SearXNG is a privacy-first metasearch engine that pulls results from 250+ search engines including Google, Bing, DuckDuckGo, Qwant, and Yahoo — without collecting or profiling user data. It’s designed for users who want to escape surveillance capitalism in search, whether through public instances or self-hosted deployments.
Built in Python with a modular architecture, SearXNG supports containerized deployment (Docker), uWSGI, Granian, and NGINX/Apache backends. It includes an administration API, plugin system, and CI/CD pipelines for professional development. The project is community-driven, with 58 language translations and over 70 public instances listed on searx.space.
What You Get
- 250+ Search Engines - Aggregates results from Google, Bing, DuckDuckGo, Qwant, Yahoo, Brave, and other search services in a single interface.
- Zero User Tracking - No cookies, no IP logging, no profiling — search queries are not stored or linked to users.
- Self-Hosted Option - Full control over your search instance with documented installation guides for Docker, uWSGI, Granian, NGINX, and Apache.
- Administration API - Programmatically manage search engines, filters, and settings via a documented REST-like API.
- 58 Language Translations - Fully localized interface with community-driven translations via Weblate.
- Tor Support - Can be accessed over the Tor network for enhanced anonymity and censorship resistance.
- Plugin System - Extend functionality with custom answerers, search engines, and result processors via modular plugins.
- CI/CD & Automated Testing - Professional development practices including automated UI tests, quality assurance, and continuous integration.
Common Use Cases
- Running a privacy-focused news portal - A journalist uses SearXNG to provide readers with unbiased, non-tracking search results embedded in their site.
- Deploying a corporate search gateway - An IT team hosts SearXNG internally to replace Google Search in the office, blocking tracking while maintaining access to all major engines.
- Building a Tor-based anonymous search node - An activist sets up a SearXNG instance accessible via Tor to help users in censored regions find information safely.
- Creating a decentralized search alternative for a community - A local library or university deploys SearXNG to offer students a private, ad-free search experience without vendor lock-in.
Under The Hood
Architecture
- Flask-based MVC structure with clear separation between routing, business logic, and templating layers
- Plugin system enabling dynamic engine registration via configuration, allowing external search engines to be added without core modifications
- Modular asset and theme management supporting theme swapping through directory-based isolation
- Dependency injection implemented through a centralized configuration system that manages engine instantiation and lifecycle
Tech Stack
- Python 3.10+ backend with Flask, Jinja2, and Flask-Babel for internationalization
- TypeScript and JavaScript frontend powered by Vite, with Pyright and ESLint for type safety and linting
- SQLAlchemy with support for SQLite and PostgreSQL, including schema migrations and versioning
- Unified development workflow orchestrated by Makefile and mise for consistent Python, Node.js, and Go tooling
- Containerized deployment via Docker and Podman, with CI/CD pipelines and automated translation management
Code Quality
- Extensive test suite covering engine parsing, preference validation, and edge-case handling with async and mocking support
- Modular test organization mirroring application structure, ensuring focused validation of components
- Robust preference validation using custom exceptions to enforce configuration integrity and prevent invalid states
- Consistent naming and test structure that prioritize clarity and diagnostic readability
- Deterministic testing via Flask test client and environment isolation, minimizing external dependencies
What Makes It Unique
- Decentralized search aggregation that routes queries across hundreds of engines while preserving user privacy by default
- Built-in query anonymization layer that strips tracking parameters and enforces secure communication protocols
- Runtime engine discovery with adaptive fallback routing based on performance and reliability metrics
- Custom deduplication and relevance scoring that cross-references metadata to surface authoritative results
- Zero-JavaScript client interface with advanced filtering and instant previews
- Native Tor and I2P gateway integration for anonymous search without external proxy configuration