We turn the firehose of dev-world signals into a feed worth reading.

Trend Snap is an independent editorial product for engineers, founders, and curious tech readers. Every day, thousands of links, repositories, threads, and videos surface across Hacker News, GitHub, and YouTube. Most of them are noise; a small fraction is genuinely worth your attention. Our job is to score the firehose, surface the substantive signal, and present it in a feed you can read in one coffee instead of one afternoon.

We focus on three topics — AI, Programming, and Tech — because those are the conversations our readers care about most. Each topic has its own dedicated feed, ordered by a transparent ranking that we describe openly below.

How our ranking works

Every item in our feed is scored by combining four observable signals:

• Trending score — a normalized engagement signal capturing community traction (upvotes, comments, stars, shares) on the originating platform.
• Quality score — an editorial / model-assessed signal that penalises clickbait, copy-paste content, and low-substance posts. We weigh quality higher than raw popularity, on purpose.
• Velocity score — how fast engagement is growing relative to the item's age. Velocity helps surface stories that are accelerating right now.
• Recency factor — exponential decay over hours since publish time, so old posts are de-prioritised even if they were once viral.

On top of those, we apply a per-source trust multiplier and a small set of guardrails that dampen items showing the late-trend pattern (very high popularity, very low velocity) and lift items showing the early-fast pattern. The result is a single ranked stream per topic — most relevant first, plenty of long-tail coverage as you scroll.

Where the content comes from

We aggregate links and metadata from public APIs and feeds: Hacker News (community, stories, comments counts), GitHub (trending repositories, stars, releases), and YouTube (creator uploads on tech and AI topics). We do not host the original articles, repositories, or videos — every card links back to the source so the original publisher gets the click and the credit.

Each card also includes an AI-generated summary, a “why it's trending” note, and (when available) a “why it matters” analysis written specifically for our readers. These editorial blurbs are ours, not the publisher's.

What we're not

We are not a social network, not an AI assistant, not a paid placement directory, and not a clone of any specific source. We do not republish full articles, we do not hide attribution, and we do not let advertisers influence ranking. Ads, when present, are clearly labelled as “Sponsored” and served by Google AdSense.

Get in touch

Tip line, takedown request, partnership idea, or feedback on the ranking? Use the contact page. We read everything.