# ihor-sokoliuk/mcp-searxng

Private web search for AI assistants via SearXNG — supports Claude, Cursor, and any MCP client

Repository: https://github.com/ihor-sokoliuk/mcp-searxng
Canonical: https://ross.abutalabs.com/products/mcp-searxng
Homepage: https://www.npmjs.com/package/mcp-searxng
Language: TypeScript
License: MIT
License Family: permissive
Topics: ai, llm, mcp, mcp-server, mcp-servers, search, searxng, claude, cursor, privacy, web-search, self-hosted
Last push: 2026-08-31T05:26:58+00:00

## Health v2 (maintenance only)
Score: 88/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 100, release rhythm 99, longevity 44
- inputs: {"age_days": 619, "days_push": 2, "days_rel": 8, "gap_med": 2, "n_releases_24m": 54}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1183, forks 155 (observed 2026-09-01T02:14:03.000325+00:00)

## What it is
An MCP server that exposes a SearXNG metasearch instance to AI assistants like Claude Desktop, Claude Code, Cursor, and other MCP clients. It provides web search with pagination, time-range/language/safe-search filters, relevance scoring, and URL reading, keeping search traffic under an operator-controlled or self-hosted instance.

## Use cases
- give Claude Desktop or Cursor privacy-respecting web search via my own SearXNG instance
- add a self-hosted web search tool to any MCP client
- search the web from an AI assistant without sending queries to third-party search APIs
- run an MCP server that proxies SearXNG with pagination and safe-search filters
- let an LLM agent fetch and read URLs through a trusted metasearch backend
- configure multiple SearXNG replicas for resilient AI assistant search

## When to choose
- you already run or trust a SearXNG instance and want AI assistants to use it
- you need MCP-compatible web search for Claude, Cursor, VS Code, Windsurf, Cline, or other MCP clients
- privacy matters and you want to avoid commercial search API keys and tracking
- you want filters like time range, language, safe search, and min-score relevance in agent search

## When to avoid
- you have no SearXNG instance and don't want to host or trust one
- you need a general-purpose search API outside the MCP ecosystem
- your assistant doesn't support the Model Context Protocol

## Facets
- artifact type: service
- maturity: active
- function: mcp, search-engine, http-client, privacy
- domain: artificial-intelligence, privacy, self-hosted, developer-tools
- platform: self-hosted, cross-platform
- tags: mcp-server, searxng, web-search, ai-assistants, claude-desktop, cursor, metasearch, model-context-protocol, search, nodejs, docker

## Member repositories
- ihor-sokoliuk/mcp-searxng (main) score 88

## Provenance
- Observed fields: from GitHub, fetched 2026-09-01T02:14:03.000325+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:31:38.448774+00:00, confidence not recorded.
  - readme: https://github.com/ihor-sokoliuk/mcp-searxng (fetched 2026-09-01T02:14:03.000325+00:00, sha fdb78f3c143b)
  - registry_npm: https://registry.npmjs.org/mcp-searxng (fetched 2026-08-29T12:36:12.498816+00:00, sha c5ed8bb4df38)
- Data as of 2026-08-30T08:39:29.467469+00:00.