Hacker News Search API: Stories and Comments

Search Hacker News stories and comments by keyword, author, date range, points and type, and get back small, clean JSON. Built for AI agents, research scripts and keyword alerts: one call, no HTML to parse, no API key.

Data comes from the official Hacker News search API (hn.algolia.com, the search linked at the bottom of every Hacker News page). That means no scraping of news.ycombinator.com, no blocked pages, no captchas and no proxy costs, which is why the price per result is this low.

What you can do with it

• Find discussions about a topic: every HN story about "vector database" with 50+ points this year.
• Monitor mentions: schedule it with Only new since last run and get just the stories or comments that appeared since the last check (brand, product, competitor or job alerts).
• Read the room: turn on Include comment threads and each story comes with its comments as plain text, in thread order, ready for an LLM to summarize.
• Track launches: newest Show HN or Launch HN posts, or what is on the front page right now.
• Follow a person's public posts: filter by HN username.

What you get

One dataset item per story or comment:

• title, url (the linked article), hnUrl (the HN discussion)
• points, numComments, createdAt (ISO 8601, UTC)
• type (story, comment, poll, job) and labels (show_hn, ask_hn, launch_hn, front_page)
• author: the poster's public HN username
• text: Ask HN or Show HN text, or the comment, as plain text (HTML and entities removed, links kept as URLs)
• For comments: storyId, storyTitle, storyUrl, parentId
• comments (only with comment threads on): id, parentId, depth, author, text, createdAt for each comment

Fields the source does not have are null, never guessed.

Input

Field	Example	Notes
query	sqlite	Keywords. Quotes for an exact phrase: "vector database". Empty = everything matching the other filters
type	story	story, comment, show_hn, ask_hn, launch_hn, job, poll, front_page, or all
author	pg	HN username, case-sensitive
dateFrom	2026-10-01 or 7 days	From this date or time (UTC). Relative: 24 hours, 7 days, 2 weeks
dateTo	2026-10-31	Up to this date (a plain date includes the whole day)
minPoints	100	At least this many points
minComments	20	At least this many comments
sortBy	relevance	relevance (best match, up to 1,000) or date (newest first, up to 5,000)
maxItems	50	1 to 5,000, default 50. You pay only for results returned
includeComments	true	Add each story's comment thread
maxCommentsPerStory	100	1 to 500, default 100
onlyNew	true	Return only results this search has not returned before

{
    "query": "postgres",
    "type": "story",
    "minPoints": 200,
    "dateFrom": "2026-01-01",
    "maxItems": 20,
    "includeComments": true,
    "maxCommentsPerStory": 50
}

Output example

{
    "id": "27016630",
    "type": "story",
    "labels": [],
    "title": "Hosting SQLite databases on GitHub Pages or any static file hoster",
    "url": "https://phiresky.github.io/blog/2021/hosting-sqlite-databases-on-github-pages/",
    "hnUrl": "https://news.ycombinator.com/item?id=27016630",
    "author": "phiresky",
    "points": 1812,
    "numComments": 244,
    "createdAt": "2021-05-02T16:43:15.000Z",
    "text": null,
    "storyId": null,
    "storyTitle": null,
    "storyUrl": null,
    "parentId": null,
    "comments": [
        {
            "id": "27016787",
            "parentId": "27016630",
            "depth": 1,
            "author": "Gehinnn",
            "text": "TL;DR:\nCompile SQLite to JS with emscripten, implement a virtual fs in JS to stream chunks of a statically hosted (readonly) SQL database. ...",
            "createdAt": "2021-05-02T17:00:45.000Z"
        }
    ],
    "query": "sqlite",
    "scrapedAt": "2026-10-31T18:48:21.631Z"
}

Pricing

Pay per result. Comment threads are included at no extra charge.

Event	Price	When
Actor start	$0.00005	Once per run (Apify's standard start event)
Result	$0.00015	Every story or comment in your dataset

Example: 1,000 results cost $0.15 plus $0.00005 for the start.

Searches that find nothing are not charged per result, and neither is a story whose comment thread could not be read (it is saved without comments, free). The actor reads your run's maximum charge before it searches and returns only as many results as that covers, so a run never goes over the limit you set.

Only-new mode for alerts

With onlyNew on, the actor remembers which results each search already returned (per search and per saved task, in a key-value store in your own Apify account) and returns only new ones. Schedule a task every hour and you get a feed of new mentions. Set maxItems above the number of new results you expect between two runs. A run with nothing new returns no items and the status message says so.

Limits

• Relevance search returns at most 1,000 results per run (the search API's limit). Sort by date to get up to 5,000.
• Data is as fresh as the HN search index, usually within minutes of posting (a cloud test on 2026-10-31 started 38 seconds after the newest comment it returned was posted). Points and comment counts are current at search time.
• Comment threads add about half a second per story, so 100 stories with threads take under a minute.
• Deleted and dead comments are left out of threads (their live replies stay).

FAQ

Where does the data come from? From the public Hacker News Search API run by Algolia at hn.algolia.com/api, which Hacker News links to as its search. The actor stays far below its published limit of 10,000 requests per hour.

Why did I get fewer results than maxItems? The search had fewer matches, your filters narrowed it, only-new mode skipped results you already have, or your run's maximum charge covered fewer results.

Can I call it from an AI agent or MCP client? Yes. Inputs are plain fields, the output is small JSON with plain-text bodies, and it runs with limited permissions, so it needs no access to your other Apify data. It also works from schedules and the API.

Is there personal data in the results? Results are public Hacker News posts. The author field is the poster's public HN username, as shown on the site. The actor adds nothing else about people: no profiles, emails or karma.

Something broke or a field is missing. What do I do? Open an issue on the Issues tab with your run link. Issues are answered within a few days.