Deploy and Host SearXNG API on Railway

A private SearXNG instance configured for backends, scripts, and LLM agents that need a plain GET JSON search API.

About Hosting SearXNG API

This template intentionally creates no public domain. SearXNG has no authentication and this configuration disables its rate limiter, so other services in the same Railway project should call it over Railway's private network:

http://searxng.railway.internal:8080

Traffic stays inside the project and does not require public networking. The image provides an exact curated engine set, bounded upstream request timeouts, and a /healthz deployment check.

Why Deploy SearXNG API on Railway?

Railway provides private service discovery, health-checked deployments, and automatic tested image updates without requiring a public endpoint or an additional database service. The generated secret and internal DNS name make the instance ready for another service in the project to use immediately.

Common Use Cases

Private web search for AI agents and LLM applications
Search aggregation for a backend service
Developer and package search through GitHub, StackOverflow, npm, PyPI, and related engines
Academic search through arXiv, Semantic Scholar, Crossref, and PubMed

Dependencies for SearXNG API Hosting

The template is a single service. Redis or Valkey is not required because this private configuration runs without SearXNG's rate limiter.

Deployment Dependencies

SEARXNG_SECRET — the only required variable; generated automatically by the template
A second service in the same Railway project if you want to consume the API over private networking
SEARXNG_BASE_URL only if you deliberately enable a public domain

Configuration

The template generates SEARXNG_SECRET automatically. It is used by SearXNG for server-side cryptography.

SEARXNG_BASE_URL is optional and is only needed when you deliberately expose the service through a public domain.

API Usage

From another service in the same Railway project:

# General web search
curl "http://searxng.railway.internal:8080/search?q=python+async&format=json"

# Developer sources
curl "http://searxng.railway.internal:8080/search?q=tokio+channels&format=json&categories=it"

# Academic sources
curl "http://searxng.railway.internal:8080/search?q=mixture+of+experts&format=json&categories=science"

# Specific engines
curl "http://searxng.railway.internal:8080/search?q=transformers&format=json&engines=arxiv,github"

The JSON response includes results, answers, infoboxes, and suggestions.

Curated Search Engines

General: DuckDuckGo, Brave, Startpage, Qwant, Mojeek, Wikipedia, Wikidata, Currency
Developer: GitHub, GitLab, StackOverflow, MDN, Hacker News, Arch Wiki, npm, PyPI, crates.io, pkg.go.dev, Docker Hub
Academic: arXiv, Semantic Scholar, Crossref, PubMed

Engines that commonly block datacenter IPs are excluded to reduce failures and latency.

Optional Public Access

If you need access from outside Railway, enable public networking on target port 8080 and set SEARXNG_BASE_URL to the generated HTTPS URL.

Anyone with that URL can query the instance. Add your own authentication layer or enable SearXNG's limiter before treating it as a public service.

Updates

The template follows the maintained 1.1 image channel and applies tested patch releases during its configured maintenance window. Each release is built and smoke-tested before publication.

Source and full documentation: https://github.com/joeychilson/railway-searxng