Skip to content

Skill and Subagent Reference

Skills

Skills are invoked automatically by agents, but you can also reference them directly in prompts.

azure-defaults

Provides regions, tags, naming conventions, AVM module references, and security baselines. This is the foundational skill — agents read it before every task.

@workspace What are the default required tags from azure-defaults?

azure-diagrams

Generates Python architecture diagrams using the diagrams library.

Generate an architecture diagram for the infrastructure in
infra/bicep/my-project/ using the azure-diagrams skill.

azure-bicep-patterns

Provides reusable Bicep patterns: hub-spoke networking, private endpoints, diagnostic settings, conditional deployments, and AVM module composition.

@workspace Show me the private endpoint pattern from azure-bicep-patterns.

terraform-patterns

Provides reusable Terraform patterns: hub-spoke networking, private endpoints, diagnostic settings, AVM-TF module composition, and known AVM pitfalls.

@workspace Show me the hub-spoke pattern from terraform-patterns.

azure-troubleshooting

KQL templates, metric thresholds, health checks, and remediation playbooks for diagnosing Azure resource issues.

@workspace What KQL queries are available in azure-troubleshooting?

azure-adr

Creates Architecture Decision Records following a structured template.

Document the decision to use Azure Front Door instead of
Application Gateway as an ADR.

git-commit

Provides conventional commit message conventions for this repository.

@workspace What commit message format does this repo use?

github-operations

Manages GitHub issues, PRs, Actions, and releases. Uses MCP tools first, falls back to gh CLI.

Create a GitHub issue for adding monitoring to the payment gateway.
Label it with 'enhancement' and 'infrastructure'.

docs-writer

Generates and maintains documentation following repository standards.

Update the docs to reflect the new Diagnose agent we added.

make-skill-template

Scaffolds a new skill directory from the template.

Create a new skill called 'azure-monitoring' for Application Insights
and Log Analytics best practices.

azure-artifacts

Artifact template structures, H2 compliance rules, and documentation styling for all agent outputs (all steps).

@workspace What H2 headings are required in the implementation plan template?

context-optimizer

Audits agent context window usage via debug logs, token profiling, and redundancy detection. Produces optimisation recommendations.

Analyse the last Copilot Chat debug log and identify context waste.

context-shredding

Runtime context compression with 3 tiers (full/summarised/minimal) and per-artifact templates to keep agents within context limits.

@workspace What compression tiers does context-shredding define
for the architecture assessment artifact?

copilot-customization

Authoritative reference for VS Code Copilot customisation mechanisms: instructions, prompt files, custom agents, skills, MCP servers, and hooks.

I want to create a new custom agent for database migration tasks.
Walk me through the steps using copilot-customization.

golden-principles

The 10 agent-first operating principles governing how agents work in this repository. Defines governance invariants and philosophy.

@workspace What are the golden principles for agent behaviour?

iac-common

Shared IaC patterns for deploy agents: CLI auth validation, deployment strategies, known issues, and governance-to-code property mapping.

@workspace What are the known deployment issues in iac-common?

session-resume

Session state tracking and resume protocol for the 8-step workflow. Manages checkpoint recovery and session-state.json schema.

Resume the workflow from step 4 using the existing session state.

workflow-engine

Machine-readable workflow DAG for the 8-step pipeline. Defines node types, edge conditions, gates, and fan-out patterns.

Subagents

Not user-invocable

Subagents are delegated to automatically by parent agents. You cannot select them from the agent picker (Ctrl+Shift+A). See Workflow Prompts for end-user scenarios.

@workspace Show the workflow graph edges and gate conditions.

Subagents

Subagents are called automatically by the Bicep CodeGen, Terraform CodeGen, Bicep Deploy, Terraform Deploy, Architect, and IaC Planner agents. You do not invoke them directly, but understanding their output helps you interpret validation results.

bicep-lint-subagent

Runs bicep lint and bicep build to validate template syntax. Returns a structured PASS/FAIL result with error counts and details.

bicep-review-subagent

Reviews Bicep templates against AVM standards, naming conventions, security baselines, and best practices. Returns an APPROVED, NEEDS_REVISION, or FAILED verdict with actionable feedback.

bicep-whatif-subagent

Runs az deployment group what-if to preview deployment changes. Analyzes policy violations, resource changes, and cost impact. Returns a structured change summary.

terraform-lint-subagent

Runs terraform fmt -check, terraform validate, and TFLint to validate configuration syntax. Returns a structured PASS/FAIL result with diagnostics.

terraform-review-subagent

Reviews Terraform configs against AVM-TF standards, CAF naming conventions, security baselines, and governance compliance. Returns APPROVED, NEEDS_REVISION, or FAILED verdict with actionable feedback.

terraform-plan-subagent

Runs terraform plan to preview infrastructure changes. Classifies resources into create/update/destroy/replace, highlights destructive operations, and returns a structured change summary.

cost-estimate-subagent

Queries Azure Pricing MCP tools for real-time SKU pricing. Compares regions and returns a structured cost breakdown.

governance-discovery-subagent

Queries Azure Policy assignments via REST API (including management group- inherited policies). Classifies policy effects and returns structured governance constraints.

Tips and Patterns

Context Priming

Open Files Before Prompting

Open relevant artifact files before starting a complex workflow step. Copilot uses open files as context, giving agents better awareness of your project state.

Before starting a complex workflow, open relevant files so Copilot has context:

  1. Open the requirements document (01-requirements.md)
  2. Open the architecture assessment (02-architecture-assessment.md)
  3. Then ask the Bicep Planner agent to create the implementation plan

Chaining Agents

You can chain agents manually by using handoff buttons in the chat, or run the Conductor for automatic orchestration. Manual chaining gives you more control over each step.

Bicep track:

  1. Run Requirements → review and approve 01-requirements.md
  2. Run Architect → review WAF scores and cost estimate
  3. Run Bicep Planner → review governance constraints and plan
  4. Run Bicep CodeGen → review generated templates
  5. Run Bicep Deploy → review what-if before approving deployment
  6. Run As-Built → generate post-deployment documentation

Terraform track:

  1. Run Requirements → review and approve 01-requirements.md
  2. Run Architect → review WAF scores and cost estimate
  3. Run Terraform Planner → review governance constraints and plan
  4. Run Terraform CodeGen → review generated configs
  5. Run Terraform Deploy → review plan output before applying
  6. Run As-Built → generate post-deployment documentation

Recovering from Errors

If an agent produces incorrect output, use specific follow-up prompts:

The VNet address space conflicts with our on-premises range (10.0.0.0/8).
Change the hub VNet to 172.16.0.0/16 and spoke VNets to 172.17.0.0/16.

Working with Existing Infrastructure

Agents can work with existing deployments, not just greenfield projects:

I have an existing resource group rg-legacy-app-prod with 15 resources.
Generate as-built documentation for this infrastructure.

Review the existing Bicep templates in infra/bicep/legacy-app/
and suggest improvements for WAF alignment.

References