output/{project}/07-ab-adr-NNN-{title}.md

📖 Definition: .github/agents/adr.agent.md

bicep-plan

Phase 4 | Implementation planning with governance discovery

Purpose: Create detailed implementation plan including module structure, resource dependencies, and governance constraints from Azure Policy.

Key Behaviors:

• Queries Azure Resource Graph for policy assignments
• Generates governance constraints file
• Plans module structure and dependencies
• Integrates with Azure Pricing MCP for cost validation

Example Prompt:

@bicep-plan Create implementation plan for patient portal architecture.
Include module structure and deployment order.

Output:

• agent-output/{project}/04-implementation-plan.md
• agent-output/{project}/04-governance-constraints.md

📖 Definition: .github/agents/bicep-plan.agent.md

bicep-code

Phase 5 | Bicep code generation

Purpose: Generate production-ready Bicep templates based on the implementation plan. Creates modular, validated code following AVM-first approach.

Key Behaviors:

• Generates main.bicep + module files
• Creates deployment script (deploy.ps1)
• Validates with bicep build and bicep lint
• Follows unique suffix pattern for resource names
• Uses Azure Verified Modules where available

Example Prompt:

@bicep-code Generate Bicep templates for patient portal.
Follow the implementation plan from Step 4.

Output:

• infra/bicep/{project}/main.bicep
• infra/bicep/{project}/modules/*.bicep
• infra/bicep/{project}/deploy.ps1
• agent-output/{project}/05-implementation-reference.md

📖 Definition: .github/agents/bicep-code.agent.md

deploy

Phase 6 | Azure deployment orchestration

Purpose: Execute Azure infrastructure deployments using generated Bicep templates. Runs deploy.ps1 scripts, performs what-if analysis, and manages deployment lifecycle.

Workflow:

1. Pre-deployment validation (Azure CLI auth, Bicep build)
2. Run what-if analysis for impact assessment
3. Execute deployment after user approval
4. Generate deployment summary

Creates:

• 06-deployment-summary.md — Deployment record with resource IDs and endpoints

Example Prompt:

@deploy Deploy the Bicep templates for the ecommerce project to Azure.
Run what-if analysis first before actual deployment.

Output: agent-output/{project}/06-deployment-summary.md

📖 Definition: .github/agents/deploy.agent.md

docs

Phase 7 | Customer-deliverable documentation

Purpose: Generate comprehensive workload documentation package for customer handoff, including design docs, runbooks, and compliance matrices.

Creates:

• 07-documentation-index.md — Master index
• 07-design-document.md — Architecture design document
• 07-operations-runbook.md — Day-2 operational procedures
• 07-resource-inventory.md — Complete resource listing
• 07-compliance-matrix.md — Security control mapping
• 07-backup-dr-plan.md — DR procedures

Example Prompt:

@docs Generate complete workload documentation package
for the deployed patient portal infrastructure.

Output: agent-output/{project}/07-*.md (6 files)

📖 Definition: .github/agents/docs.agent.md

diagnose

Phase 8 | Post-deployment health assessment and diagnostics

Purpose: Interactive diagnostic agent that guides users through Azure resource health assessment, issue identification, and remediation planning. Operates on a single resource or entire resource group.

Key Behaviors:

• Approval-First Execution: All diagnostic commands require user approval before execution
• Interactive Guidance: Asks clarifying questions at each phase transition
• Educational Approach: Explains what each diagnostic step reveals and why it matters
• Flexible Scope: Supports single-resource or resource-group-level diagnostics

Diagnostic Workflow:

1. Resource Discovery - Identify and validate target Azure resources
2. Health Assessment - Evaluate availability, performance, and error rates
3. Log Analysis - Analyze logs and telemetry for patterns and issues
4. Issue Classification - Categorize by severity and root cause
5. Remediation Planning - Create and execute approved remediation actions
6. Report Generation - Document findings and actions taken

When to Use:

• Post-deployment validation before go-live
• Troubleshooting production issues
• Periodic health assessments of production workloads
• Before major changes or scaling events

Handoff Options:

• To architect - For architectural issues requiring WAF evaluation
• To docs - To incorporate health findings into as-built documentation

Example Prompt:

@diagnose Diagnose health of my-app-prod web app
in rg-myapp-prod. Check availability, performance, and recent errors.

Output: agent-output/{project}/08-resource-health-report.md

📖 Definition: .github/agents/diagnose.agent.md

Quick Reference Guide

By Task

By Workflow Phase

By Output Artifact

Related Documentation

Core Workflow Documents

• Workflow Guide — Complete 8-phase workflow with detailed examples
• Defaults Reference — Regional standards, naming conventions, tags
• Bicep Patterns — Infrastructure code best practices

Agent Definitions

All agent definitions are located in .github/agents/:

• requirements.agent.md
• architect.agent.md
• diagram.agent.md
• adr.agent.md
• bicep-plan.agent.md
• bicep-code.agent.md
• deploy.agent.md
• docs.agent.md
• diagnose.agent.md

Supporting Resources

• Azure Pricing MCP Server — Real-time pricing integration
• Copilot Instructions — Repository-level guidance
• Repository README — Project overview and getting started

Templates & Prompts

• Requirements Template — Phase 1 starting point
• Artifact Templates — Canonical output structures