A Model Context Protocol (MCP) server providing tools for document reading, web fetching, and web searching over HTTP.
- Document Tools: Read and extract content from various document formats (PDF, DOCX, XLSX, PPTX, TXT, CSV, MD, RTF)
- Web Fetch: Retrieve web content from URLs in multiple formats (markdown, html, text)
- Web Search: Search the web using DuckDuckGo search engine
go installOr build from source:
go build -o fkmcps# Start with all tools enabled (default: localhost:8000)
fkmcps server
# Specify host and port
fkmcps server --host 0.0.0.0 --port 8080
# Enable specific tools only
fkmcps server --tools doc,search
# Interactive tool selection
fkmcps server --interactive# Check for updates and install
fkmcps updateget_document_info- Get document metadata (type, size, pages, etc.)read_document_smart- Intelligently read document content with automatic chunkingread_document_by_page- Read specific page rangesread_document_by_line- Read specific line ranges
fetch- Fetch web content from URL with customizable output format
search- Search the web using DuckDuckGo
Server flags:
--host- Host to listen on (default:localhost)--port- Port number (default:8000)--tools- Comma-separated list of tools to enable--interactive/-i- Interactive tool selection
Client flags:
--host- Host to connect to (default:localhost)--port- Port number (default:8000)--proto- Protocol (default:http)
MIT