How skills/create-applications-with-the-copilot-cli works
Project type and purpose (from README + tree)
This repo is not the Node.js calculator app itself. It’s a GitHub Skills training/exercise template whose purpose is to teach developers how to use the standalone GitHub Copilot CLI for:
• creating/managing GitHub Issues (from templates),
• generating code iteratively,
• collaborating from the terminal,
• and progressing through a multi-step exercise that is orchestrated by GitHub Actions.
The “Node.js calculator app” is something the learner will create during the exercise; it’s not present in this repository’s file tree.
What the file tree tells you about the architecture
Key directories/files and what they imply:
GitHub Actions-driven exercise engine
• .github/workflows/0-start-exercise.yml
• .github/workflows/1-step.yml
• .github/workflows/2-step.yml
• .github/workflows/3-step.yml
• .github/workflows/4-step.yml
These are the “control plane.” They likely:
• detect learner actions (issue created, file changes, etc.),
• update the repo state (issues, PRs, comments, branches),
• and “unlock” next steps.
Curriculum content & guidance
• .github/steps/1-step.md ... 4-step.md
• .github/steps/x-review.md
These are the “content plane”: step instructions that the workflow(s) probably reference or post to the learner.
Issue templates for structured work items
• .github/ISSUE_TEMPLATE/feature_request.md
This enables the “create an issue from a template” part of the exercise. It’s a key artifact in the system.
Dev environment scaffolding
• .devcontainer/devcontainer.json
• .vscode (editor defaults)
This tells you the exercise expects an optional containerized dev environment, which is part of the learner-side architecture.
Static documentation assets
• README.md, images/, .github/images
• Developer/Learner workstation (terminal + Copilot CLI + Node.js runtime if/when the calculator is created)
• GitHub repository (content, issue template, workflows)
• GitHub Actions runners (execute workflows, check conditions, write back to repo)
• GitHub APIs (Issues, Checks, PRs, Contents)
• Copilot services (Copilot CLI calling model endpoints; auth via GitHub)
• Optional: Dev Container environment (Docker/VS Code Remote Containers)
Learner Environment
• Terminal
• GitHub Copilot CLI (standalone)
• Git client
• Node.js 22+ runtime (used when building the calculator)
• (Optional) VS Code + Dev Container
GitHub Platform
• GitHub Repository (this repo)
• .github/workflows/* (exercise automation)
• .github/steps/* (lesson content)
• .github/ISSUE_TEMPLATE/* (issue template)
• .github/instructions/* (CLI reference)
• GitHub Issues (issue objects created from template)
• GitHub Actions (workflow orchestrator)
• GitHub Actions Runner (execution environment)
Copilot/AI Services
• Copilot CLI backend / model endpoints (OpenAI/Google models abstracted behind Copilot)
• AuthN/AuthZ (GitHub identity + Copilot subscription entitlement)
External/Optional Tooling
• Model Context Protocol (MCP) servers (mentioned in README; optional integration point)
• Gists (for /share)
Primary flows explicitly:
Flow 1: Exercise initialization
• Learner → GitHub (copies template repo)
• GitHub Actions (0-start-exercise.yml) → GitHub Repo state (creates instructions, sets initial issue/comments, or prepares the next step)
Flow 2: Issue-driven workflow
• Learner (Copilot CLI or GitHub UI) → GitHub Issues (create issue from template)
• GitHub Actions (step workflow) → reads Issue events / repo contents → posts guidance/comments → marks step complete
Flow 3: Copilot CLI interactive development
• Learner Terminal → Copilot CLI → Copilot service (model inference)
• Copilot CLI → local filesystem changes (generates/edits Node.js calculator app files once created)
• Git client → pushes commits to GitHub Repo
• GitHub Actions → validates expected artifacts (files present, tests pass, etc.) → updates next step
Flow 4: Sharing/delegation (if used)
• Copilot CLI /share → GitHub Gist or markdown output
• Copilot CLI /delegate → Copilot coding agent → produces patches/commands → applied back to repo
Important architectural patterns
• Event-driven automation: GitHub Actions triggered by repo/issue events drives progression.
• Separation of planes:
• Content plane: .github/steps, docs/images
• Control plane: .github/workflows
• Interaction plane: Issues + Copilot CLI sessions
• Human-in-the-loop AI: Copilot CLI proposes; learner reviews/applies; workflows validate.
Technologies/frameworks that matter here
• GitHub Actions (YAML workflows)
• GitHub Issues templates
• Copilot CLI (standalone)
• Node.js 22+ (target runtime for the app being built)
• Dev Container (Docker + VS Code Remote, implied)
• GitHub API surfaces (Issues/Checks/Contents)
Elements explicitly:
• “creates issue”
• “triggers workflow”
• “calls Copilot inference”
• “pushes commits”
• “posts step feedback”
• Blue rectangles: compute/execution (Runner, workstation)
• Green rectangles: automation/control (Workflows)
• Orange rectangles: external services (Copilot, Gist, MCP)