.github Folder - GitHub Copilot Configuration

This folder contains GitHub Copilot configuration files that control how the AI assistant behaves in this repository. It is intentionally structured as a learning example of modern Copilot customisation.

---

File and Folder Reference

copilot-instructions.md

What it is: Repository-wide instructions that Copilot includes automatically in every conversation in this workspace.

When Copilot reads it: Always — no trigger needed. Any chat or inline suggestion in this repo will have these instructions in context.

What to put here: Project-wide conventions that should always apply, such as code style rules, naming conventions, technology choices, and tone guidelines.

This repo uses it for: Enforcing the no-emoji policy, TypeScript conventions, and reminding Copilot that this is an educational codebase.

Note: The canonical location is .github/copilot-instructions.md.

---

instructions/ folder

What it is: Scoped instruction files (.instructions.md) that Copilot applies automatically when working on files that match the applyTo pattern in the frontmatter.

When Copilot reads it: When the file being edited matches the applyTo glob pattern, or when manually referenced with # in the chat.

Frontmatter fields:

---
description: "Human-readable explanation shown in the UI"
applyTo: "glob/pattern/**/*.md"
---

What to put here: Conventions that only apply to a subset of files — for example, formatting rules for lab READMEs, API handler conventions, or test file standards.

This repo uses it for: lab-reviewer.instructions.md — rules for reviewing and writing lab README files, applied automatically whenever a labs/**/README.md file is in context.

---

prompts/ folder

What it is: Reusable prompt files (.prompt.md) that act as saved, parameterised commands you can invoke from the Copilot Chat UI.

When Copilot reads it: When you select the prompt from the chat input using the / command or the paperclip attach menu.

Frontmatter fields:

---
description: "What this prompt does"
argument-hint: "Description of the argument the prompt expects"
mode: "agent"    # "ask" (default), "edit", or "agent" for multi-step agentic tasks
tools:
  - read_file
  - run_in_terminal
  - mcp_github_create_issue   # MCP tools can be listed here too
---

What to put here: Multi-step workflows that you run repeatedly, such as auditing a folder, generating a report, or scaffolding a new lab. Think of these as reusable scripts written in natural language.

When mode: "agent" is set, the prompt runs with full tool access. You can restrict this by listing specific tools in the tools array — including any MCP server tools that are configured in your workspace. For example, the file-dependency-issues prompt in this repo uses GitHub MCP tools to create issues directly from the prompt.

This repo uses it for: