Hacker News Archive: Stories, Comments & Hiring

Hacker News is a living archive of technology discussion: front-page links, Ask HN and Show HN threads, job ads, polls, user karma, and years of comments. This actor collects all of it from the public Hacker News surfaces — live feeds, keyword search, historical front pages, individual items, user profiles, and the monthly “Who is hiring?” threads. No login, cookies, API key, or proxy setup is needed.

What it does

• 11 modes — six live feeds, item lookup, user profiles, keyword search, historical front pages, and monthly hiring threads.
• Complete comment trees — flat rows or nested comments, with no duplicated replies and a correct maxItems cap.
• Deep search — relevance or newest-first ordering, tags, author, story, date range, points, and comment filters.
• Search past 1,000 results — adaptive date-window splitting keeps collecting when a query is larger than the search API’s per-query ceiling.
• Typed records — stories, comments, jobs, polls with vote options, poll options, user profiles, and parsed job postings.
• Historical front pages — read any day back to 2007 with its original ranking, score, author, and comment count.

Services

The actor reads three public Hacker News services:

Service	What it provides
Firebase Hacker News API	Live feeds, individual items, user profiles, scores, comment counts, and exact deleted/dead flags
Algolia Hacker News Search	Keyword search, date ordering, tag and author filters, and fast comment-tree reads
Hacker News HTML	Historical /front?day= pages and their original story rankings

All three are reachable without credentials.

How it works

Pick a mode and the actor chooses the right source for it:

• Feeds read the complete live list for the chosen section, then fetch item details and apply your filters. It keeps walking until the feed ends or maxItems is reached, so filtered-out stories do not silently shorten the result.
• Search sends your query, tags, author, story, and date range to Hacker News search. With exhaustiveSearch enabled it splits the date range whenever a window would exceed 1,000 hits, then merges the windows without duplicates.
• Comments use the fast threaded read first. When a thread is missing live replies, the actor checks the thread’s comment search and merges the gap. Asking for dead or deleted content switches to the exact live item walk.
• Historical front pages read page after page of the chosen day, then deduplicate stories that appear twice.
• Who is hiring? finds the requested number of monthly threads, reads their top-level posts, and extracts the common hiring fields without inventing missing details.

Every mode is filter-aware: a story is checked before its comments are fetched, so filtered-out stories cost nothing extra.

Input

Field	Type	Default	Description
mode	string	topStories	topStories / newStories / bestStories / askStories / showStories / jobStories / item / user / search / frontPage / whoIsHiring
itemIds	array	[]	Item IDs for mode=item
usernames	array	[]	Usernames for mode=user
searchQueries	array	[]	Keywords for mode=search; several queries are merged
searchSort	string	relevance	relevance or date
contentTypes	array	["story"]	story, comment, poll, pollopt, job, show_hn, ask_hn, front_page
author	string	–	Restrict search to one username
storyId	string	–	Restrict search to one story
exhaustiveSearch	boolean	false	Date-slice past the 1,000-hit search ceiling
dateRangeFrom	string	–	Earliest date (YYYY-MM-DD, UTC)
dateRangeTo	string	–	Latest date (YYYY-MM-DD, UTC)
minScore	integer	–	Minimum points for stories, jobs, and polls
minCommentCount	integer	–	Minimum comment count for stories and polls
domainAllowlist	array	[]	Keep only these link-host fragments
domainBlocklist	array	[]	Drop these link-host fragments
excludeDeadOrDeleted	boolean	true	Drop dead/deleted items
commentAuthorFilter	array	[]	Keep only comments by these authors
commentSort	string	rank	rank or newest sibling order
maxComments	integer	0	Comments collected per story; 0 skips comments
maxDepth	integer	5	Maximum reply depth; roots are depth 0
enableCommentHierarchy	boolean	false	Nested comments instead of flat comment rows
includeSubmissions	boolean	false	Also emit a user’s submissions
submissionTypes	string	all	all, story, or comment
frontPageDate	string	–	Day for mode=frontPage; empty means yesterday
hiringMonths	integer	1	Recent monthly hiring threads to parse
hiringThreadIds	array	[]	Specific hiring threads, overriding the month count
hiringKeywords	array	[]	Keep matching hiring posts
remoteOnly	boolean	false	Keep hiring posts that mention remote work
maxItems	integer	50	Total records emitted; nested comments count too
includeAuthorProfiles	boolean	false	Add author karma and account creation date
startUrls	array	[]	Item, user, front-page, or hiring-thread links

Empty fields are omitted from the output — there are no nulls or blank strings.

Example: search with filters

{
  "mode": "search",
  "searchQueries": ["rust async runtime", "tokio"],
  "searchSort": "date",
  "contentTypes": ["story", "comment"],
  "dateRangeFrom": "2025-01-01",
  "minScore": 50,
  "maxItems": 200
}

Example: full comment tree

{
  "mode": "item",
  "itemIds": ["49922569"],
  "maxComments": 300,
  "maxDepth": 10,
  "enableCommentHierarchy": true,
  "commentSort": "newest"
}

Example: a historical front page

{
  "mode": "frontPage",
  "frontPageDate": "2024-01-01",
  "maxItems": 500
}

Example: this month’s hiring posts

{
  "mode": "whoIsHiring",
  "hiringMonths": 1,
  "hiringKeywords": ["python", "backend"],
  "remoteOnly": true,
  "maxItems": 100
}

Output per story, job, or poll

• id, hnUrl, type (story, job, or poll), title
• url, domain for link posts
• score, author, numComments
• createdAt (ISO-8601 UTC), createdAtEpoch, ageHours
• text for Ask HN, Show HN, jobs, and polls
• kids — top-level comment IDs
• rank — feed, search, or front-page position
• frontPageDate for historical rows
• pollOptions[] — id, text, votes, author, createdAt, hnUrl
• fetchedCommentCount when comments were read
• comments[] — nested replies when hierarchy is enabled
• authorKarma, authorCreatedAt when author enrichment is enabled
• dead / deleted only when true
• recordType: "story", scrapedAt

Output per comment

• id, hnUrl, type, author, text
• storyId, parentId, depth
• createdAt, createdAtEpoch, ageHours
• kids — reply IDs
• storyTitle, storyUrl when a search hit supplies them
• replies[] in nested mode only
• dead / deleted only when true
• recordType: "comment", scrapedAt

Output per user

• id, profileUrl
• karma, createdAt, createdAtEpoch, about
• submittedCount, submitted (first 50 recent IDs)
• recordType: "user", scrapedAt

With includeSubmissions, normal story and comment rows follow the profile.

Output per poll option

• id, hnUrl, type: "pollopt", text, votes, pollId
• author, createdAt, createdAtEpoch, ageHours
• recordType: "pollOption", scrapedAt

Output per hiring post

• id, hnUrl, threadId, threadTitle, threadMonth
• author, company, role, location, remote, salary
• techStack, visa, email, applyUrl
• text, postedAt
• recordType: "jobPosting", scrapedAt

Parsed hiring fields are omitted when the post does not state them.

Usage cost

There is no paid data source, proxy, or credential in this actor. Usage is ordinary Apify compute plus the records you choose to collect. The cheapest useful run is the default prefill: 20 top stories with comments and author enrichment off. Cost grows with maxItems, maxComments, maxDepth, multiple search queries, exhaustiveSearch, and includeAuthorProfiles because each adds fetches. Small maxItems values keep runs short and inexpensive.

FAQ

Do I need an account, cookie, or API key? No. Every mode reads public Hacker News data.

Do I need a proxy? No. The actor connects directly and needs no proxy configuration.

Why is a field missing from some records? Hacker News does not store that field for every item. For example, self posts have text but no link URL, and many users have no “about” text. Missing fields are omitted instead of filled with a placeholder.

Why do comment scores never appear? Hacker News does not publish per-comment scores through its feeds, search, or HTML. Stories, polls, jobs, and poll options do have scores.

Why did a search stop at 1,000 results? Hacker News search returns at most 1,000 hits per query window. Turn on exhaustiveSearch to split the date range into smaller windows and collect deeper results.

Why is relevance order different when exhaustiveSearch is on? The deep search collects several date windows. Each window is ranked by relevance, so the final list follows window order rather than one global relevance ordering. Use searchSort: "date" for a stable chronological order.

How fresh is the data? Feeds and item lookups are live. Search and comment search can lag the live site by a short period, so brand-new comments or score changes may appear slightly later.

What is the difference between maxItems and maxComments? maxItems caps every emitted record, including nested comments. maxComments caps how many comments are read for each story before depth filtering.

Can I read a historical best, Ask HN, or Show HN list? No. Hacker News archives only the historical front page by date. Live Ask HN, Show HN, best, and job lists are available through their normal modes.

Is this affiliated with Hacker News or Y Combinator? No. This is an independent actor that reads Hacker News’ public data.

Notes

• All timestamps are UTC and ISO-8601, with a matching Unix epoch.
• Records are snapshots: scores, comment counts, and comment trees can change after a run.
• Flat comment output contains one row per comment and never repeats a reply inside another row.
• Nested comment output is one story row; every reply carries its own type and scrape timestamp.
• item returns a comment row for a comment ID, a story-style row for a story, job, or poll, and a poll-option row for a poll-option ID.
• startUrls accepts item, user, historical front-page, and hiring-thread links alongside the mode-specific lists.
• Historical front pages are deduplicated when a story appears twice on the same day.

Limitations

• No comment scores or flags. Hacker News does not expose per-comment scores, moderation flags, or edit history publicly.
• Search ceiling. Each Hacker News search window returns at most 1,000 hits. exhaustiveSearch works around this by date slicing, but a single second with more than 1,000 hits cannot be split further.
• Historical lists. Only the front page is archived by day. Historical best, Ask HN, Show HN, and jobs pages are not available.
• Feed caps. Hacker News itself limits the live feeds: top and new stories usually expose 500 IDs, best about 200, Ask HN 20, Show HN up to about 150, and jobs about 30. The actor reports when a feed is exhausted.
• Comment-tree lag. The fast tree can trail the live thread by a short period, and deleted branches may be absent. The actor merges a comment-search fallback when live replies are missing; turning off excludeDeadOrDeleted uses the exact slower walk.
• Hiring parsing is heuristic. Company, role, location, salary, tech stack, visa, email, and apply link are extracted from free-form posts. Fields the post does not state are omitted; no value is invented.
• Very large user histories. A user can have tens of thousands of submissions. maxItems and submissionTypes keep the run bounded.
• Relevance is not deterministic. Search relevance can shift between runs as the index changes; date order is stable.