What It Does
- Fetches live front-page stories from the Algolia Hacker News API and delivers them as a concise spoken digest
- Drills into any story on request for a fuller breakdown
- Searches Hacker News for any topic by name
- Pages through the front page for more stories
- Uses the LLM only for narration — titles, points, and comment counts come directly from the API
Voice Commands
Example Conversation
User: What’s on Hacker News? AI: A new memory allocator for Rust is getting attention — it cuts allocation overhead by forty percent in benchmarks, and the thread is full of performance engineers comparing notes. Up next, PostgreSQL 17 shipped… AI: Want me to explain one of those? Say a number, name a topic, or say stop. User: Two. AI: PostgreSQL 17’s headline feature is a revamped vacuum process… User: Search for WebAssembly. AI: Five stories on WebAssembly right now — the top one covers the new WASI 0.2 spec that just hit stable… User: Stop. AI: Later. Stay curious.
How It Works
- Low-latency startup — the API fetch and the spoken intro run concurrently, so the stories are ready by the time the intro finishes.
- Background prefetch — once the digest is spoken, all five story expansions are pre-generated in the background, so asking for a story by number is usually instant.
- Intent classification — each response is routed through an LLM classifier (
EXPAND,SEARCH,MORE,EXIT,CONVERSE,UNKNOWN); bare numbers bypass the LLM via fast-path detection. - Pagination — “more” fetches the next Algolia page, so stories never repeat across batches.
Trigger Phrases
Hacker NewsHN digestwhat's on Hacker Newstech digestHN today
Setup
No API key required — the ability uses the free, public Algolia Hacker News Search API, which needs no sign-up or token. It requires Python 3.10+, thehttpx package, and an OpenHome Agent with an LLM configured (used for narration and intent classification).
APIs Used
Developer Credit
Developed by Kuber Mehta.View on GitHub
Source code for the Hacker News Digest community ability.

