Hacker News Data

Pull stories, comments, and other items from Hacker News in two flexible ways: browse the live category lists (Top, New, Best, Ask, Show, Jobs), or scan every item posted on a specific date. Optionally fetch nested comment threads several levels deep.

Introduction

This actor queries the official Hacker News API to give you clean, structured access to stories and discussions — no scraping the website required. It works in two modes:

• Category mode: fetch the current stories in a given HN list (Top, New, Best, Ask HN, Show HN, or Jobs).
• Date mode: fetch every item (story, comment, job, or poll) posted on one specific UTC day, useful for historical analysis or building an archive.

Both modes support fetching nested comment replies to a configurable depth, so you can pull a story with its full conversation tree, not just the top-level comments.

Features

• Two query modes — pull the current front page, or scan a specific calendar day.
• Category filtering — Top, New, Best, Ask HN, Show HN, or Jobs.
• Nested comment threads — fetch replies up to 5 levels deep, with configurable breadth per level.
• Item type filtering — in Date mode, narrow results to just stories, comments, jobs, or polls.
• Honest completion signal — every Date mode run reports range_complete so you always know whether you got the whole day or need to page further, instead of silently returning a partial slice.
• Resumable scans — a next_page_token lets you continue a day that had more items than fit in one run.
• HTML-stripped comment text — comment bodies come back as clean, readable plain text.
• Budget-aware — automatically caps the number of results returned to stay within any spending limit you've set for the run.

Use Cases

• Content research — pull the current top stories on a topic-adjacent category (e.g. Ask HN) for trend analysis or newsletter curation.
• Historical archiving — scan a specific day of HN activity for research or dataset-building.
• Discussion analysis — fetch a story along with its full comment tree to analyze sentiment, common themes, or community reaction.
• Monitoring — periodically pull "New" stories to track when specific topics or domains get submitted.
• Dataset creation — build a training or evaluation dataset of real-world discussion threads with reply structure intact.

Input (UI & JSON)

The actor has two modes, chosen automatically based on whether you fill in Date to scan:

• Leave it empty → Category mode (pulls current stories from a category list).
• Fill it in → Date mode (scans every item posted on that one UTC day).

Field	Mode	Description
category	Category	Which HN list to pull from: top, new, best, ask, show, or job. Default: top.
date	Date	The single UTC day to scan, in YYYY-MM-DD format (e.g. 2026-07-01).
limit	Both	Max number of results to return in this run. Default 30, max 200.
item_type	Date	Restrict results to story, comment, job, or poll. Leave blank for all types.
comment_limit	Category	Number of top-level comments to fetch per story. Default 10, max 100, set 0 to skip comments.
reply_depth	Both	How many levels of replies to fetch below each comment. Default 0 (no replies), max 5.
replies_per_comment	Both	Max replies to fetch per comment at each level. Default 5, max 20. Only matters if reply_depth > 0.
page_token	Date	Resume token from a previous run's next_page_token, used to continue scanning the same day. Leave empty on the first run for a day.

Example — Category mode (JSON input):

{
  "category": "top",
  "limit": 20,
  "comment_limit": 5,
  "reply_depth": 1,
  "replies_per_comment": 5
}

Example — Date mode (JSON input):

{
  "date": "2026-07-01",
  "limit": 100,
  "item_type": "story",
  "reply_depth": 0
}

Example — continuing a day that wasn't fully scanned:

{
  "date": "2026-07-01",
  "limit": 100,
  "page_token": 48752931
}

Output (UI & JSON)

Results are pushed to the Apify dataset — one row per story (Category mode) or per matched item (Date mode). The Overview tab in the dataset UI shows all fields in a table; nested fields like comments and replies display as expandable JSON.

Common fields (all rows):

Field	Description
id	Hacker News item ID.
item_type	story, comment, job, or poll.
author	Username of the poster.
posted_at	Unix timestamp (UTC) when the item was posted.

Category mode row fields:

Field	Description
category	The category this story was fetched from.
title	Story title.
url	External link, if any.
points	Score.
comment_count	Total number of replies (all levels) on HN.
comments	Array of resolved top-level comments (up to comment_limit), each with nested replies if reply_depth > 0.

Date mode row fields:

Field	Description
title, url, points, comment_count	Present only on story/job/poll rows.
body	Comment or Ask HN text, HTML-stripped.
parent_id	Parent item ID, for comments.
comment_ids	Raw IDs of direct replies.
replies	Array of resolved replies (if reply_depth > 0).

Run metadata (key-value store, key OUTPUT):

Every run also writes a small summary object, separate from the dataset rows:

Field	Applies to	Description
mode	Both	get_stories or find_by_date.
date	Date mode	The day that was scanned.
count	Both	Number of rows returned in this run.
scanned	Date mode	Number of raw HN items internally checked.
next_page_token	Date mode	Pass this as page_token to continue scanning the same day. null if the day was fully covered.
range_complete	Date mode	true if this run scanned the entire requested day. false means the day had more items than fit in this run — check next_page_token.

Example output row (Category mode):

{
  "id": 48753715,
  "category": "top",
  "item_type": "story",
  "title": "ZCode – Harness for GLM-5.2",
  "author": "chvid",
  "url": "https://zcode.z.ai/en",
  "points": 234,
  "posted_at": 1782943416,
  "comment_count": 223,
  "comments": [
    {
      "id": 48752128,
      "author": "seizethecheese",
      "posted_at": 1782935074,
      "body": "I'm somewhat surprised that this is not open source...",
      "reply_count": 9,
      "replies": []
    }
  ]
}

Example run metadata (Date mode, day not fully scanned):

{
  "mode": "find_by_date",
  "date": "2026-07-01",
  "count": 100,
  "scanned": 2000,
  "next_page_token": 48752931,
  "range_complete": false
}

Support

For bugs, feature requests, or questions — reach us at https://docs.google.com/forms/d/e/1FAIpQLSfsKyzZ3nRED7mML47I4LAfNh_mBwkuFMp1FgYYJ4AkDRgaRw/viewform?usp=dialog