A self-improving AI agent that creates and refines skills from experience, runs on many model providers, and works through CLI and messaging apps.
ihor-sokoliuk/mcp-searxng
An MCP server that lets AI assistants like Claude and Cursor run web searches through a SearXNG instance you operate or trust.
- Stars
- 1,316
- Forks
- 178
- Open issues
- 5
- Last commit
- 2 days ago
What is it?
mcp-searxng is a Model Context Protocol server that connects AI assistants to SearXNG, a self-hostable metasearch engine. It exposes tools for searching the web, fetching query suggestions, inspecting an instance's categories and engines, and reading the text of a known URL, so an assistant can gather sources while answering a question.
The server does not install SearXNG; you need an existing instance with JSON output enabled. It runs locally over STDIO through npx, a global npm install, or Docker, or as an independently run Streamable HTTP service with optional static bearer or OAuth protection. Settings such as replica failover, caching, proxies and browser solver integrations are configured through environment variables.
Privacy depends on the SearXNG deployment. An operator-controlled instance avoids sending queries to a third-party search operator, while a public instance receives and may log the query.
Key features
- 01Web search with pagination, filters, direct answers and text or JSON output.
- 02Reads HTML, structured text and bounded PDF content from a given URL.
- 03Returns search suggestions and reports the instance's categories and engines.
- 04Supports replica failover, HTML fallback, caching and proxy settings.
- 05Runs over local STDIO or Streamable HTTP with optional bearer or OAuth protection.
Use cases
- Giving Claude or Cursor web search without a paid search API key.
- Running private research in an assistant against your own SearXNG instance.
- Letting an AI agent read a specific web page or PDF during a task.
- Hosting a shared HTTP MCP endpoint for a team's assistants.
Topics
FAQ
- What is mcp-searxng?
- An MCP server that lets AI assistants like Claude and Cursor run web searches through a SearXNG instance you operate or trust. mcp-searxng is a Model Context Protocol server that connects AI assistants to SearXNG, a self-hostable metasearch engine. It exposes tools for searching the web, fetching query suggestions, inspecting an instance's categories and engines, and reading the text of a known URL, so an assistant can gather sources while answering a question.
- Is mcp-searxng free to use? What license does it have?
- mcp-searxng is open source under the MIT license. Check the license terms before using it in commercial products.
- Is mcp-searxng still maintained?
- Actively maintained. The last commit to mcp-searxng was on Oct 19, 2026.
- What can I use mcp-searxng for?
- Giving Claude or Cursor web search without a paid search API key. Running private research in an assistant against your own SearXNG instance. Letting an AI agent read a specific web page or PDF during a task. Hosting a shared HTTP MCP endpoint for a team's assistants.
- Where can I get mcp-searxng?
- The source code and installation instructions are on GitHub: https://github.com/ihor-sokoliuk/mcp-searxng
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source personal AI assistant and Agent Harness that plans tasks, runs tools and skills, and keeps memory across web and IM channels.
Self-hostable AI answering engine that searches the web via SearxNG and cites sources, working with local LLMs through Ollama or cloud providers.
On-screen AI assistant for desktop that reads the current app's context and uses multiple LLMs and MCP tools to help you act across apps.
Open-source platform combining AI agents and workflow automation, self-hostable or embeddable in SaaS products, with 250+ connectors.
An agent workspace for knowledge work and software development, offering a choice of models and open-source agent harnesses, runnable locally or hosted.