Hacker News Topic Intelligence

Type a topic and get the Hacker News stories people actually argued about, ranked, with the top comments next to each one. No story IDs, no API code. Leave the default input as it is and you get the 30 most-discussed "AI agents" stories from the last 30 days in about 15 seconds.

What you get (real output, Oct 15 2026, default input)

Rank	Story	Points	Comments	Points/hour	Domain
1	Feds Target AI Critics as "Foreign Agents"	396	457	0.75	kenklippenstein.com
2	AI coding agents generate more code, but not more software	364	310	4.45	arstechnica.com
3	There are no "rogue" AI agents	396	269	0.89	eoinhiggins.substack.com
4	Early rogue AI agent activity and attempts to hack found on urlquery.net	267	313	0.51	transluce.org
5	Nvidia wants to put a watchdog chip next to every AI agent	230	298	0.55	cnbc.com

Each row also has the top 3 comments in HN's own ranking order, as flat columns. For example, the top comment on story #2:

"This is from March 2026, but it does match what I've encountered professionally. Code review was always a significant bottleneck and is even more so with agentic coding…"

You also get a plain-text SUMMARY report in the key-value store: top stories, demand signals and the domains that get shared most.

Why this one

	This Actor	Typical HN scrapers
Search by any topic, ranked by engagement	✅ points + 2× comments, or by velocity	Usually front page or keyword list only
Top comments in the same row (CSV-ready)	✅	Separate comment dumps or none
Velocity (points per hour) to spot rising stories	✅	❌
Demand signals: "is there a tool…", "alternatives to…", Ask HN questions	✅ flagged per row	❌
Monitor mode: only stories you haven't seen	✅	Rare
Precise title matching or broad matching	✅	❌
Official public APIs only (Algolia HN Search + HN Firebase)	✅	Varies

Use cases

• Founders and PMs: see which pains people complain about, mine Ask HN for "is there a tool for X" requests, and validate ideas before you build.
• Marketers and DevRel: find which domains and angles get traction for your topic, and track mentions of your product or competitors on a schedule.
• Investors and analysts: weekly trend reports on any technology, with the real sentiment from comments.
• AI agents and RAG: clean, flat JSON from one call, so it works well as an MCP or agent tool.

Input

Field	Default	What it does
topics	["AI agents"]	One ranked list per topic (a product, company, domain or keyword)
daysBack	30	Time window
maxStoriesPerTopic	30	Top N after ranking
sortBy	engagement	engagement, points, comments, velocity or newest
matchIn	title	title is precise; everywhere also searches story text and URLs
storyTypes	all	Limit to ask_hn to mine questions, or show_hn to track launches
topCommentsPerStory	3	0 to 10 (0 is cheaper)
minPoints	5	Drops noise
onlyNewSinceLastRun	false	Monitor mode. Schedule it daily to get only new stories

Pricing (pay per result)

• Story with top comments: $2.00 per 1,000 ($0.002 each)
• Story without comments: $1.00 per 1,000

The default run (30 stories with comments) costs about $0.06. A daily monitor of 3 topics at roughly 20 new stories a day comes to about $3.60 a month. You never pay for empty runs or errors.

Output fields

rank, topic, title, storyType, points, comments, engagementScore, pointsPerHour, commentToPointRatio, demandSignal, url, domain, hnUrl, author, createdAt, ageHours, storyText, topComment1..N, topComment1By..N, storyId

API / MCP

curl -X POST "https://api.apify.com/v2/acts/forgelabs~hacker-news-topic-intelligence/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" -d '{"topics":["Postgres","SQLite"],"daysBack":7}'

You can also call it from any MCP client through Apify's MCP server (mcp.apify.com).

FAQ

Is this allowed? Yes. It uses only the official public Hacker News APIs (hn.algolia.com and hacker-news.firebaseio.com). No login and no scraping of HTML pages. How is demandSignal computed? A phrase heuristic ("is there a tool", "looking for", "alternatives to", "I'd pay"…) applied to the title and text, plus Ask HN questions. Treat it as a filter that helps you find candidates, not as ground truth. Why fewer stories than I asked for? Only stories matching your topic inside the time window, and above minPoints, are returned. Widen daysBack, lower minPoints or set matchIn to everywhere.