Track
Newbie to Tech
Comfortable with tech, then pick a path
Pick the tool you like. Learn the habits that work in all of them. Build one small app, step by step, until it's live on the web with a tiny AI helper. We don't promise a job. We do promise you'll know your next step.
Practice project: https://github.com/jazken/mood-habit-tracker (Daily Mood and Habit Tracker, plain HTML/JS, MIT licence)
Kenneth's story of how he switched is coming soon.
The steps
Step 0 of 8
Pick your tool and get the practice project
Choose a tool, install it, turn on "ask me first", fork and clone the repo, open `index.html`.
Step 1 of 8
Read what your AI coding tool is telling you
Ask your tool to explain `index.html`, and spot the 5 kinds of messages.
Step 2 of 8
Stay in control: plans, approvals, undo
Plan, approve and then undo a "Stretch" habit button.
Step 3 of 8
Git part 1: save points you control
On branch `my-save`, add saving with localStorage and make your first commit (solo).
Step 4 of 8
Git part 2: work together
With a partner on GitHub's website: two branches, two PRs, one real merge conflict.
Step 5 of 8
Put your tracker online
Push `my-save`, merge it into your fork's `main` with a PR, turn on GitHub Pages.
Step 6 of 8
Your first real code: a week chart
The AI builds a 7-day chart with no library. You type the "X of 7 days" line yourself.
Step 7 of 8
A small AI build: your weekly coach
A practice coach note with no key. Optional: a real local AI with Ollama (free, no key). Secrets warning.
Step 8 of 8
Pick your next path
Write one page: what I built, what I liked, what's next. Pick roadmap.sh, Odin, CS50P, fast.ai or AIAP.
Lesson numbers vs repo branch names (they don't match)
| Repo branch/tag | What it holds | Used in lesson |
|---|---|---|
main | Starter: nothing saved | 0, 1, 2 |
step-1 | Saving with localStorage | 3 (answer key) |
step-2 | 7-day chart | 6 (answer key) |
step-3 | Partner conflict setup (built on step-2) | 4 (setup) |
step-4 | GitHub Pages instructions in the README | 5 (instructions) |
step-5 | Practice coach stub, no key | 7 (answer key) |
Why we kept the lesson order: the teaching order (Git solo → Git with a partner → online → code → AI) is the right order for learners. The repo's README already names "the Step 3 lesson" and "the Step 4 lesson" in this lesson order. Each lesson page says which branch it uses, and says plainly that the names don't line up. Renaming branches is optional.
The habits that run through every step
- Plan first, then approve, then undo with Git. A tool's own undo only covers some changes. A commit covers everything.
- The AI works on a branch. Never straight on
main. - Changes reach
main(and your live site) through a PR that a human reads. - Never "always allow" push or merge while learning.
- Never put a key in the repo. Your site is public, and Git remembers.
Pick-your-tool summary (prices re-checked 24 Oct 2026, SGT)
| Tool | Free or low-cost start |
|---|---|
| Kiro | Kiro Free $0/month, 50 credits · Pro $20 |
| OpenAI Codex | ChatGPT Free: Codex in the desktop app, "subject to rollout" · Go $8 · Plus $20 (adds the CLI, IDE extension, web) |
| Claude Code | Not on the free plan · Pro $17/month yearly or $20 monthly |
| Google Antigravity | Individual $0/month, "basic weekly rate limits" |
Sources: https://kiro.dev/pricing/ · https://learn.chatgpt.com/docs/pricing · https://claude.com/pricing · https://antigravity.google/pricing
How it fits with the rest
- Track 1, One Useful Task, teaches the same habits without code: ask before you write, check before you send.
- The Quest is a shared warm-up, which can sit before Step 0 or next to Step 7.