Hacker News — AI Navigation Skill

What This Skill Does

Provides AI browser agents with pre-mapped selectors, page structures, and navigation flows for news.ycombinator.com. Eliminates the need to visually scan the page — use these selectors directly.

Site Overview

Property	Value
Domain	news.ycombinator.com
Site Type	Multi-page / MPA
Auth Required	No (public access)
Mapped	2026-02-20

Site Map

Page	URL	Purpose
Homepage	https://news.ycombinator.com	Hacker News
news	https://news.ycombinator.com/news	Hacker News
newest	https://news.ycombinator.com/newest	New Links
front	https://news.ycombinator.com/front	2026-02-19 front

Page Structures

Homepage: Hacker News

Form Inputs

Element	Selector	Notes
q	[name="q"]

Content Links (Sample)

Link Text	Selector	Notes
Hacker News	a:has-text("Hacker News")	news
comments	a:has-text("comments")	newcomments
submit	a:has-text("submit")	submit
gist.github.com	a:has-text("gist.github.com")	from?site=gist.github.com
interpidused	a:has-text("interpidused")	user?id=interpidused
5 hours ago	a:has-text("5 hours ago")	item?id=47082496
88 comments	a:has-text("88 comments")	item?id=47082496
bbc.com	a:has-text("bbc.com")	from?site=bbc.com
tartoran	a:has-text("tartoran")	user?id=tartoran
2 hours ago	a:has-text("2 hours ago")	item?id=47083735
19 comments	a:has-text("19 comments")	item?id=47083735
together.ai	a:has-text("together.ai")	from?site=together.ai
zagwdt	a:has-text("zagwdt")	user?id=zagwdt
6 comments	a:has-text("6 comments")	item?id=47083648
gustedt.wordpress.com	a:has-text("gustedt.wordpress.com")	from?site=gustedt.wordpress.com

For content-heavy sites, iterate with: page.locator('a').filter({ hasText: /pattern/ })

news: Hacker News

Inputs

Element	Selector	Notes
q	[name="q"]

newest: New Links | Hacker News

Inputs

Element	Selector	Notes
q	[name="q"]

front: 2026-02-19 front | Hacker News

Inputs

Element	Selector	Notes
q	[name="q"]

Navigation Flows

Flow 1: Browse Homepage Content

1. Navigate to https://news.ycombinator.com
   wait: networkidle OR domcontentloaded

2. Content loads automatically
   wait: page has settled (500ms)

3. Scroll to explore content
   action: page.evaluate(() => window.scrollBy(0, 500))
   wait: 300ms for lazy content

Authentication

Public access — no authentication required for main content.

Dynamic Content Patterns

Pattern	Description	How to Handle
Lazy loading	Content loads on scroll	page.evaluate(() => window.scrollBy(0, 500)) then wait 500ms
SPA navigation	URL changes without full reload	Wait for networkidle or specific element visibility
Infinite scroll	More items load at bottom	Scroll loop until no new items appear

Timing & Waiting

Action	Wait For	Typical Delay
Page navigation	waitUntil: 'domcontentloaded'	1-3s
After click	Element visibility change	300-800ms
After form submit	URL change or success message	1-4s
After scroll	New content appearance	300-600ms
Dynamic content	waitForSelector() on target element	500-2000ms

Common Gotchas

1. Rate limiting — news.ycombinator.com may throttle rapid requests. Add 1-2s delays between page loads.
2. Cookie consent — If a consent banner appears, click "Accept all" or equivalent before interacting: page.locator('[class*="consent"] button, button:has-text("Accept")').first().click()
3. Dynamic class names — Avoid selectors based on hashed class names (e.g. .a3f8b). Use aria-label, data-testid, or text-based selectors instead.
4. Viewport-dependent layout — These selectors were mapped at 1280×800. Mobile layout may differ.
5. Stale selectors — Sites update their markup. If a selector fails, fall back to text-based: page.locator('text="Button Label"')

Element Finding Strategy (Fallback Order)

1. data-testid / data-cy        → most stable
2. aria-label / role            → semantic, stable
3. id attribute                 → usually stable
4. name attribute (forms)       → stable for inputs
5. text content                 → :has-text("...")
6. CSS class (first class only) → fragile, last resort