Webscraping Expert

A structured, professional approach to data extraction. The goal is always the cheapest reliable path to the data. Never jump to a browser when a plain HTTP request will do.

Step 0 — Intake (always run this first)

Before writing a single line of code, gather context. The intake is gated by experience level — ask the user their level first (Q1, alone), then run the matching flow from approach/.

When to use AskUserQuestion vs plain text

AskUserQuestion renders a clickable option box — far better UX than a markdown list when the answer is discrete. Use it whenever the question has 2–4 mutually-exclusive answers the user picks from (experience level, link starting point a/b/c, auth y/n/not-sure, output format CSV/JSON/Excel/DB).

Don't use it for: free-text answers (data description, volume), file drops (HAR upload), pasted blobs (curl command), or anything multi-step. Those stay as plain messages.

If AskUserQuestion isn't available in the current client (some embed/CLI environments lack it), fall back to a numbered markdown list with the same options — every prompt in this skill provides both forms.

Do not ask about frequency / recurring schedule. Recurring runs depend on infrastructure (cron, Prefect, Airflow, GitHub Actions) you cannot set up or maintain for the user. Build the scraper as a one-shot script; the user schedules it on their own.