0xSearchstr

Decentralized search aggregator. Nostr first, web when needed. No backend required.

Live: https://0xSearchstr.shakespeare.wtf · nsite mirror: https://0xsearchstr.shakespeare.to

Nostr: npub1z2k4ttglmwgc75c5e856tngnt05mw3hxams4lkr3muf354nh6xvskk2ew6

Sister app: 0xPresearchstr — same engine, shared index (see Federation)

How It Works

User Search
       │
       ▼
 ┌─────────────── All providers run in parallel ──────────────┐
 │                                                             │
 │  Nostr (NIP-50)  SearXNG   Wikipedia   Hacker News   Tor   │
 │       │              │          │           │          │    │
 │  Web Index     DuckDuckGo  Stack Overflow   Community      │
 │  Cache Index        │          │              Index         │
 │       ▼              ▼          ▼              ▼             │
 │   SearchResult[] from each provider                         │
 │                                                             │
 └──────────────────────┬──────────────────────────────────────┘
                        │
              Instant Answer? (calculator · npub · Wikipedia)
                        │
                   Merge + Deduplicate + Rank
                        │
                        ▼
                   Display Results
                        │
                   Still nothing?
                        │
                        ▼
              Trending cached queries + Browser Fallback Links

Instead of building another centralized search engine, 0xSearchstr is a search aggregator with a plugin-based provider architecture:

Every source is a provider — each returns a universal SearchResult[]
All providers run in parallel — results stream in as each completes
Nostr scores highest — decentralized results are prioritized
Auto-indexing — every search publishes discovered pages back to Nostr as document observations (never the query)
Community-curated — any Nostr user can submit links to the shared index
Never leaves you empty — trending cached queries + fallback links to privacy-respecting search engines

Everything runs in the browser. No backend, no crawler, no tracking.

Features

🔦 Honest Privacy (traffic-light indicator)

Every provider is classified by who can see your query, and a traffic-light indicator sits next to the search bar at all times:

Light	Tier	Providers	Who sees the query
🟢	Nostr	Cache Index, Nostr, Community	Relay operators only (query + IP, no account linked)
🟡	Direct	Wikipedia, Hacker News, Stack Overflow	The API operator (query + IP in standard logs)
🔴	Proxied	SearXNG, DuckDuckGo, Ahmia	A CORS proxy and the destination service

Privacy Mode (Settings) — one switch to run Nostr-tier providers only. Zero third-party exposure, at the cost of fewer results.
The full, honest threat model lives on the About page. "No backend" ≠ "no one sees anything" — we document exactly who does.

⚡ Instant Answers

Direct answers above the result list, no waiting:

Calculator — 2^10, (3+4)*5, 15% of 80 — safe recursive-descent parser, computed locally, never sent anywhere
Nostr profiles — paste a bare npub1… / nprofile1… → rich profile card (avatar, banner, NIP-05, bio)
Wikipedia summaries — strong title matches render the article's first paragraph + thumbnail (auto-disabled in Privacy Mode)

🗂 Community Index (curated by Nostr users)

The index isn't just a bot cache — any logged-in Nostr user can submit links (kind 30078, signed with their own key) via the Submit button. Submissions support https://, magnet:, ipfs://, and .onion links with automatic content-type detection (Torrent, IPFS, Video, Audio, PDF badges).

URL allowlist blocks javascript:/data: at parse time; onion submissions render behind a Tor warning interstitial
Independent project, credit where due: 0xSearchstr was built on its own, but the idea of letting every user curate the index — not just bots and crawlers — was adopted after we discovered Nostra Search, a project exploring the same territory. Thanks for the spark. Our implementation is our own, with an improved schema (unique per-URL d-tags fix their one-entry-per-author limitation)
Nostra read-interop: we also read Nostra Search's own index (nostra:index), including their AES-GCM obfuscated payloads — their community's curation shows up here, attributed as "Nostra Index"

🌐 Federation: One Index, Many Clients

The Search Index Protocol (kind 39697, canonical spec: SIP-01/SIP-01) is shared with 0xPresearchstr, the UNCAGED Index Relay, the Crawlstr + indexstr crawler network, and open to any fork or third-party engine. Every browser is its own indexer — there is no central signing key:

Per-device identity — each browser generates its own indexer keypair on first use (Settings → Indexing). Pseudonymous, replaceable, exportable.
Independent observations — when indexer A and indexer B both see example.com, they publish separate events with the same d tag. Search nodes count distinct indexers per document — "7 independent indexers saw this page."
Any relay works — plain Nostr filters (#d, #t) on ordinary relays; NIP-50 is an optional acceleration, never a requirement.

The legacy query cache (kind 30078, 0xsearchstr:cache:*) remains federated with 0xPresearchstr during migration — readers trust both bot pubkeys, and old clients keep a warm cache. Full legacy spec in NIP.md.

🔍 Explore the Index

/explore turns the index into discoverable content: recently indexed pages (with independent-indexer counts), trending terms, staked keywords, and aggregate stats. Every search becomes content.

💎 Keyword Staking (Presearch-style, Nostr-native)

Stake your identity on a keyword instead of tokens: sign an addressable kind 30078 event (0xsearchstr:stake:<keyword>) binding a keyword to your link. Anyone searching that keyword — on any compatible client — sees your link as the top "Community Stake" placement. One stake per keyword per npub; re-staking replaces it. No tokens, no auction.

👍 Voting & Reporting

Every result card carries 👍/👎 votes and a report flag:

Votes are NIP-25 reactions (kind 7; e tag for events, r tag with the SIP-01-normalized URL for pages). Anonymous by default — signed by this device's indexing identity, never your npub. Settings → Indexing has an opt-in "vote with my npub" toggle.
Reports are NIP-56 kind 1984 events (0xsearchstr.abuse namespace), landing in the team's /admin inbox.

📈 Trending Terms (k-anonymity)

Trending searches with no plaintext queries anywhere: each device publishes only sha256(normalized query) under its indexing identity (0xsearchstr:term:*). A term's plaintext is revealed by the network only once 3+ independent devices searched it — the reveal is self-verifying (readers re-hash and compare). Rare or sensitive queries (NIP-19 ids, NIP-05 addresses, URLs, math) are never signaled at all.

🛡 Team Console (`/admin`)

A hidden, role-gated dashboard (not linked in nav — team members see "Admin console" in their account menu when logged in with a team key):

Stats — indexed pages, stakes, open reports, relay pool sizes
Reports — the NIP-56 abuse inbox, one-click "hide from results"
Moderation — team-signed NIP-32 labels (kind 1985, 0xsearchstr.moderation) hide URLs/event ids from every user's results; un-hiding publishes a NIP-09 deletion. Clients trust labels from the owner + role-list pubkeys only
Roles (owner) — add/remove admins & moderators by npub (owner-signed 0xsearchstr:admin-roles / 0xsearchstr:mod-roles addressable events)
Filter test — check whether a URL or event id is currently filtered

🤖 AI Answers (optional, off by default)

An AI answer layer sits on top of the federation: the top results become a numbered evidence pack, the model answers only from it, with clickable [n] citations. Ephemeral — never indexed into SIP-01. Tiers: your own key (BYOK — PPQ.ai, OpenRouter, Ollama, any OpenAI-compatible endpoint) → a built-in free tier (shared rate-limited PPQ key, public by design). Privacy boundaries hold: NIP-19/05, URLs, and math keep their deterministic paths; Nostr results are excluded from evidence unless you opt in.

⌨️ Quality of Life

OpenSearch — add 0xSearchstr as a browser search engine (/opensearch.xml)
Keyboard shortcuts — Ctrl+K / Cmd+K or / focuses the search bar from anywhere
Shareable searches — ?q=query&source=web URL params
Smart snippets — Nostr notes show the relevant window around your query terms (Google-style), not just the note head; hashtag/link-stuffed spam is filtered out
Graceful degradation — a failing provider shows a muted "skipped" state instead of a red error when others delivered

Auto-Indexing (Shared Web Index)

The killer feature: every search grows the index.

When you search, useful web results are published to Nostr as document observations — one lightweight, addressable event per URL (kind 39697, SIP-01), signed by this browser's own dedicated indexing identity. No account, no login, no central signing key — and your search query is never published, only the pages' public metadata.

Search "best monero wallet"
       │
       ├─→ Read the shared web index (kind 39697) + legacy cache (kind 30078)
       │     └─→ HIT? → instant results from Nostr
       │
       ├─→ Run all providers in parallel
       │     └─→ Merge + deduplicate + rank
       │
       └─→ Index the useful pages (auto-index)
             ├─→ kind 39697 per URL, signed by your device's indexer key
             └─→ every compatible client can now find those pages

Indexing identity: each browser generates its own keypair on first use (Settings → Indexing). It is pseudonymous, replaceable, exportable, and never linked to your personal Nostr identity. Key separation is guaranteed; network anonymity is not (relays see IP/timing — be honest about that).

The more people use any compatible client, the smarter every client gets. No crawler. No database. Just Nostr.

Migration note: the legacy query→results cache (kind 30078) is frozen legacy data (SIP-01 §17) — 0xSearchstr still reads it for backward compatibility with 0xPresearchstr and older clients, but no longer writes it. The only signer in this codebase is the per-device indexing identity.

Quick Start

git clone https://github.com/NostrDanish/0xSearchstr.git
cd 0xSearchstr
npm install
npm run dev

Open http://localhost:8080 and search.

Provider Architecture

src/lib/providers/
├── types.ts          ← SearchResult, SearchProvider interface (privacy tiers!)
├── web-index.ts      ← Shared web index — SIP-01 kind 39697 observations (reads first)
├── cached-index.ts   ← Legacy federated query cache (kind 30078, read for compatibility)
├── nostr.ts          ← NIP-50 relay search
├── stakes.ts         ← Keyword stakes (Presearch-style top placements)
├── community.ts      ← Community-curated index (+ Nostra Search interop + NIP-B0)
├── git.ts            ← NIP-34 repos/issues/PRs (ngit/GRASP pool, read-only)
├── wiki.ts           ← NIP-54 wiki articles (wikifreedia pool, read-only)
├── brave.ts          ← Brave Search API (BYOK, off by default)
├── searxng.ts        ← SearXNG meta-search with failover
├── duckduckgo.ts     ← DuckDuckGo HTML scraper
├── wikipedia.ts      ← MediaWiki API
├── hacker-news.ts    ← Algolia HN Search API
├── stackoverflow.ts  ← StackExchange API
├── tor.ts            ← Ahmia.fi .onion search
├── registry.ts       ← Provider catalog (+ per-engine enable/disable)
└── index.ts          ← Barrel export

src/lib/
├── webIndex.ts        ← SIP-01 protocol: URL normalization, build/parse/verify
├── indexerIdentity.ts ← Per-device anonymous indexer keypair
├── keywordStakes.ts   ← Stake schema (0xsearchstr:stake:*)
├── termSignals.ts     ← k-anonymity trending (0xsearchstr:term:*)
├── votes.ts           ← NIP-25 result votes (device identity by default)
├── reports.ts         ← NIP-56 abuse reports (0xsearchstr.abuse)
├── moderation.ts      ← NIP-32 team labels (0xsearchstr.moderation) + roles
├── queryClassify.ts   ← Query understanding (URL/nip19/nip05/math detection)
├── resultRank.ts      ← Coverage-ranked result merging
└── ai/                ← Optional AI answer layer (BYOK + built-in free tier)

Adding a Provider

Create src/lib/providers/my-provider.ts implementing SearchProvider
Import it in registry.ts and add to ALL_PROVIDERS
Done — the orchestrator picks it up automatically

SearchProvider Interface

interface SearchProvider {
  id: string;
  name: string;
  source: SearchSource;
  /** Extra tabs this provider also runs under (e.g. community runs under 'tor' too) */
  additionalSources?: SearchSource[];
  /** Privacy tier — who can observe the query. Drives the traffic-light + Privacy Mode. */
  privacy: 'nostr' | 'direct' | 'proxied';
  /** Honest one-liner about who sees the query (shown in the privacy popover). */
  privacyNote: string;
  search(options: SearchOptions): Promise<ProviderSearchResponse>;
}

Live Providers

Provider	Source	API	Privacy	Notes
Web Index	Nostr kind 39697	WebSocket	🟢 Nostr	Shared document observations from all indexers (SIP-01), integrity-verified (`d`↔`u`, `x`↔content)
Cache Index	Federated Nostr index	WebSocket	🟢 Nostr	Legacy query cache, read-only (frozen) — off by default
Nostr	NIP-50 relays	WebSocket	🟢 Nostr	UNCAGED + community default relays, fully user-editable
Keyword Stakes	Nostr kind 30078	WebSocket	🟢 Nostr	Community-staked keyword placements
Community	Nostr kind 30078 + 39701	WebSocket	🟢 Nostr	User-submitted links + NIP-B0 bookmarks + Nostra Search index
Git Repos	ngit/GRASP relays (NIP-34)	WebSocket, read-only	🟢 Nostr	Repos, issues, PRs & patches — pool editable in Settings
Nostr Wiki	Wiki relays (NIP-54)	WebSocket, read-only	🟢 Nostr	Wikifreedia corpus — pool editable in Settings
SearXNG	Dynamic instance pool	CORS proxy	🔴 Proxied	DDG, Brave, Wikipedia, and dozens more
DuckDuckGo	HTML scraper	CORS proxy	🔴 Proxied	Direct DDG fallback when SearXNG is slow
Brave	Brave Search API	CORS proxy	🔴 Proxied	Off by default — BYOK in Settings → Engines
Wikipedia	MediaWiki API	Direct (CORS)	🟡 Direct	Off by default — enable in Settings → Engines
Hacker News	Algolia API	Direct (CORS)	🟡 Direct	Stories with points/comments
Stack Overflow	StackExchange API	Direct (CORS)	🟡 Direct	Off by default — enable in Settings → Engines
Tor (Ahmia)	HTML scraping	CORS proxy	🔴 Proxied	Off by default — policy-compliant .onion search

Every engine can be toggled off in Settings → Search Engines — off engines never run and never see your query. The tab bar is modular too: Settings → Search Tabs picks visible tabs, order, and the default landing tab.

Dynamic SearXNG Instance Pool (searxist-style)

Instead of a hardcoded instance list, the SearXNG provider uses a self-healing dynamic pool (inspired by searxist):

┌── Tier 1: Custom ──────┐   Your self-hosted / trusted instances (always first)
├── Tier 2: Discovered ──┤   Live from searx.space, privacy-filtered:
│                        │     • no analytics  • clearnet  • ≥80% search success
└── Tier 3: Seeds ───────┘   Hardcoded bootstrap fallback

Auto-discovery — the pool refreshes from searx.space every 24h, client-side
Quality-aware health tracking — per-instance success/failure/latency and result-count stats (EMA) in localStorage; failing or thin instances sink, fast and complete ones rise
Self-hosting friendly — add your own instance in Settings and it runs first on every search
Zero backend — discovery, health, and ranking all happen in the browser

Relay Management

Two layers, both manageable at /settings:

Your Relays (NIP-65) — your personal relay list with read/write flags; publishes kind 10002 when logged in. Defaults to the 0xSearchstr app relays for new users.
Search Relays (index pool) — the pool that powers Nostr search, the shared web index, and index publishing. Defaults: the UNCAGED index relay (relay-na1.metanomalist.com), community NIP-50 relays (ditto, jskitty.cat, nos.today, nostr.band, noswhere), Primal + Hifish for replication, and a Tor index relay (.onion). Every default is removable — removed defaults can be restored, and customs (including your own self-hosted NIP-50 relay) can be added. A built-in latency tester checks reachability and round-trip times. The .onion relay is only contacted when the app itself runs on a .onion origin — on clearnet it's skipped silently (a connection could never succeed, and on devices with a local Tor resolver the attempt trips Chrome/Brave's local-network permission prompt).

Incremental Results

All providers run in parallel. The UI shows live status:

✔ Nostr (124ms)  ✔ Wikipedia (230ms)  ⏳ SearXNG...  ⏳ HN...

Results appear as each provider finishes — no waiting for the slowest one.

Search Tabs

The tab bar is fully modular — Settings → Search Tabs lets every user pick which tabs show, reorder them, and star the tab a fresh visit starts on. Deep links (/?source=tor&q=…) keep working even for hidden tabs.

Tab	Sources
Web	Web Index (SIP-01) + Stakes + Community + SearXNG + DuckDuckGo + Brave (BYOK)
Index	The community index only — SIP-01 observations + legacy cache
All	All providers merged + ranked (stakes on top)
Nostr	Profiles, notes, articles, Wikifreedia, files, torrents, code snippets
News	Hacker News stories
Wiki	Nostr wiki articles (NIP-54 pool); Wikipedia engine off until enabled
Code	Git repos/issues/PRs (NIP-34 via ngit/GRASP) + NIP-C0 snippets; Stack Overflow off until enabled
Tor	.onion hidden services via Ahmia + index-observed onion pages
I2P	Eepsite directory links + index-observed eepsites

Self-Hosted Backend (Optional)

The backend/ directory contains a full self-hosted stack for when you want to run your own search infrastructure:

Service	Description
Meilisearch	Full-text search index engine
Nostr Crawler	NIP-01 subscriber indexing kinds 0/1/30023/1063
Clearnet Crawler	Polite web crawler (robots.txt, rate-limited)
Tor/I2P Crawler	Hidden service crawler with content policy enforcement
NIP-50 Relay	Search relay proxy bridging Meilisearch to Nostr
Abuse API	REST search API + abuse report management

cp .env.example .env   # Edit MEILI_API_KEY + ABUSE_ADMIN_TOKEN
docker compose up -d

See the backend README and Content Policy for details.

Protocol Spec

The interoperable web-document index is specified by SIP-01 — canonical spec at github.com/SIP-01/SIP-01 (kind 39697) — designed so an independent developer can implement a compatible indexer, search node, or search engine without reading this codebase. 0xSearchstr's implementation is byte-compatible with the spec's §13 test vectors (see docs/SEARCH_INDEX_PROTOCOL.md).

App-specific legacy schemas — the federated query cache, community submissions, trusted indexer list, and Nostra Search interop — are documented in NIP.md.

Content Policy

The self-hosted backend enforces content policy modeled on Ahmia. Hard-blocked categories: CSAM, human trafficking, weapons sales, drug marketplace listings. See the Policy page for details.

Tech Stack

React 19 + TypeScript + Vite
TailwindCSS 4 + shadcn/ui
Nostrify — NIP-50 relay search
SearXNG — meta-search fallback
Wikipedia — MediaWiki API
Hacker News — Algolia search
TanStack Query — data fetching + caching

License

MIT

Vibed with Shakespeare

Name		Name	Last commit message	Last commit date
.agents/skills		.agents/skills
.github/workflows		.github/workflows
.nsite		.nsite
.vscode		.vscode
backend		backend
docs		docs
eslint-rules		eslint-rules
public		public
src		src
.env.example		.env.example
.gitignore		.gitignore
.gitlab-ci.yml		.gitlab-ci.yml
.mcp.json		.mcp.json
AGENTS.md		AGENTS.md
CONTRIBUTING.md		CONTRIBUTING.md
NIP.md		NIP.md
README.md		README.md
SECURITY.md		SECURITY.md
components.json		components.json
docker-compose.yml		docker-compose.yml
eslint.config.js		eslint.config.js
index.html		index.html
opencode.json		opencode.json
package-lock.json		package-lock.json
package.json		package.json
tsconfig.json		tsconfig.json
vite.config.ts		vite.config.ts

Navigation Menu

Repository navigation

About

Repository files navigation

0xSearchstr

How It Works

Features

🔦 Honest Privacy (traffic-light indicator)

⚡ Instant Answers

🗂 Community Index (curated by Nostr users)

🌐 Federation: One Index, Many Clients

🔍 Explore the Index

💎 Keyword Staking (Presearch-style, Nostr-native)

👍 Voting & Reporting

📈 Trending Terms (k-anonymity)

🛡 Team Console (`/admin`)

🤖 AI Answers (optional, off by default)

⌨️ Quality of Life

Auto-Indexing (Shared Web Index)

Quick Start

Provider Architecture

Adding a Provider

SearchProvider Interface

Live Providers

Dynamic SearXNG Instance Pool (searxist-style)

Relay Management

Incremental Results

Search Tabs

Self-Hosted Backend (Optional)

Protocol Spec

Content Policy

Tech Stack

License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

0xSearchstr

How It Works

Features

🔦 Honest Privacy (traffic-light indicator)

⚡ Instant Answers

🗂 Community Index (curated by Nostr users)

🌐 Federation: One Index, Many Clients

🔍 Explore the Index

💎 Keyword Staking (Presearch-style, Nostr-native)

👍 Voting & Reporting

📈 Trending Terms (k-anonymity)

🛡 Team Console (/admin)

🤖 AI Answers (optional, off by default)

⌨️ Quality of Life

Auto-Indexing (Shared Web Index)

Quick Start

Provider Architecture

Adding a Provider

SearchProvider Interface

Live Providers

Dynamic SearXNG Instance Pool (searxist-style)

Relay Management

Incremental Results

Search Tabs

Self-Hosted Backend (Optional)

Protocol Spec

Content Policy

Tech Stack

License

About

Resources

Contributing

Security policy

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

🛡 Team Console (`/admin`)