What can mcp-searxng do?

web-search-via-searxng-metasearch, mcp-tool-registration-for-search, pagination-aware-search-result-retrieval, multi-engine-metasearch-aggregation, self-hosted-search-endpoint-configuration, mcp-protocol-server-implementation, claude-desktop-integration-via-mcp

mcp-searxng

MCP ServerFree

MCP server for SearXNG integration

Open Source

/ 100

7 capabilities

Capabilities7 decomposed

web-search-via-searxng-metasearch

Medium confidence

Executes web searches through a SearXNG instance (self-hosted or public) using the MCP protocol, enabling Claude and other MCP clients to query multiple search engines simultaneously without direct API dependencies. Implements MCP tool registration to expose search as a callable function with query and optional pagination parameters, abstracting away HTTP communication with the SearXNG backend.

Solves for

I want Claude to search the web in real-time without using OpenAI's Bing integrationI need to route search queries through my self-hosted SearXNG instance for privacyI want to give my AI agent access to metasearch results from multiple engines at onceI need to integrate web search into an MCP-compatible application without writing custom HTTP clients

Best for

developers building privacy-focused AI agents with Claude

teams running self-hosted SearXNG instances who want MCP integration

builders creating multi-tool AI applications using the Model Context Protocol

Requires

Node.js 16+ (MCP SDK requirement)

Running SearXNG instance (self-hosted or public endpoint)

MCP-compatible client (Claude Desktop, custom MCP host, or compatible AI application)

Limitations

Depends on external SearXNG instance availability — no built-in fallback if SearXNG is down

Search quality and result freshness tied to SearXNG configuration and underlying engine sources

No built-in result ranking or deduplication across multiple search engines

What makes it unique

Bridges SearXNG (privacy-focused metasearch engine) with MCP protocol, enabling declarative search tool registration for Claude and other MCP clients without requiring custom HTTP wrapper code or API key management for individual search engines

vs alternatives

Provides privacy-preserving web search for MCP agents without Bing/Google API dependencies, unlike Claude's native search which relies on commercial APIs and cannot be self-hosted

mcp-tool-registration-for-search

Medium confidence

Registers search functionality as an MCP tool with schema validation, parameter definitions, and callable interface that MCP clients (like Claude) can discover and invoke. Uses MCP's tool definition format to expose search with typed parameters (query string, pagination options) and structured response schemas, enabling semantic understanding of search capabilities by AI clients.

Solves for

I want Claude to automatically discover that search is available and understand its parametersI need to define search input/output schemas so Claude can validate queries before sending themI want to expose search as a callable tool in an MCP-compatible applicationI need search to appear in Claude's tool list with proper documentation

Best for

MCP server developers integrating third-party capabilities

teams building multi-tool AI agents with proper capability discovery

developers who want type-safe tool invocation with schema validation

Requires

MCP SDK for Node.js

Understanding of MCP tool definition format (JSON schema)

MCP-compatible client that supports tool discovery

Limitations

Tool schema must be defined statically at server startup — no dynamic schema updates

Parameter validation happens at MCP level, not at SearXNG level — mismatches may cause runtime errors

No built-in error handling for malformed search queries or SearXNG API failures

What makes it unique

Implements MCP's tool registration pattern specifically for SearXNG, handling schema definition, parameter validation, and client-side tool discovery without requiring manual tool binding code in client applications

vs alternatives

Enables automatic tool discovery and invocation in MCP clients (like Claude) without manual function binding, unlike direct HTTP clients which require explicit endpoint configuration and parameter handling

pagination-aware-search-result-retrieval

Medium confidence

Handles paginated search results from SearXNG by accepting page parameters and returning result sets with metadata about total results and current page position. Implements offset-based or cursor-based pagination depending on SearXNG API capabilities, allowing clients to retrieve large result sets incrementally without loading all results into memory at once.

Solves for

I want to retrieve the next page of search results without re-queryingI need to know how many total results are available for a search queryI want to iterate through search results in batches to avoid memory overheadI need to implement search result browsing in my AI agent

Best for

applications handling high-volume search queries with large result sets

AI agents that need to explore multiple result pages for comprehensive research

builders implementing search result browsing UIs

Requires

SearXNG instance with pagination support enabled

MCP client capable of passing pagination parameters

Limitations

Pagination implementation depends on SearXNG version and configuration — may not be consistent across instances

No built-in result caching between page requests — each page fetch hits SearXNG

Page size and maximum page depth may be limited by SearXNG configuration

What makes it unique

Abstracts SearXNG's pagination API into MCP tool parameters, allowing clients to request specific result pages without understanding SearXNG's underlying pagination mechanism or managing state between requests

vs alternatives

Provides stateless pagination through MCP parameters rather than requiring clients to manage session state or cursor tokens, simplifying integration with stateless AI clients like Claude

multi-engine-metasearch-aggregation

Medium confidence

Leverages SearXNG's ability to query multiple search engines (Google, Bing, DuckDuckGo, etc.) simultaneously and returns aggregated results through a single MCP interface. SearXNG handles engine selection, result deduplication, and ranking internally; this capability exposes that aggregation to MCP clients without requiring separate API calls to individual engines.

Solves for

I want search results from multiple engines without managing multiple API keysI need to compare results across different search engines programmaticallyI want to avoid search engine bias by querying multiple sources at onceI need comprehensive search coverage without direct dependencies on Google/Bing APIs

Best for

research-focused AI agents that need comprehensive result coverage

privacy-conscious teams avoiding single-engine dependency

applications requiring search engine diversity for bias mitigation

Requires

SearXNG instance with multiple search engines enabled

Network connectivity to all configured search engines

Limitations

Result quality varies by engine — SearXNG aggregation may not optimize for relevance

No control over which engines are queried — determined by SearXNG configuration

Deduplication logic is SearXNG-specific and may miss near-duplicate results

What makes it unique

Exposes SearXNG's multi-engine aggregation as a single MCP tool, eliminating the need for MCP clients to manage multiple search engine integrations or API keys while maintaining result diversity

vs alternatives

Provides multi-engine search through one MCP tool without API key management, unlike integrating Google/Bing/DuckDuckGo separately which requires multiple credentials and custom aggregation logic

self-hosted-search-endpoint-configuration

Medium confidence

Allows configuration of a custom SearXNG endpoint (self-hosted or public instance) at MCP server initialization, enabling organizations to route all search queries through their own infrastructure. Configuration is typically passed via environment variables or config files, and the MCP server maintains a persistent connection to the configured endpoint for all subsequent search requests.

Solves for

I want to run search through my organization's self-hosted SearXNG instanceI need to ensure all search queries stay within my network for complianceI want to customize SearXNG engine configuration for my use caseI need to point my MCP server at a specific SearXNG endpoint without code changes

Best for

enterprises with data residency or privacy requirements

teams running self-hosted SearXNG for cost control

organizations wanting to customize search engine selection

Requires

Running SearXNG instance (self-hosted or public endpoint URL)

Environment variable or config file support in MCP server

Network access from MCP server to SearXNG endpoint

Limitations

Endpoint configuration is static — requires server restart to change endpoints

No built-in endpoint health checking or failover to backup instances

Self-hosted SearXNG requires separate infrastructure management and maintenance

What makes it unique

Enables MCP server to be configured with custom SearXNG endpoints via environment variables, allowing deployment flexibility without code changes and supporting both self-hosted and public SearXNG instances

vs alternatives

Provides endpoint configuration at server level rather than client level, enabling centralized search routing and compliance enforcement across all MCP clients using this server

mcp-protocol-server-implementation

Medium confidence

Implements the Model Context Protocol (MCP) server specification in Node.js, handling MCP message serialization/deserialization, tool registration, request routing, and response formatting. Uses MCP SDK to manage the server lifecycle, client connections, and protocol compliance, abstracting away low-level MCP communication details from the search integration logic.

Solves for

I want to expose SearXNG search as an MCP tool to Claude and other MCP clientsI need a standards-compliant MCP server without building protocol handling from scratchI want my search integration to work with any MCP-compatible clientI need proper error handling and response formatting for MCP clients

Best for

developers building MCP servers for AI integrations

teams standardizing on MCP for tool integration

builders who want to avoid custom protocol implementation

Requires

Node.js 16+

MCP SDK (@modelcontextprotocol/sdk)

MCP-compatible client (Claude Desktop, custom host, etc.)

Limitations

MCP protocol overhead adds latency compared to direct HTTP calls

Server must maintain MCP connection state — not suitable for serverless deployments

Error handling is MCP-specific — may not map cleanly to SearXNG errors

What makes it unique

Implements MCP server specification using the official MCP SDK, handling protocol compliance, message routing, and client lifecycle management without requiring custom protocol implementation

vs alternatives

Uses standard MCP SDK rather than custom protocol implementation, ensuring compatibility with all MCP-compliant clients and reducing maintenance burden compared to custom HTTP wrappers

claude-desktop-integration-via-mcp

Medium confidence

Registers the MCP server with Claude Desktop through MCP's client discovery mechanism, making search available as a native tool within Claude's interface. Claude Desktop automatically discovers the MCP server, loads tool definitions, and enables users to invoke search directly in conversations without manual tool binding or configuration.

Solves for

I want to use SearXNG search directly in Claude Desktop conversationsI need Claude to automatically discover my search tool without manual setupI want to give Claude real-time web search capability through my self-hosted SearXNGI need search to appear in Claude's tool list alongside other capabilities

Best for

Claude Desktop users wanting self-hosted search integration

teams deploying MCP servers for Claude integration

developers building privacy-focused Claude workflows

Requires

Claude Desktop 0.2.0 or later with MCP support

MCP server running and accessible to Claude Desktop

MCP server configuration in Claude's config file (typically ~/.claude/config.json)

Limitations

Requires Claude Desktop 0.2.0+ with MCP support — not available in web version

MCP server must be running locally or accessible from Claude Desktop's environment

Tool discovery happens at Claude startup — changes require restart

What makes it unique

Integrates with Claude Desktop's MCP discovery mechanism, enabling automatic tool registration without manual configuration and allowing Claude to invoke search as a native capability within conversations

vs alternatives

Provides seamless Claude Desktop integration through MCP protocol rather than custom Claude API wrappers, enabling native tool discovery and invocation without code changes to Claude

Capabilities are decomposed by AI analysis. Each maps to specific user intents and improves with match feedback.

Related Artifactssharing capabilities

Artifacts that share capabilities with mcp-searxng, ranked by overlap. Discovered automatically through the match graph.

MCP Server27

@brave/brave-search-mcp-server

Brave Search MCP Server: web results, images, videos, rich results, AI summaries, and more.

web-results-retrieval-via-mcpsearch-result-filtering-and-parametersimage-search-results-retrieval

3 shared capabilities

MCP Server23

Brave Search

** - Web and local search using Brave's Search API. Has been replaced by the [official server](https://github.com/brave/brave-search-mcp-server).

web-search-via-brave-apilocal-search-indexing

2 shared capabilities

MCP Server23

Search1API

** - One API for Search, Crawling, and Sitemaps

multi-engine web search with filtering and time-range constraints

1 shared capability

MCP Server43

firecrawl-mcp-server

🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.

web search with result ranking and snippet extraction

1 shared capability

MCP Server38

@z_ai/mcp-server

MCP Server for Z.AI - A Model Context Protocol server that provides AI capabilities

web search integration with semantic result ranking

1 shared capability

MCP Server22

FetchSERP

** - All-in-One SEO & Web Intelligence Toolkit API [FetchSERP](https://www.fetchserp.com)

real-time serp data fetching with multi-engine support

1 shared capability

Best For

✓developers building privacy-focused AI agents with Claude
✓teams running self-hosted SearXNG instances who want MCP integration
✓builders creating multi-tool AI applications using the Model Context Protocol
✓organizations avoiding direct dependencies on commercial search APIs
✓MCP server developers integrating third-party capabilities
✓teams building multi-tool AI agents with proper capability discovery
✓developers who want type-safe tool invocation with schema validation
✓applications handling high-volume search queries with large result sets

Known Limitations

⚠Depends on external SearXNG instance availability — no built-in fallback if SearXNG is down
⚠Search quality and result freshness tied to SearXNG configuration and underlying engine sources
⚠No built-in result ranking or deduplication across multiple search engines
⚠Pagination support may be limited by SearXNG API response structure
⚠No caching layer — each query hits SearXNG directly, adding latency
⚠Tool schema must be defined statically at server startup — no dynamic schema updates

Requirements

Node.js 16+ (MCP SDK requirement)Running SearXNG instance (self-hosted or public endpoint)MCP-compatible client (Claude Desktop, custom MCP host, or compatible AI application)Network access to SearXNG endpointMCP SDK for Node.jsUnderstanding of MCP tool definition format (JSON schema)MCP-compatible client that supports tool discoverySearXNG instance with pagination support enabled

Input / Output

Accepts: text (search query string), optional parameters (page number for pagination, result count limit), JSON schema definitions, tool parameter specifications, search query string, page number or offset, optional result count per page, optional engine selection filter (if SearXNG supports it), SearXNG endpoint URL (environment variable or config parameter), optional authentication credentials if SearXNG requires them, MCP protocol messages (JSON-RPC format), tool invocation requests from MCP clients, natural language search queries from Claude conversations

Produces: structured JSON (search results with title, URL, snippet, source engine), paginated result sets, MCP tool definitions, callable tool interface for MCP clients, paginated result set with metadata, total result count, current page indicator, aggregated results from multiple engines, engine source attribution per result, validated endpoint connection, search requests routed to configured endpoint, MCP protocol responses, tool results formatted for MCP clients, search results displayed in Claude conversation, formatted result summaries for Claude to process

UnfragileRank

Adoption15%(30% weight)

Quality16%(25% weight)

Ecosystem50%(25% weight)

Match Graph10%(15% weight)

Freshness75%(5% weight)

UnfragileRank is computed from adoption signals, documentation quality, ecosystem connectivity, match graph feedback, and freshness. No artifact can pay for a higher rank.

Type: MCP Server

7 capabilities

Visit mcp-searxng→

Package Details

npm

Registry

1.0.3

Version

Weekly Downloads

About

MCP server for SearXNG integration

Alternatives to mcp-searxng

wink-embeddings-sg-100d24Repository

100-dimensional English word embeddings for wink-nlp

Compare →

voyage-ai-provider30API

Voyage AI Provider for running Voyage AI models with Vercel AI SDK

Compare →

@vibe-agent-toolkit/rag-lancedb27Agent

LanceDB implementation of RAG interfaces for vibe-agent-toolkit

Compare →

vectra41Repository

A lightweight, file-backed vector database for Node.js and browsers with Pinecone-compatible filtering and hybrid BM25 search.

Compare →

Are you the builder of mcp-searxng?

Claim this artifact to get a verified badge, access match analytics, see which intents users search for, and manage your listing.

Claim this artifact →Verification via email

Get the weekly brief

New tools, rising stars, and what's actually worth your time. No spam.

Data Sources

mcp registry

Looking for something else?

Search →

Capabilities7 decomposed

web-search-via-searxng-metasearch

Medium confidence

Solves for

Best for

developers building privacy-focused AI agents with Claude

teams running self-hosted SearXNG instances who want MCP integration

builders creating multi-tool AI applications using the Model Context Protocol

Requires

Node.js 16+ (MCP SDK requirement)

Running SearXNG instance (self-hosted or public endpoint)

MCP-compatible client (Claude Desktop, custom MCP host, or compatible AI application)

Limitations

Depends on external SearXNG instance availability — no built-in fallback if SearXNG is down

Search quality and result freshness tied to SearXNG configuration and underlying engine sources

No built-in result ranking or deduplication across multiple search engines

What makes it unique

vs alternatives

Provides privacy-preserving web search for MCP agents without Bing/Google API dependencies, unlike Claude's native search which relies on commercial APIs and cannot be self-hosted

mcp-tool-registration-for-search

Medium confidence

Solves for

Best for

MCP server developers integrating third-party capabilities

teams building multi-tool AI agents with proper capability discovery

developers who want type-safe tool invocation with schema validation

Requires

MCP SDK for Node.js

Understanding of MCP tool definition format (JSON schema)

MCP-compatible client that supports tool discovery

Limitations

Tool schema must be defined statically at server startup — no dynamic schema updates

Parameter validation happens at MCP level, not at SearXNG level — mismatches may cause runtime errors

No built-in error handling for malformed search queries or SearXNG API failures

What makes it unique

vs alternatives

pagination-aware-search-result-retrieval

Medium confidence

Solves for

Best for

applications handling high-volume search queries with large result sets

AI agents that need to explore multiple result pages for comprehensive research

builders implementing search result browsing UIs

Requires

SearXNG instance with pagination support enabled

MCP client capable of passing pagination parameters

Limitations

Pagination implementation depends on SearXNG version and configuration — may not be consistent across instances

No built-in result caching between page requests — each page fetch hits SearXNG

Page size and maximum page depth may be limited by SearXNG configuration

What makes it unique

vs alternatives

Provides stateless pagination through MCP parameters rather than requiring clients to manage session state or cursor tokens, simplifying integration with stateless AI clients like Claude

multi-engine-metasearch-aggregation

Medium confidence

Solves for

Best for

research-focused AI agents that need comprehensive result coverage

privacy-conscious teams avoiding single-engine dependency

applications requiring search engine diversity for bias mitigation

Requires

SearXNG instance with multiple search engines enabled

Network connectivity to all configured search engines

Limitations

Result quality varies by engine — SearXNG aggregation may not optimize for relevance

No control over which engines are queried — determined by SearXNG configuration

Deduplication logic is SearXNG-specific and may miss near-duplicate results

What makes it unique

Exposes SearXNG's multi-engine aggregation as a single MCP tool, eliminating the need for MCP clients to manage multiple search engine integrations or API keys while maintaining result diversity

vs alternatives

Provides multi-engine search through one MCP tool without API key management, unlike integrating Google/Bing/DuckDuckGo separately which requires multiple credentials and custom aggregation logic

self-hosted-search-endpoint-configuration

Medium confidence

Solves for

Best for

enterprises with data residency or privacy requirements

teams running self-hosted SearXNG for cost control

organizations wanting to customize search engine selection

Requires

Running SearXNG instance (self-hosted or public endpoint URL)

Environment variable or config file support in MCP server

Network access from MCP server to SearXNG endpoint

Limitations

Endpoint configuration is static — requires server restart to change endpoints

No built-in endpoint health checking or failover to backup instances

Self-hosted SearXNG requires separate infrastructure management and maintenance

What makes it unique

vs alternatives

Provides endpoint configuration at server level rather than client level, enabling centralized search routing and compliance enforcement across all MCP clients using this server

mcp-protocol-server-implementation

Medium confidence

Solves for

Best for

developers building MCP servers for AI integrations

teams standardizing on MCP for tool integration

builders who want to avoid custom protocol implementation

Requires

Node.js 16+

MCP SDK (@modelcontextprotocol/sdk)

MCP-compatible client (Claude Desktop, custom host, etc.)

Limitations

MCP protocol overhead adds latency compared to direct HTTP calls

Server must maintain MCP connection state — not suitable for serverless deployments

Error handling is MCP-specific — may not map cleanly to SearXNG errors

What makes it unique

Implements MCP server specification using the official MCP SDK, handling protocol compliance, message routing, and client lifecycle management without requiring custom protocol implementation

vs alternatives

Uses standard MCP SDK rather than custom protocol implementation, ensuring compatibility with all MCP-compliant clients and reducing maintenance burden compared to custom HTTP wrappers

claude-desktop-integration-via-mcp

Medium confidence

Solves for

Best for

Claude Desktop users wanting self-hosted search integration

teams deploying MCP servers for Claude integration

developers building privacy-focused Claude workflows

Requires

Claude Desktop 0.2.0 or later with MCP support

MCP server running and accessible to Claude Desktop

MCP server configuration in Claude's config file (typically ~/.claude/config.json)

Limitations

Requires Claude Desktop 0.2.0+ with MCP support — not available in web version

MCP server must be running locally or accessible from Claude Desktop's environment

Tool discovery happens at Claude startup — changes require restart

What makes it unique

vs alternatives

Provides seamless Claude Desktop integration through MCP protocol rather than custom Claude API wrappers, enabling native tool discovery and invocation without code changes to Claude

Capabilities are decomposed by AI analysis. Each maps to specific user intents and improves with match feedback.

Alternatives to mcp-searxng

wink-embeddings-sg-100d24Repository

100-dimensional English word embeddings for wink-nlp

Compare →

voyage-ai-provider30API

Voyage AI Provider for running Voyage AI models with Vercel AI SDK

Compare →

@vibe-agent-toolkit/rag-lancedb27Agent

LanceDB implementation of RAG interfaces for vibe-agent-toolkit

Compare →

vectra41Repository

A lightweight, file-backed vector database for Node.js and browsers with Pinecone-compatible filtering and hybrid BM25 search.

Compare →

mcp-searxng

Capabilities7 decomposed

web-search-via-searxng-metasearch

mcp-tool-registration-for-search

pagination-aware-search-result-retrieval

multi-engine-metasearch-aggregation

self-hosted-search-endpoint-configuration

mcp-protocol-server-implementation

claude-desktop-integration-via-mcp

Related Artifactssharing capabilities

@brave/brave-search-mcp-server

Brave Search

Search1API

firecrawl-mcp-server

@z_ai/mcp-server

FetchSERP

Best For

Known Limitations

Requirements

Input / Output

UnfragileRank

Package Details

About

Categories

Alternatives to mcp-searxng

Are you the builder of mcp-searxng?

Get the weekly brief

Data Sources

mcp-searxng

Capabilities7 decomposed

web-search-via-searxng-metasearch

mcp-tool-registration-for-search

pagination-aware-search-result-retrieval

multi-engine-metasearch-aggregation

self-hosted-search-endpoint-configuration

mcp-protocol-server-implementation

claude-desktop-integration-via-mcp

Related Artifactssharing capabilities

@brave/brave-search-mcp-server

Brave Search

Search1API

firecrawl-mcp-server

@z_ai/mcp-server

FetchSERP

Best For

Known Limitations

Requirements

Input / Output

UnfragileRank

Package Details

About

Categories

Alternatives to mcp-searxng

Are you the builder of mcp-searxng?

Get the weekly brief

Data Sources