Glossary
Agent mode
Section titled “Agent mode”VS Code chat mode where Copilot can plan, edit files, run commands, and use tools after you approve them. You first check it in Lab 00.
AGENTS.md
Section titled “AGENTS.md”A repository guide for AI coding agents. This repo’s root AGENTS.md summarizes setup, validation, boundaries, and custom agents. You read it in Lab 03.
Agent Package Manager. This lab uses it to install Skills for Fabric and discuss sharing skills. See APM and plugins and Lab 06.
Checkpoint
Section titled “Checkpoint”A known-good upstream tag that can restore the model state for a lab. Use Restore-LabCheckpoint.ps1 from Scripts and the catch-up block on lab pages.
Copilot CLI
Section titled “Copilot CLI”GitHub Copilot’s terminal surface. This lab uses it with hooks in Lab 11.
Copilot cloud agent
Section titled “Copilot cloud agent”A GitHub-hosted Copilot agent that researches, plans, changes code in its own environment, and can open a pull request. See Lab 13.
Copilot code review
Section titled “Copilot code review”Copilot’s pull request review feature. It reviews diffs, can use repo instructions, and consumes paid-plan or organization-enabled access. See Lab 12.
Custom agent
Section titled “Custom agent”A repo-defined agent in .github/agents/*.agent.md with a description, tools, instructions, and optional handoffs. See Agents and Lab 06.
Custom instructions
Section titled “Custom instructions”Guidance files Copilot reads as context. Repo-wide instructions live in .github/copilot-instructions.md; path instructions live under .github/instructions/. See Lab 03.
Display folder
Section titled “Display folder”A Power BI measure organization property. The lab requires measure display folders such as Volume, Refills, Processing, and Time Comparison. See Lab 08.
Grader
Section titled “Grader”The deterministic Python evaluator that scores question run files against expected answers. See Business questions and Lab 05.
Handoff
Section titled “Handoff”A custom agent button that switches to another agent and pre-fills a prompt. You use the assessor-to-improver handoff in Lab 07.
A deterministic guardrail that runs on Copilot events such as pre-tool-use or post-tool-use. Hooks can deny risky edits and run lint. See Hooks and Lab 11.
LineageTag
Section titled “LineageTag”A stable Power BI object identifier stored in TMDL. Keep it unchanged when renaming objects so Power BI can track identity. See Lab 07.
Marked date table
Section titled “Marked date table”A Power BI table designated as the canonical date table for time intelligence. The lab adds this in Lab 08.
Model Context Protocol, a way for Copilot to use external tools with approval. The lab connects Power BI Modeling MCP in Lab 04.
Measure
Section titled “Measure”A DAX calculation stored in the semantic model, usually used instead of implicit aggregations. The core measures are added in Lab 08.
Path instructions
Section titled “Path instructions”.instructions.md files with an applyTo glob, such as fabric/**/*.tmdl or rules/**. You create one in Lab 03.
Power BI Project, the folder-based format for a report and semantic model. You open fabric/ContosoPharmacy.pbip in Lab 02.
Power BI report files as JSON. This repo stores one visual.json per visual. Report field bindings are fixed in Lab 07.
Power BI Desktop
Section titled “Power BI Desktop”The Windows app that opens PBIP projects, refreshes the local model, and hosts the local Analysis Services instance for Desktop MCP and DAX tests. You install it in Lab 00.
Power BI Modeling MCP
Section titled “Power BI Modeling MCP”An MCP server that lets Copilot inspect or modify a Power BI model through Desktop or a PBIP folder. See MCP and Lab 04.
Prompt file
Section titled “Prompt file”A reusable prompt in .github/prompts/*.prompt.md, run with a slash command such as /assess-model. See Prompts.
Question run
Section titled “Question run”A JSON file under evaluation/runs/ that records answers, assumptions, and grades for the five business questions. See Lab 05 and Lab 10.
Semantic model
Section titled “Semantic model”The Power BI model that defines tables, columns, relationships, measures, and metadata. In this repo it lives under fabric/ContosoPharmacy.SemanticModel/definition/ as TMDL.
A reusable Copilot capability in .github/skills/<name>/SKILL.md, loaded when its description matches the task. See Skills and Lab 06.
Skills for Fabric
Section titled “Skills for Fabric”Microsoft-provided skills for Fabric and Power BI authoring. This repo installs the semantic-model-authoring skill with APM in Lab 06.
Synthetic data
Section titled “Synthetic data”Generated, deterministic, fictional pharmacy data. It lets the lab run without a Fabric tenant or real data. See Synthetic data model and Lab 01.
Tabular Model Definition Language, the text format for Power BI semantic model objects. You inspect it in Lab 02 and edit it in later labs.
Tool approval
Section titled “Tool approval”The prompt you see before an agent runs a terminal command, edits a file, or uses an MCP tool. It gives you a chance to read the action before allowing it. See Lab 04 and Lab 11.