How it works

Hacker Approved lists the books, gear, software and services that people recommend in Hacker News comments, ranked by how many separate people recommend each one.

Where the numbers come from

We read 1,795,640 Hacker News comments from November 2025 to November 2026. A language model (Claude Haiku) marked every product each comment recommends or criticizes, with the words that say so. 652,515 comments carried a product opinion.

For each product we count:

• Recommend: separate people who recommended it. One enthusiastic person counts once.
• Criticize: separate people who warned against it or regretted it.
• Months recommended: how many of the 13 months it was recommended in, which separates lasting favorites from a week of news.
• Years owned: when owners say how long they have used it, the median of what they say.

Lists are sorted by the number of people who recommend each product. Summaries and reasons are written by Claude from the comments, and every reason links to the comments it rests on.

The badge

A product is Hacker Approved when at least 10 different people recommend it, recommenders outnumber critics at least 2 to 1, and it was recommended in at least 4 different months. It is recalculated whenever the data refreshes. It cannot be bought, requested or traded for anything. 2,909 products meet the rule today.

How we make money

Some buy buttons are affiliate links: if you buy through them, the store pays us a commission at no cost to you. As an Amazon Associate we earn from qualifying purchases. Commissions never change a ranking, a summary or the badge, and free products are listed alongside paid ones.

The source

The recommendations come from public comments on Hacker News. Hacker Approved is independent and not affiliated with Y Combinator or Hacker News. We quote short passages and link every quote to its comment; we don't show usernames.

Remove your quotes

If you wrote a comment we quote and want it removed, email optout@hackerapproved.com with the comment's link. We remove it at the next refresh.

For AI agents and developers

Every product page has a markdown version: add .md to its address. Start from llms.txt.

AI assistants can connect to our MCP server at https://hackerapproved.com/mcp (Streamable HTTP, no sign-in) to search products, read one, compare two or list the best of a type. The same data is plain JSON: /api/search?q= for search, /api/products/<slug>.json for one product and /api/catalog.json for the whole list.