This walks you through setting up academic paper search via MCP, specifically steering you toward the paper-search server instead of the buggy arxiv-mcp-server that fails during initialization. You get guided through config location choices (project, local, or global), Docker MCP verification, and server installation with a test search to confirm it works. The real value is multi-source access in one shot: arXiv, PubMed, bioRxiv, Semantic Scholar, Google Scholar, and more, all with download and read capabilities. It ends by updating your CLAUDE.md with the available tools and usage patterns. Solid onboarding if you're doing literature review or research and want to avoid the manual troubleshooting of broken MCP servers.
npx skills add https://github.com/neolabhq/context-engineering-kit --skill setup-arxiv-mcp