A tiny review collector

You are helping me — a regular person, not a developer — build a small PERSONAL
tool that collects reviews from a website into a tidy file I can read. Go one
step at a time and wait for my OK before each step.

WHAT WE'RE BUILDING
A little script that opens a REAL browser (Playwright) to visit a page I choose
and pull the reviews off it — rating, text, author, date — then saves them to a
file. We use a real browser because many review pages only load their content
with JavaScript, so simple scrapers get an empty page; a real browser sees what
I'd see.

RESPONSIBLE USE (build this in):
- For MY OWN small, personal use, on sites I'm allowed to look at.
- Go gently: one page at a time, with pauses — never hammer a site.
- Respect the site: if it asks me to log in or solve a captcha, or says don't
  automate it, we STOP and I decide — we don't try to break past that.
- Only the public review text on the page; no personal-data harvesting, no
  reselling, no bulk crawling.

STEP 1 — calibrate to me (ask, then wait):
1. How much help? (a) hold my hand (b) normal (c) quicker
2. Explanations: (a) plain (b) technical (c) just do it
3. Want the "why", or just the tool?

STEP 2 — design MY version (ask, then build to my answers):
A. WHICH PAGE(S): I'll give you one or more URLs to start with.
B. WHAT TO CAPTURE: rating / text / author / date — skip any I don't want.
C. HOW MANY: just the first page, or follow "next page" up to a limit I set.
D. OUTPUT: (1) CSV for a spreadsheet  (2) JSON  (3) a simple local HTML table.
E. RUN: just when I run it by hand (default) is plenty for now.

STEP 3 — build it small and clear:
- Use Playwright (Node). Everything runs locally with commands you show me.
- No cloud, no accounts, no keys — runs on my machine, writes a local file.
- Build in stages I can watch: open the page and print the title, then find
  one review, then all reviews on the page, then save to my format, then (if
  I chose) a "next page" loop with a delay and a hard limit.
- Keep it small and tidy; explain each part in plain words. When the site
  changes and it breaks, show me the one line to fix.

STEP 4 — make it mine and save it:
- Run it on my real target, check the output together.
- When I'm happy, put it in a NEW, PRIVATE GitHub repo and push. Private by
  default. I do logins myself; never ask me for a password or token.

RULES: official tools only; keep everything on my machine; explain before you
install or run anything and wait for my OK. Start with STEP 1.