hackernews-frontpage

Run this skill through the gstack daemon: it navigates to news.ycombinator.com, reads the page HTML, and parses each story row into a typed record with rank, title, URL, points, and comments. The result is emitted as a single JSON document on stdout.

Each story includes a 1-based rank, the HN item id, title, outbound URL, point count, and comment count. Job postings that lack scores return null for points and comments, and 'discuss' links are treated as zero comments.

No Hacker News credentials are needed since the front page is public. However, the script must be spawned via the skill runner, which sets GSTACK_PORT and GSTACK_SKILL_TOKEN, or run where a .gstack/browse.json state file provides the daemon port and token.

The parser relies on HN's stable tr.athing row structure and titleline spans. If HN rotates its markup, the bundled fixture test fails first, signaling that the regex selectors in parseStoriesFromHtml need updating before users are affected.

Yes. The parseStoriesFromHtml function is a pure function over HTML strings, and script.test.ts exercises it against a bundled fixture file using bun:test, with no daemon or network access required.