Home

web-search-mcp

web-search-mcp is a FastMCP server providing web search, content extraction, and research tools for LLM clients. It implements the Model Context Protocol (MCP) to give LLMs access to real-time web information across multiple free data sources.

Quick Start

Prerequisites

• Python 3.11+
• uv for dependency management

Installation

git clone https://github.com/sydasif/web-search-mcp.git
cd web-search-mcp
uv sync

Running the Server

uv run web-search-mcp

Core Capabilities

• Web Search: Fast, raw search results via DuckDuckGo.
• Content Extraction: Clean text extraction from URLs.
• Community Signal: Reddit search for real-user experiences.
• Tech Discourse: Hacker News search for developer opinions and startup discussions.
• Code Discussions: GitHub Issues and PRs search across repositories.
• Prediction Signals: Polymarket search for market odds and probabilities.
• AI Research: Deep research and page analysis via Groq GPT-OSS and Compound systems.

Sources at a Glance

Source	Tools	Auth
DuckDuckGo	web_search, fetch_page, search_docs	None
Reddit	reddit_search	None
Hacker News	hackernews_search	None
GitHub	github_search	Optional token
Polymarket	polymarket_search	None
Groq	groq_browse, groq_research, groq_analyze_page	GROQ_API_KEY

---

Architecture | Tools | Development | Configuration