Shelpuk-AI-Technology-Consulting
kindly-web-search-mcp-server
Web search and content retrieval MCP server for AI coding tools, delivering full conversations and structured content from specialized sources.
Repository last updated Jul 14, 2026
Overview
Kindly Web Search is an MCP server that provides web search and robust content retrieval for AI coding assistants and agents. It supports multiple search providers—Serper, Tavily, and SearXNG—and returns full conversations, answers, and metadata in a single call, eliminating the need for secondary scraping.
The server uses a priority chain of specialized handlers to extract structured content from StackExchange, GitHub Issues, GitHub Discussions, Wikipedia, and arXiv. For other URLs, it falls back to a headless Chromium browser via nodriver to parse and return clean Markdown content.
Designed for developers using Claude Code, Codex, Cursor, GitHub Copilot, Gemini, and other AI coding tools, Kindly Web Search improves code generation quality by providing up-to-date API documentation and solution discussions directly to the AI.
Highlights
- Supports Serper, Tavily, and SearXNG search providers
- Returns full conversations and metadata in one call
- Specialized handlers for StackExchange, GitHub, Wikipedia, arXiv
- Headless browser fallback for any webpage
- Markdown fast paths for doc sites like Aliyun and Cloudflare
Features
Multiple Search Providers
Supports Serper, Tavily, and self-hosted SearXNG with automatic fallback priority.
Specialized Content Handlers
Extracts structured Markdown from StackExchange, GitHub Issues/Discussions, Wikipedia, and arXiv via native APIs.
Universal HTML Extraction
Uses a headless Chromium browser to parse any webpage and return clean Markdown content.
Markdown Fast Paths
Skips the browser for supported doc sites by requesting Markdown directly via HTTP, reducing latency.
Proxy Support
All HTTP-based handlers support HTTP and SOCKS5 proxies via environment variables; the browser supports proxy via KINDLY_CHROME_PROXY.
Installation
Commands are reproduced exactly as published by the project maintainers. Always check the upstream repository for the current instructions.
curl -LsSf https://astral.sh/uv/install.sh | shirm https://astral.sh/uv/install.ps1 | iexuvx --versionbrew install --cask chromiumRequirements
- Python 3.13+
- Serper API key, Tavily API key, or self-hosted SearXNG instance
- Chromium-based browser (Chrome, Chromium, Edge, or Brave) for universal page extraction
- GitHub token (recommended for improved GitHub Issue extraction)
How it's built
Kindly Web Search is written in Python and uses the MCP protocol to expose two tools: web_search and get_content. The web_search tool returns search results with page content, while get_content retrieves content from a given URL. The server is run via uvx with the command kindly-web-search-mcp-server start-mcp-server.
The repository is organized into modules under src/kindly_web_search_mcp_server/: search/ contains provider implementations (Serper, Tavily, SearXNG), content/ has specialized handlers for StackExchange, GitHub, Wikipedia, and arXiv, and scrape/ handles universal HTML extraction using nodriver for headless Chromium. Configuration is managed via environment variables defined in settings.py.
Integration requires a search provider API key (Serper or Tavily) or a self-hosted SearXNG instance, plus a Chromium-based browser for universal page extraction. A GitHub token is recommended for better GitHub Issue extraction. The server can be configured with proxy settings for both HTTP and SOCKS5 protocols.
Source
This summary was written for Neura Market from the project's own repository. Install commands, identifiers and licence details are reproduced unchanged. The Shelpuk-AI-Technology-Consulting/kindly-web-search-mcp-server repository is the authoritative source and the place to file issues or contribute.
Category: MCP Servers · Licensed under MIT