How Chuninja Works

The four steps

🎯

Pick a privacy mode

Private Web for everyday results from the live web, or PIR Vault for cryptographic retrieval over curated indexes.

🔍

Search the live web

Web, news, papers, code, images, Wikipedia, Hacker News, and arXiv all route through our SearXNG backend so the engines never see your IP.

🥷

Open through Ninja View

Click 🥷 Ninja View on any result. We fetch the page on our server, scrub the dangerous bits, and serve it to you sandboxed.

🔐

Switch to PIR Vault for curated indexes

For Wikipedia and other curated corpora, the document you fetch is retrieved via SimplePIR — the server can't tell which one you opened.

Architecture

Chuninja is a Rust (Axum) backend that fronts a self-hosted SearXNG instance and a headless Chromium renderer. The frontend is plain HTML/CSS/JS with no framework — easy to audit, easy to host.

Components

Axum server — handles search routing, PIR queries, and the Ninja View proxy reader
SearXNG — meta-search backend that fans out to upstream engines
Headless Chromium — stealth browser for rendering JS-required pages server-side
SimplePIR — cryptographic single-server PIR
Candle + BERT — sentence embeddings for the ranking index

What sees what

Component	Private Web	PIR Vault
Your browser → Chuninja	Query text in clear	Query text in clear (for ranking)
Chuninja → SearXNG / engines	Query text	Not used
Chuninja → PIR DB	Not used	Encrypted one-hot vector
Destination → you	Sees Chuninja's IP via Ninja View	Sees Chuninja's IP via Ninja View

Limitations

Chuninja is a frontend — it protects your IP from upstream engines and your reading from destination sites, but it does not anonymize your network traffic (use a VPN or Tor for that). And PIR only works on curated corpora, not on the live web.

Read more in our about page or PIR technology explainer.

HowChuninjaworks