rgem227/knoweldge-base

What it solves

This project provides a centralized system for managing and storing documents across various formats (PDF, Word, Excel, PowerPoint, text, and images) that can be accessed by both human users via a web interface and AI agents via a specialized protocol.

How it works

The system uses a FastAPI backend with a SQLite database for metadata and a local filesystem for file storage. It features a web-based frontend for document upload, organization into nested folders, and in-browser previews. To enable AI integration, it includes a Model Context Protocol (MCP) server that allows agents to perform read-only operations such as searching for knowledge, listing documents, and retrieving document content using JWT authentication for security.

Who it’s for

It is designed for users who need a structured, private knowledge base for their files and developers who want to provide their AI agents with a secure, read-only interface to access that specific data.

Highlights

• Multi-format Support: Handles a wide range of files including PDF, DOCX, XLSX, PPTX, and legacy Office formats.
• Agent-Ready: Implements an MCP server allowing AI agents to search and retrieve documents via stdio or HTTP transports.
• Organized Storage: Supports nested folders, breadcrumb navigation, and batch uploads.
• Secure Access: Uses JWT authentication and role-based access control to ensure agents and users only see authorized documents.