arxiv

This ArXiv MCP Server provides a set of tools to search, fetch, parse, and convert arXiv papers into Chinese Markdown and微信文章 formats. It exposes capabilities such as search_arxiv for locating papers by keyword, download_arxiv_pdf for retrieving the PDF, parse_pdf_to_markdown to generate Chinese Markdown from the PDF, and convert_to_wechat_article to produce a公众号-friendly article. The full process can be orchestrated with process_arxiv_paper to optionally include a WeChat article, or you can mix and match individual tools to fit your automation workflow. Each tool returns the outcome filenames (e.g., {arxivId}.pdf, {arxivId}_md_zh.md, {arxivId}_wechat.md), allowing smooth integration into existing pipelines.

Prerequisites:

• Node.js v18.x or newer
• npm (comes with Node.js)
• A SiliconFlow API Key and a configured WORK_DIR for storage

Install steps:

1. Clone the repository (or install via npm if published): git clone https://github.com/yzfly/arxiv-mcp-server.git cd arxiv-mcp-server

2. Install dependencies: npm install

3. Set required environment variables (example): export SILICONFLOW_API_KEY="your_api_key" export WORK_DIR="/path/to/your/data/directory"

4. Run in development mode (hot reloading / ts-node-style workflow depending on project setup): npm run dev

5. Build and run the production version (example): npm run build npm start

If you prefer NPX-based usage without local install, you can use: npx @langgpt/arxiv-mcp-server

Tips andNotes:

• Ensure SILICONFLOW_API_KEY is valid and WORK_DIR points to a writable directory.
• The MCP client configuration (e.g., Claude Desktop) can point to the server using command npx and pass env vars similarly to the README example.
• The output files are named according to the arXiv ID, such as {arxivId}.pdf, {arxivId}_md_zh.md, and {arxivId}_wechat.md; these names help integrate results into automated workflows.
• If you encounter API or download errors, check network connectivity, arXiv IDs, and that the WORK_DIR has proper permissions.
• You can enable verbose logs for debugging by prefixing commands with DEBUG=arxiv-mcp-server (e.g., DEBUG=arxiv-mcp-server npx @langgpt/arxiv-mcp-server).
• This server relies on SiliconFlow for AI-driven content understanding and generation; ensure your API quota aligns with expected usage.