AI Coding Workstation Starter Kit

A beginner-friendly guide to set up your computer for AI coding in a safe, cloud-first way.

Who This Is For

• You use macOS, Linux, or Windows
• You can use a terminal, but want a simple setup path
• You want to learn and build without compromising your computer

Why Use This Kit

Use this kit if you want to:

• set up AI coding tools safely with clear boundaries
• build small AI helpers for real tasks (summaries, extraction, drafting, automations)
• start with cloud APIs first, then add complexity later only if needed
• keep your setup portable so you can rebuild it on another machine
• optionally run more privacy-focused local workflows on your own machine

Privacy: Cloud vs Local (Simple Version)

• Cloud mode: your prompts/data are sent to a provider API.
• Local mode: inference runs on your own machine.

Both are valid. This kit starts cloud-first for speed, then supports local options when they are practical.

Voice-to-Text Option (WhisperKit)

WhisperKit is an open-source toolkit for running Whisper speech-to-text locally (on-device).

In plain terms:

• it turns audio into text
• local use can be more private because audio does not need to leave your machine
• local use does not charge per API call from cloud providers

Money note:

• local Whisper/WhisperKit software is generally free to run
• if you use a hosted transcription API instead, that is usually paid per usage

Do You Need Docker?

Docker is optional for this starter kit.

Install Docker when you want to run local helper services on your machine, for example:

• a local app database (store project data like users, tasks, notes)
• a local document-search backend for “chat with my files”
• a self-hosted tool you want to run locally

If you are only calling cloud model APIs from Python/CLI, you do not need Docker yet.

Start Here (4 Steps)

1. Open the checklist: docs/checklist.md
2. Run discovery with Prompt A: docs/full-guide.md#prompt-a---hardware-and-system-discovery
3. Run design with Prompt B: docs/full-guide.md#prompt-b---safe-llm-workshop-design
4. Build your baseline and run one cloud test: docs/full-guide.md#first-post-build-cloud-win

Workspace at a Glance

/LLM-Workshop
├── repo/      # clone this repo here
├── work-rw/   # generated editable work
├── drop-ro/   # read-only source inputs
├── secrets/   # live keys and .env files (not in Git)
└── data/      # persistent service/app data

Document Map

• Full guide: docs/full-guide.md
• Beginner checklist: docs/checklist.md
• Visual folder diagram + mounts: docs/full-guide.md#visual-layout
• Prompt A command fallback: docs/full-guide.md#prompt-a---manual-command-fallback
• Next learning path: NEXT-STEPS.md
• Contribution notes: CONTRIBUTING.md

Checklist Tip

In repository Markdown files, task boxes are plain text until you edit the file.

• To mark progress in GitHub files, edit [ ] to [x] and commit
• For easier day-to-day tracking, open docs/checklist.md in Obsidian, VS Code, or another Markdown app
• If you want clickable task boxes on GitHub itself, use Issues or Pull Request task lists

Repository Structure

.
├── README.md
├── docs/
│   ├── checklist.md
│   └── full-guide.md
├── NEXT-STEPS.md
├── bootstrap.sh
├── CONTRIBUTING.md
├── LICENSE
└── .env.example

License

Licensed under CC BY 4.0.