web-researcher-mcp

web-researcher-mcp is an open-source Model Context Protocol (MCP) server written in Go that equips AI assistants like Claude, Cursor, and any MCP-compatible client with honest, citation-verified web research capabilities. Built by Zohar Babin and released under the MIT license, it runs entirely on the user's machine — queries never pass through a third-party server. The latest release is v1.34.0, published June 16, 2026, which added a Python SDK.

What It Is

web-researcher-mcp is an MCP server that plugs into AI assistants and gives them a suite of research tools: web search, full-page scraping, academic paper lookup, patent search, SEC filing retrieval, US court record search, citation verification, and more. Unlike AI search products that synthesize results from opaque indexes, this tool fetches and reads primary sources directly, returning working links and properly formatted citations (APA, MLA, BibTeX, RIS, CSL-JSON). The core design goal is eliminating hallucinated citations — every result must be a real, verifiable URL.

How Search Lenses Work

A central feature is "search lenses" — curated domain lists that restrict which websites an AI is allowed to search. Instead of querying the open web and surfacing SEO spam or AI-generated content, a lens limits results to trusted sources for a given field:

• medical — PubMed, WHO, NIH, and clinical sources
• legal — law databases, case repositories, statutes
• academic — preprint servers, open-access journals, repositories
• security — CVEs, advisories, vulnerability research
• finance — SEC EDGAR, market filings
• journalism — public records, FOIA, corporate filings

Users can also define custom lenses by listing trusted domains in a JSON file. Lenses work with every supported search provider.

Tool Suite and Research Capabilities

The server exposes over 20 tools to connected AI assistants, covering the full research workflow:

• web_search / search_and_scrape — web search with optional lens filtering and quality scoring
• scrape_page — full-text extraction of web pages, PDFs, Word docs, YouTube transcripts, and Hacker News threads
• academic_search — real papers with DOIs, authors, citation counts, and open-access links via OpenAlex/CrossRef
• citation_graph — walks a paper's citation neighborhood with intent/influence signals
• patent_search — US, European, and international patent offices
• filing_search — SEC EDGAR filings (10-K, 10-Q, 8-K) and structured XBRL company facts
• legal_search — US court opinions and dockets via CourtListener
• verify_citation / audit_bibliography — checks whether citations exist, match real records, and are not retracted
• answer — synthesized factual answers with citations (requires Exa provider)
• sequential_search — multi-step deep research with session persistence
• research_export — exports sessions as markdown or JSON reports with full provenance

Guided prompt templates (comprehensive-research, fact-check, competitive-analysis, literature-review) are also bundled and surface as slash commands in supported AI apps.

Deployment and Setup Paths

The server ships as a single signed binary with multiple install paths:

• Python/uvx — uvx web-researcher-mcp (no Go required, cross-platform)
• Homebrew — brew install zoharbabin/tap/web-researcher-mcp (macOS/Linux)
• One-command shell installer — curl/PowerShell scripts for macOS, Linux, and Windows
• WinGet / Scoop — Windows package managers
• Docker — zoharbabin/web-researcher-mcp image with Chromium bundled
• Go install — go install from source
• One-click buttons — deep-link installers for Cursor, VS Code, and LM Studio

DuckDuckGo is the zero-config default (no API key needed). Optional providers — Google PSE, Brave, Serper, Tavily, Exa, SearXNG, SearchAPI.io, and Hacker News — can be added by setting environment variables. Multiple providers can be configured for automatic failover via SEARCH_ROUTING.

Update: v1.34.0 — Python SDK

The v1.34.0 release (June 16, 2026) introduced a Python SDK (web-researcher-mcp on PyPI) with an async WebResearcherClient and a sync wrapper for scripts and notebooks. The SDK is installable via pip, uv, or uvx, and the PyPI wheels vendor the compiled Go binary so no Go toolchain is required. The project also ships an HTTP server mode with OAuth 2.1 support for team deployments, and the repository shows active development with 27 stars and 4 forks as of the latest data.