Skip to content

Glossary

Key terms and definitions for Agentic InfraOps.

Note

Canonical vocabulary: The terms below are the agreed names used throughout all participant-facing docs, facilitator guides, and scripts. Use these exact terms when writing or updating workshop content.

Naming Conventions

Concept Canonical term Do not use
The event MicroHack or microhack hackathon, workshop (except in general descriptions)
Event organiser Facilitator coach, instructor, proctor
Event participant Participant attendee, student, user (in workshop context)
Challenge tasks Challenge 1, Challenge 2, etc. Task, Exercise, Lab
The surprise DR requirement Curveball (Challenge 4) twist, surprise, pivot
The fictional company Nordic Fresh Foods NFF, the customer (reserve "customer" for the C8 role)
The platform being built FreshConnect the app, the project (when referring to the scenario)
Azure deployment language Bicep ARM templates (unless comparing)
AI assistant GitHub Copilot Copilot (alone is acceptable), AI, the agent (when referring to the product)
Specific AI persona Agent (e.g., requirements agent) bot, assistant
Scoring document Scoring rubric rubric, scorecard
Working repository Template repo (azure-agentic-infraops-accelerator) starter repo, base repo
This documentation repo Docs repo (microhack-agentic-infraops) main repo, source repo
Team subscription model One subscription per team shared subscription (prohibited)
Post-event resource deletion Cleanup teardown, decommission

Azure Product Names

Use the full product name on first reference, then the short form:

Full name Acceptable short form
Azure App Service App Service
Azure SQL Database SQL Database
Azure Blob Storage Blob Storage, Storage Account
Azure Key Vault Key Vault
Azure Policy Policy
Azure Monitor / Application Insights App Insights
Azure Resource Manager ARM
GitHub Copilot Pro+ Copilot Pro+
GitHub Copilot Enterprise Copilot Enterprise
Azure Well-Architected Framework WAF

A · B · C · D · F · G · H · I · L · M · P · S · V · W


A

ADR — Architecture Decision Record. A structured document capturing a design decision, its context, and consequences. Generated by the azure-adr skill.

Admin — A Leaderboard App role (admin in SWA). Admins can review and validate team submissions, manually override published scores, assign awards, manage teams, and upload JSON score files. Compare with Member.

Agent — A specialized AI assistant with a defined role, model, and set of tools. Agents are defined in .github/agents/*.agent.md.

Approval Gate — A mandatory pause point in the 7-step workflow where the user reviews and approves output before the next step proceeds.

AVM — Azure Verified Modules. Microsoft's official library of pre-tested Bicep modules published to mcr.microsoft.com/bicep.

B

Bicep — Azure's domain-specific language for deploying Azure resources declaratively. Files use the .bicep extension.

C

Challenge — A time-boxed task in the microhack aligned to one or more workflow steps. There are 8 challenges progressing from requirements through deployment, documentation, diagnostics, and a final showcase.

Conductor — The InfraOps Conductor agent (🎼 Maestro). Master orchestrator that coordinates all other agents through the 7-step workflow.

Conventional Commits — A commit message format (type: description) used for automated versioning. Enforced by commitlint and lefthook.

Curveball — A surprise requirement change announced mid-microhack (Challenge 4). Simulates real-world scope changes by introducing multi-region disaster recovery requirements.

D

Dev Container — A Docker-based development environment defined in .devcontainer/devcontainer.json. Provides consistent tooling for all contributors.

F

Facilitator — The event organiser who runs the microhack: manages the schedule, announces the curveball, coaches teams, and runs the scoring scripts.

FreshConnect — The project codename for the Nordic Fresh Foods delivery platform used as the microhack scenario. All agent output goes to agent-output/freshconnect/.

G

Governance Constraints — Azure Policy assignments that restrict resource configurations. Discovered during Step 4 (Planning) and saved to 04-governance-constraints.md.

H

Handoff — The mechanism by which the Conductor agent delegates work to a specialized agent and passes context (artifact file paths).

I

Instruction File — A .instructions.md file in .github/instructions/ that provides file-type-specific coding standards to Copilot.

L

Leaderboard App — An optional browser-based HackerBoard web application built on Azure Static Web Apps. Some workshop packages use it for score submission and review, but it is not included in this repository.

M

MCP — Model Context Protocol. A standard for connecting AI models to external tools and data sources. This repo uses Azure MCP and a custom Pricing MCP server.

Member — A Leaderboard App role (member in SWA). Members can view the leaderboard, see their own team's score breakdown, register their attendee profile, and submit scores for their own team. Compare with Admin.

P

Persona — A named character assigned to each agent for easy recognition. For example, the Requirements agent is 📜 Scribe and the Architect agent is 🏤 Oracle. See the agent overview.

S

Scoring — WAF-aligned evaluation of microhack submissions. Base score is 105 points across eight scored categories plus up to 25 bonus points for advanced features like zone redundancy and private endpoints.

Skill — A reusable knowledge module defined in .github/skills/*/SKILL.md. Skills provide domain-specific instructions that agents invoke based on keyword matching.

Submission — A score payload submitted by a team member through the Leaderboard App. Submissions enter a Pending state and must be approved by an admin before they affect the published leaderboard.

Subagent — A lightweight agent invoked by another agent for a specific validation task (lint, what-if, review).

V

Validation — In the Leaderboard App context, the admin review step where a pending submission is approved or rejected before scores are published. In the infrastructure context, the az deployment group validate or bicep build step that checks template correctness.

W

WAF — Well-Architected Framework. Microsoft's design framework with five pillars: Security, Reliability, Performance Efficiency, Cost Optimization, and Operational Excellence.

What-if — An Azure deployment preview that shows what resources would be created, modified, or deleted without actually making changes. Run via az deployment group what-if.