Aetherium Browser Search
nomad-actual/aetherium-browser-search
Buscador privado com resumos IA, integrável ao navegador via SearXNG
Ver no GitHubMCP server que integra buscas web descentralizadas via SearXNG com resumos gerados por IA. Funciona como provedor de busca para navegadores, proxiando queries para SearXNG e enriquecendo resultados com análises automáticas de LLMs compatíveis com OpenAI.
Instalação
npx convez-ai nomad-actual/aetherium-browser-search1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Integrar buscas descentralizadas como motor padrão do navegador
- Gerar resumos automáticos de resultados de busca com LLMs locais
- Hospedar instância privada de busca web com IA na própria infraestrutura
Veredito

Solução bem arquitetada para quem quer controle total sobre buscas e resumos IA. Requer infraestrutura própria (SearXNG + LLM) mas oferece privacidade e flexibilidade genuínas, ideal para selfhosting.
README
A tradução para o português deste README ainda está na fila.
Aetherium Browser Search
A privacy-focused browser search bar service that proxies queries to SearXNG for decentralized web search results and enhances them with AI-generated overviews from any OpenAI-compatible LLM (llama.cpp, Ollama, vLLM, etc.).
Features
- Browser Search Bar Integration -- Configure any browser to use your service as a search engine via the OpenSearch / keyword URL format
- SearXNG Proxy -- Forward queries to a self-hosted or remote SearXNG instance, supporting categories (general, images, news, video, science, etc.)
- AI Overview -- Automatically generate a concise AI summary from top search results using any OpenAI-compatible LLM endpoint, streamed live via SSE
- SPA Architecture -- Client-side single-page application bundled with esbuild for fast, reactive rendering
- Switchable Themes -- Built-in gruvbox, tokyonight, and dark-aero themes with cookie-persisted user preference
- Collapsible Thinking Block -- View the LLM's reasoning process in a collapsible sidebar panel
- Docker Ready -- Multi-stage build with health checks, ready to deploy via
docker-compose
Architecture
Browser (search provider URL)
│
│ GET /search?q=<query>
▼
┌─────────────────────────────────┐
│ Fastify Service (TypeScript) │
│ - SearXNG proxy │
│ - LLM AI overview │
│ - SPA HTML shell │
└─────────┬───────────────┬───────┘
│ │
▼ ▼
┌──────────┐ ┌──────────────┐
│ SearXNG │ │ OpenAI-Like │
│ Instance │ │ LLM API │
└──────────┘ └──────────────┘
│ │
└───────┬───────┘
▼
Client SPA (esbuild bundle)
- Search results rendering
- Theme switcher
- SSE streaming client
Quick Start
Prerequisites
- Node.js 22+ (for local development)
- Docker & Docker Compose (for production)
- A running SearXNG instance
- An OpenAI-compatible LLM API (optional, e.g., llama.cpp, Ollama)
Docker Compose (Recommended)
- Copy and customize the environment file:
cp .env.example .env
- Edit
.envwith your SearXNG URL and optional LLM settings:
SEARXNG_URL=http://localhost:8080
LLM_API_URL=http://localhost:8081
LLM_MODEL=llama3.1-8b-instruct
- Start the service:
docker compose up -d
- Visit
http://localhost:3000/healthto verify.
Local Development
npm install
cp .env.example .env
# Edit .env with your configuration
npm run dev
Browser Configuration
Chrome / Edge / Brave
- Open Settings → Search engine → Manage search engines
- Under Site search, click Add
- Fill in:
- Search engine: Aetherium Search
- Keyword:
@aether - URL with %s in place of query:
https://your-domain.com/search?q=%s
- Now typing
@aether somethingin the address bar will search via your service.
Firefox
- Open Settings → Search
- Scroll to Search Shortcuts, click Add
- Fill in:
- Name: Aetherium Search
- Shortcut:
@aether - URL:
https://your-domain.com/search?q=%s
Safari
Safari uses your default search engine. Set your default to point to:
https://your-domain.com/search?q=
Configuration
All settings are environment variables (see .env.example):
| Variable | Required | Default | Description |
|---|---|---|---|
SEARXNG_URL | Yes | — | Base URL of your SearXNG instance |
SEARXNG_API_KEY | No | — | API key if SearXNG requires authentication |
LLM_API_URL | No | — | Base URL of the OpenAI-compatible LLM API |
LLM_API_KEY | No | — | API key for the LLM endpoint |
LLM_MODEL | No | llama3.1-8b-instruct | Model name to use |
LLM_MAX_TOKENS | No | 1024 | Max tokens for LLM response |
LLM_TEMPERATURE | No | 0.7 | LLM sampling temperature |
PORT | No | 3000 | Server port |
HOST | No | 0.0.0.0 | Server bind address |
AI_OVERVIEW_PROMPT | No | Built-in | Custom prompt template for AI overview |
AI Overview Prompt
Use {{query}} and {{results}} as placeholders:
AI_OVERVIEW_PROMPT=Synthesize the following search results into a concise answer for: {{query}}
Results:
{{results}}
API Reference
GET /search
Returns an HTML page with search results and AI overview.
Query Parameters:
| Param | Type | Description |
|---|---|---|
q | string | Search query (required) |
category | string | SearXNG category: general, images, news, video, science, it, music, files, maps |
engines | string | Comma-separated SearXNG engines |
language | string | Language code (e.g., en, en-US) |
pageno | number | Page number |
Example:
GET https://search.yourdomain.com/search?q=what+is+aetherium&category=general
GET /config
Returns current configuration (API keys masked).
Response:
{
"searxngUrl": "http://localhost:8080",
"llmApiUrl": "http://localhost:8081",
"llmModel": "llama3.1-8b-instruct",
"llmApiKey": null,
...
}
POST /config
Validate configuration changes (requires restart to apply).
GET /health
Health check endpoint.
Response:
{
"status": "ok",
"timestamp": "2025-01-15T12:00:00.000Z"
}
Project Structure
├── src/
│ ├── server/
│ │ ├── app.ts # Fastify server bootstrap
│ │ ├── config.ts # Environment schema, config loader, AppConfig
│ │ ├── routes.ts # Fastify route handlers
│ │ ├── searxng.ts # SearXNG URL builder, result formatting
│ │ └── llm.ts # OpenAI-compatible LLM call
│ ├── client/
│ │ ├── app.ts # Client SPA entry point
│ │ ├── search.ts # Search page rendering, SSE client
│ │ ├── ui.ts # Theme switching, escapeHtml
│ │ ├── markdown.ts # markdownToHtml (marked + DOMPurify)
│ │ ├── autocomplete.ts # Autocomplete dropdown
│ │ └── types.ts # Client-side SearXNGResult type
│ └── shared/
│ └── types.ts # Shared TypeScript interfaces
├── public/
│ ├── index.html # HTML shell
│ └── css/
│ ├── base.css # Base styles, layout, result styling
│ └── themes.css # Theme color palettes
├── docker-compose.yml
├── Dockerfile
├── tsconfig.server.json
├── tsconfig.client.json
└── .env.example
License
ISC