---

Type a word and you get two things at once: a month-by-month trend line of how often Hacker News talked about it, and the real posts and comments behind every spike. Search one term, or stack several on the same chart to see which one the internet actually cared about.

The whole thing is a thin UI over Upstash Redis Search. There's no separate search engine to run - you store plain Redis hashes and define one index over them, and every search and every trend line is a single @upstash/redis call.

What you can do

• 🔎 Search ~45M Hacker News posts and comments, full-text, in milliseconds.
• 📈 See the trend - mentions per month for any term, so you know exactly when it took off.
• 🆚 Compare several terms on one chart and watch them rise and fall against each other.
• 🧵 Drill in to the actual stories and comments behind each spike - who wrote them, and how the conversation splits between posts and replies.

How it works

1. Ingest. scripts/ingest.ts streams monthly Hacker News Parquet files from HuggingFace for backfills, and a daily GitHub Action pulls new items from the official HN API (--live). Each item is HSET as a plain hn:<id> hash in Upstash Redis.

2. Index. Before loading, the same script defines one search index over those hashes:

await redis.search.createIndex({
  name: "hn",
  dataType: "hash",
  prefix: "hn:",
  schema: s.object({
    title: s.string(),
    text: s.string(),
    by: s.keyword(),
    type: s.keyword(),
    time: s.date().fast(),
    score: s.number("F64"),
    ndesc: s.number("F64"),
    parent: s.number("F64"),
  }),
});

3. Query. Every search the UI runs is one hn.query({ filter, ... }), and every trend line is one hn.aggregate({ filter, aggregations }). Both are built from the same options in src/lib/hn-query.ts.

const hn = redis.search.index({ name: "hn", schema });

// matching posts, newest first
const { documents } = await hn.query({
  filter: { title: { $eq: "rust" }, time: { $gte: "2024-01-01" } },
  orderBy: { time: "DESC" },
  limit: 30,
});

// mentions per month
const { aggregations } = await hn.aggregate({
  filter: { title: { $eq: "rust" } },
  aggregations: {
    by_month: { $dateHistogram: { field: "time", fixedInterval: "30d" } },
  },
});

Run it locally

Add your Upstash Redis REST credentials to .env.local:

UPSTASH_REDIS_REST_URL="..."
UPSTASH_REDIS_REST_TOKEN="..."

Then install, load some data, and start the dev server:

bun install

# ingest one quarter to play with (creates the index on first run)
bun scripts/ingest.ts 2026 Q1

# build the "Who is hiring?" postings index for the same months
# (/who-is-hiring reads it by default; set NEXT_PUBLIC_JOBS_INDEX_READY=0 to skip)
bun scripts/ingest-jobs.ts 2026-01 2026-03

bun dev

Open localhost:3000 and start searching.

Tech stack

• Upstash Redis Search via @upstash/redis - the entire search + analytics backend.
• Next.js (App Router) + React, deployed on Vercel.
• Hacker News data from the monthly Parquet dumps on HuggingFace, kept current from the official HN Firebase API.

License

MIT © Upstash