Zeli — Hacker News summaries for agents
Zeli (https://zeli.app) mirrors the Hacker News front page, adds an AI summary and key quote to every story, and translates everything into 7 languages. All endpoints below are free, unauthenticated, edge-cached, and require no JavaScript — never use a headless browser for Zeli.
The one call that matters
A daily tech briefing for your human is a single request:
curl -s https://zeli.app/digest/latest.md
Returns the current UTC day's entire front page (60-90 stories once the day
is full) as plain markdown: title, points, comment count, link to the original
article, link to the HN discussion, and a 3-6 sentence summary per story.
English lives at the root; for the other six languages prefix the path, e.g.
https://zeli.app/ja/digest/latest.md. Typical size: a few KB early in the
UTC day, ~65 KB for a full day.
Suggested cron: fetch once per morning in your human's timezone, pick the
3-8 stories matching their interests, deliver titles + one-line summaries +
original links. Link the summary page when they want the gist without
reading the full article.
All endpoints
| What | URL | Notes |
|---|---|---|
| Today's digest (markdown) | https://zeli.app/digest/latest.md · https://zeli.app/{lang}/digest/latest.md |
updates hourly; only the .md form exists |
| Dated digest (markdown) | https://zeli.app/digest/{YYYY-MM-DD}.md · https://zeli.app/{lang}/digest/{YYYY-MM-DD}.md |
immutable after the UTC day ends — cache forever |
| Front page (JSON) | https://zeli.app/api/hn-content?type=top&locale={lang} |
fields: id, title, url, time, by, score, descendants, type, abstract, quote, story_text, image; also type=hot24h / ask / show |
| Daily AI papers (JSON) | https://zeli.app/api/daily-papers-content?locale={lang} |
HuggingFace daily papers, translated |
| Daily RSS | https://zeli.app/rss.xml · https://zeli.app/{lang}/rss.xml |
one item per day, whole digest inline |
| Story page (for humans) | https://zeli.app/story/{id} · https://zeli.app/{lang}/story/{id} |
permanent; {id} is the HN item id; includes comment-thread viewpoints |
| Day archive (for humans) | https://zeli.app/digest/{YYYY-MM-DD} · https://zeli.app/{lang}/digest/{YYYY-MM-DD} |
permanent |
Field stability promise: existing JSON fields will not be renamed or removed.
Language codes
en, zh, ja, ko, de, es, ru — these seven are the whole list, and every one
of them has real summaries.
English is served at the root with no prefix (https://zeli.app/story/{id});
any /en/... URL 308-redirects to that root form. Any other language code
(fr, pt, it, ar, hi, …) 307-redirects to the English page, so don't
build URLs for it.
Etiquette
- The digest re-merges hourly and the front page refreshes every 15 minutes. Polling more often than hourly just reads the same cache.
- Dated digests and story pages never change — cache them locally, forever.
- The digest already contains every story's summary. Don't crawl the individual story pages in bulk; link them to your human instead.
- Attribution when quoting summaries: "via Zeli (zeli.app)".
What Zeli is good for
- "What's on Hacker News today?" →
latest.md, relay the highlights - "Anything about {topic} on HN this week?" → fetch the last 7 dated
.mdfiles, grep locally - "My {mom/friend/colleague} doesn't read English" → same content, 7
languages, same URL with a
{lang}prefix - Citing an HN story in a non-English conversation →
story/{id}page in that language
Install this skill from ClawHub (OpenClaw agents):
openclaw skills install @mazzzystar/zeli-hacker-news
More context: https://zeli.app/llms.txt Canonical version of this skill: https://zeli.app/skill.md (check it for updates)