explaingit

zoharbabin/web-researcher-mcp

★ 11GoAudience · developerComplexity · 3/5Setup · moderate

TLDR

A Go server that connects AI assistants like Claude to the internet, letting them search the web, extract page content, browse PDFs, and find academic papers or patents during a conversation.

Mindmap

mindmap
  root((web-researcher-mcp))
    What it does
      Web search
      Page extraction
      PDF and doc support
      Academic paper search
    Search providers
      Google Custom Search
      Brave Search
      Serper
      SearXNG self-hosted
    Tech stack
      Go binary
      MCP protocol
    Setup
      Install binary
      Configure Claude settings
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Add real-time web search to a Claude conversation so it can look up current information.

USE CASE 2

Extract readable text from PDFs, Word files, or PowerPoint presentations using an AI assistant.

USE CASE 3

Search academic papers on arXiv or PubMed through an AI chat interface.

USE CASE 4

Route search queries through Brave, Google, or a self-hosted SearXNG instance with automatic fallback.

Tech stack

GoMCP

Getting it running

Difficulty · moderate Time to first run · 30min

Requires a search API key from Google Custom Search, Brave, Serper, or SearchAPI.io before the server can run.

In plain English

Web Researcher MCP is a server that gives AI assistants like Claude the ability to search the internet and extract content from web pages during a conversation. It communicates through the Model Context Protocol, a standard that lets AI tools connect to external data sources and services. The project is built in Go, which produces a single executable file with no complex installation requirements. Once connected to an AI assistant, the server provides eight distinct research tools the AI can call on your behalf. General web search returns results from your choice of search provider. Page extraction pulls readable content from any URL, including PDF documents, Word files, PowerPoint presentations, and YouTube transcripts. There are also dedicated search modes for news, academic papers (covering sources like arXiv and PubMed), and patents. A combined search-and-extract mode runs both steps together and scores results for quality. A session-based tool supports multi-step research where each query builds on the previous one. For fetching page content, the server tries four different approaches in order: requesting the page in a clean text format directly, then a quieter HTTP request designed to avoid blocking, then extracting the text from raw HTML, and finally using a headless browser that can run JavaScript-heavy pages. This tiered approach means it can handle most websites including those that require JavaScript to display their content. Search requests can be routed through several providers: Google Custom Search, Brave Search, Serper, SearchAPI.io, and a self-hosted option called SearXNG. You configure which provider or providers to use via environment variables, and the server can fall back to a second provider automatically if the first fails. Setup for Claude Code takes about two steps: install the binary and add a few lines to the Claude settings file pointing to it along with your search API credentials. Docker and source builds are also available.

Copy-paste prompts

Prompt 1
Search the web for the latest news about [topic] and summarize the key developments from the past week.
Prompt 2
Extract the main content from this URL and list the five most important points: [URL]
Prompt 3
Find recent academic papers about [topic] on arXiv and summarize the top 3 results with their key findings.
Prompt 4
Search for patents related to [technology] and list the main claims of the most relevant ones.
Prompt 5
Do a combined search and extract on [topic], score the results for quality, and recommend the three best sources.
Open on GitHub → Explain another repo

← zoharbabin on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.