How this works

HackerProducts answers one question for each product: why do Hacker News people use this, and who says so? It is a prototype built from one year of comments (Nov 2025 to Oct 2026).

Pipeline

  1. Ingest. 3,768,623 comments from the open-index/hacker-news dataset and the Algolia HN API.
  2. Prefilter. Rules keep comments with recommendation phrasing ("I use", "switched to", "highly recommend"), links to non-news sites, or a place in an Ask HN thread about tools: 347,262 candidates.
  3. Extract. Claude Sonnet labels the highest-priority candidates (75,185 so far, 34,513 of which name a product): each product named, the commenter's stance, a verbatim quote, what they use it for, alternatives, and whether they built it. On a 500-comment check against Claude Opus, Sonnet scored 87.8% F1 and Haiku 78.9%, so Sonnet does this step.
  4. Resolve. Names are merged by link domain or GitHub repo first, then by normalized name; Sonnet reviews near-duplicates ("Postgres" and "PostgreSQL").
  5. Score. One vote per HN user per product, their latest stance counts, warnings subtract, recent comments count more, and comments by the product's own makers are excluded.
  6. Synthesize. For products with at least 5 recommenders, Claude Opus groups quotes into praise and caveat themes and writes the two-sentence verdict. Any clause without a citation to a real comment is rejected.

What to keep in mind

Keyboard

j/k move through lists, o opens, / searches, ? shows everything.

Data

Comments come from the open-index/hacker-news dataset (ODC-BY license) and the HN Algolia API. Quotes are short excerpts, each linked to its comment on news.ycombinator.com. HackerProducts is an independent project, not affiliated with Y Combinator or Hacker News.

Everything is static. Download products.json, or append .json data from any product page.

Keyboard shortcuts
j / kNext / previous item
o or EnterOpen selected item
/Search
g hHome
g tTop
g rRising
g cCategories
tToggle dark mode
?This help