Web Researcher MCP

This is a web research tool that lets Claude search specific domains you trust instead of the entire web. It exposes operations like web_search with configurable "lenses" (curated site lists for medical, legal, academic sources), scrape_page for full text extraction from PDFs and web pages, and academic_search that returns actual DOIs. The pitch is citation accuracy: if you're writing something that gets published or submitted to court, you need sources you can verify, not AI hallucinations. It runs locally via stdio, keeps queries private, and includes sequential_search for multi-step research where the AI builds on previous findings. Supports patent lookups, news filtering, and image search alongside the core web scraping.