Pretty Hacker News — Privacy Policy

Last updated: June 2026

Pretty Hacker News is a browser extension that renders a calmer reader for news.ycombinator.com. It is local-first: it has no servers of its own, no analytics, and no tracking. This page lists every network request it makes and exactly what triggers each one, so you can verify the claims against your browser's network tab.

What we collect

Nothing. Pretty Hacker News operates no backend and runs no analytics. We do not collect, store, or transmit your personal data or browsing activity to any server we control — because we don't have one.

Where your data lives

Your preferences and state — theme, read/seen stories, bookmarks, muted domains, default feed, and (if you use AI) your API key, profile text, interest tags, and local token-usage counts and estimated daily spend — are stored only in your browser via chrome.storage.local. They never leave your device except as described below.

Network access — the complete list

• news.ycombinator.com — to read the feed/comments and, when you choose to, vote or comment using your own existing HN login. Pretty Hacker News uses the cookie already in your browser (exactly as clicking the real arrows does); it never sees, stores, or transmits your HN password or credentials.
• hacker-news.firebaseio.com and hn.algolia.com — Hacker News's own official, public data APIs (run by Y Combinator and its partner). Anonymous reads only, no cookies, no personal data — used to fetch feed rankings and comment threads as structured data.
• No third-party font or icon hosts. The reader's font (Inter) is bundled inside the extension, and site icons are drawn locally as coloured letter tiles — so Pretty Hacker News contacts no Google (or other) font or favicon service at all.
• Article previews (optional, off by default) — only if you turn this on, and only when you click “Preview,” Pretty Hacker News fetches that one linked page to show its text inline. Nothing is fetched otherwise, and the result is not stored or sent anywhere.
• AI features (optional, off by default — only if you want them) — see below.

AI is opt-in

Pretty Hacker News makes no AI calls at all unless you choose to enable AI and provide your own API key. There is no AI by default.

• When (and only when) you click AI brief or Suggest tags, Pretty Hacker News sends a request directly from your browser to the provider you chose — api.anthropic.com or api.openai.com — using your key. The request passes through no server of ours.
• That request contains the post, the linked article's text, and the comments (for a brief) or your profile text (for tag suggestions), so it can be summarized or matched to your interests. Your key is stored only in your browser and is never sent anywhere except to your chosen provider.
• AI is used only to help you read — summarizing threads and matching stories to your interests. Pretty Hacker News does not use AI to generate, draft, or post comments, replies, or submissions, or to vote, on your behalf. (This is a standing commitment of the project — see the LICENSE.)

Permissions

• news.ycombinator.com access — to render the reader and act on your behalf when you vote/comment.
• api.anthropic.com / api.openai.com access — used only when you turn on AI and trigger it, to call the provider you chose with your key.
• storage — to save your preferences and (if you use AI) your key locally.
• declarativeNetRequest — to relax HN's content-security-policy on the pages Pretty Hacker News renders so its bundled font and article preview images can load. No requests are blocked or redirected. (It also strips the Origin header on Pretty Hacker News's own calls to api.anthropic.com so keys on zero-data-retention orgs work; this is scoped to Pretty Hacker News's own requests and never affects other sites.)
• Broad site access (optional) — requested only if you turn on Article previews; used solely to fetch a page you click Preview on.

Contact

Questions? Open an issue on the project's repository.

Not affiliated with Y Combinator.