Custom MCP Server
by degenhero
A custom Machine Conversation Protocol (MCP) server that integrates various tools for AI assistants. It provides functionalities like web search, file operations, GitHub integration, and sequential thinking.
Last updated: N/A
What is Custom MCP Server?
This is a custom MCP server designed to enhance AI assistants with a range of tools, including web search, file system access, GitHub integration, sequential thinking capabilities, and a download utility. It comprises a core server and an MCP protocol adapter.
How to use Custom MCP Server?
First, clone the repository and install the dependencies using npm run install:all. Then, start both servers with npm run start:both. To integrate with Claude, install it as an MCP server using npx @anthropic-ai/install-mcp custom-mcp-server http://localhost:3001.
Key features of Custom MCP Server
Brave Search integration for web and local searches
Secure filesystem operations with access controls
GitHub integration for repository and code management
Sequential thinking for step-by-step problem solving
Download utility for fetching files from the internet
Use cases of Custom MCP Server
Enabling AI assistants to perform web searches and retrieve information.
Allowing AI assistants to securely manage and manipulate files.
Integrating AI assistants with GitHub for code management and collaboration.
Providing AI assistants with the ability to solve complex problems through sequential reasoning.
Enabling AI assistants to download files from the internet for various purposes.
FAQ from Custom MCP Server
What is MCP?
What is MCP?
MCP stands for Machine Conversation Protocol, a standard for communication between AI models and external tools.
How do I configure the server?
How do I configure the server?
Configuration is managed through environment variables in a .env file. You need to set API keys and tokens for services like Brave Search and GitHub.
How do I add a new function?
How do I add a new function?
Create a handler in the appropriate file in the functions directory, add the function definition to mcp-schema.json, and update the routing in index.js.
How can I test the functions?
How can I test the functions?
You can test functions directly against the MCP server using curl with a POST request to the /mcp endpoint.
Where can I find detailed documentation?
Where can I find detailed documentation?
Detailed documentation on all available functions can be found in the documentation.md file.