arxiv-mcp-server vs pdf-reader-mcp
arxiv-mcp-server: The arxiv-mcp-server is an AI-powered platform designed to streamline academic research by enabling efficient arXiv paper search, citation analysis, and trend tracking. It offers robust data processing and intelligent discovery, allowing researchers to export findings in multiple formats and focus on their core work.; pdf-reader-mcp: PDF Reader MCP is a production-ready Model Context Protocol server for AI agents. It extracts text, images, and metadata with 5-10x faster parallel processing and Y-coordinate content ordering. With 94%+ test coverage and full TypeScript support, it ensures reliability and performance.
TL;DR
Assemble literature reviews.
- You have a config built on pdf-reader-mcp and only need PDFs: one `read_pdf` tool reads text, images, metadata and page ranges from paths or URLs.
- You want per-page error isolation, so one bad page does not fail the file.
- You can set `--allow-dir` or `MCP_PDF_ALLOWED_DIRS` to limit file access.
- You are starting fresh: the project was renamed to anymd, which covers more formats.
- You need OCR for scanned PDFs: the README does not describe it.
- You would run it with defaults: it can read any local file and fetch any URL until you set allowed directories.
Side-by-Side Comparison
Features
Use Cases
FAQ
Deploy on DigitalOcean
Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.