jsboige-mcp-servers

What is jsboige-mcp-servers?

jsboige-mcp-servers is a collection of Model Context Protocol (MCP) servers that extend LLM capabilities by providing tools for file operations, Jupyter notebook interaction, and web page conversion to Markdown. It is aimed at developers using LLMs that support the MCP protocol.

How to use jsboige-mcp-servers?

Clone the repository, install dependencies with npm install and npm run install-all, then configure with npm run setup-config. Start individual servers by navigating to their directory (e.g., servers/quickfiles-server) and running npm start. Connect your LLM client to the server URL (typically http://localhost:3000).

Key features of jsboige-mcp-servers

• QuickFiles Server: read, list, edit, delete multiple files in one request
• Jupyter MCP Server: create, modify, execute notebooks and manage kernels
• JinaNavigator Server: convert web pages to Markdown via Jina API
• Unit testing, performance testing, and error handling for each server
• Standardized architecture with shared configuration and documentation
• Continuous integration via GitHub Actions

Use cases of jsboige-mcp-servers

• Bulk read or edit configuration files across a project
• Automate Jupyter notebook creation and execution from an LLM
• Capture web page content as Markdown for summarization or analysis
• Perform code analysis by reading multiple source files at once

FAQ from jsboige-mcp-servers

What are the prerequisites to use jsboige-mcp-servers?

Node.js 14.x or higher, npm 6.x or higher, and Git. Python with Jupyter is required only for the Jupyter MCP Server.

Can I run individual servers without the full repository?

Yes, each server is self-contained in servers/<server-name>/. You can install and run it independently after cloning.

Does the Jupyter MCP Server have known issues?

Yes, it may present authentication problems in certain configurations. See the Jupyter MCP troubleshooting guide for recommended solutions.

How do I configure authentication for Jupyter MCP?

Set the same token in config.json as used when starting Jupyter Notebook (e.g., --NotebookApp.token=test_token) and ensure --NotebookApp.allow_origin='*' is used.

What transport does jsboige-mcp-servers use?

The servers expose an HTTP endpoint, typically on port 3000. Your LLM client connects to this URL.