GDnD Wiki

GDnD Wiki

This site is for Ulster University students and staff. Enter the class access password to continue.

Beginner's Guide to the Terminal for Game Development, Git, and AI/LLM Work

3 min read

Source metadata

  • Type: structured beginner reference guide (programme teaching material).
  • Author: unattributed; compiled with links to official primary docs (Microsoft Learn, git-scm, GitHub, Unity, Unreal, Python, Ollama, LM Studio).
  • Audience: beginners working on games, code, Git, and AI/LLM workflows on Windows, macOS, or Linux.
  • Scope: cross-platform (PowerShell, Bash, Zsh) with worked command pairs throughout.

Key takeaways

  • Separates three commonly conflated ideas: the terminal emulator (the window), the shell (interprets typed commands — PowerShell, Bash, Zsh), and the command-line tool (does the work — git, dotnet, python, ollama, Unity).
  • Gives an opinionated, beginner-safe default setup per OS: on Windows, Windows Terminal + PowerShell 7+, with WSL 2 and Git Bash as options and winget for installs; on macOS, Terminal/iTerm2 + Zsh + Homebrew; on Linux, the distro terminal + Bash + the native package manager.
  • Covers the practical core: navigation and file commands (with PowerShell equivalents), opening tools (code .), checking installs (git --version), and package managers (winget, brew, apt, pip with venvs, npm).
  • Git for game projects is treated specifically: .gitignore for Unity, Git LFS for large binary assets, always committing .meta files, not committing generated builds, and writing small specific commit messages. This reinforces the wiki’s existing Git pages rather than replacing them.
  • Game-dev CLI patterns: Unity batch-mode builds (-quit -batchmode -projectPath … -executeMethod), Unreal command-line launches, and the .NET CLI (dotnet new/build/run/publish), plus wrapping repeated commands in build.ps1 / build.sh scripts.
  • AI/LLM workflows: isolated Python virtual environments, running local models with Ollama (ollama run/serve, API at localhost:11434) and LM Studio (lms), using terminal coding agents with a review-first loop, and keeping API keys in environment variables or ignored .env files.
  • Closes with safety guidance (treat rm -rf, sudo, curl … | sh with care; never commit secrets), practice exercises, and quick-reference command tables.

Notable claims

  • “The terminal is a text-based way to control your computer. It is not a replacement for Unity, Unreal, VS Code, Rider, Blender, or GitHub Desktop. It is the reliable layer underneath them.”
  • On agents: “Treat them as powerful assistants, not as unquestioned authorities” — inspect, plan, make a small change, review the diff, run tests yourself, commit only after review.
  • On secrets: “Do not paste API keys into source code. Use environment variables or ignored .env files.”

Relevance

This source underpins the new command-line-basics concept page and reinforces the existing Git track:

  • git-concepts — the guide’s Git model diagram matches the snapshots / three-areas model.
  • git-workflow — the same daily command set, plus git switch and the gh CLI.
  • git-github-unity — .gitignore, Git LFS, and .meta handling (the guide and this page agree closely).
  • source-control — binary-asset handling and LFS in a team context.

Its AI/LLM section also aligns with this project’s local-model workflow (Ollama via the Windows host) noted in the agent docs.

Open questions raised

  • The guide is intentionally tool-agnostic and broad; it does not commit to a single recommended toolchain for a specific module. Which subset should a CRE module standardise on for assessment?
  • It mentions terminal coding agents but does not address academic-integrity boundaries for their use in coursework — a policy question for the programme rather than the guide.

Graph View

Backlinks