Vulnerable MCP Server
This project is an intentionally vulnerable MCP (Model Context Protocol) app, designed for security research.
DO NOT use in production environments.
It executes raw SQL and system commands with no authentication or restrictions.
⚙️ MCP SERVER
A command execution server that combines:
- FastAPI for the HTTP interface
- SQLite as a persistent database
- Ollama LLM to interpret natural language queries
- JSON-RPC as the main API protocol
This system routes natural language input to either SQL queries or shell commands, using a locally running LLM via Ollama.
Built to test SQL Injection (SQLi) and Remote Code Execution (RCE) vulnerabilities
via FastAPI, JSON-RPC, and LLM-based decision logic.
⚙️ WARNING: Security Notice
This app is intentionally insecure:
- ❗ No authentication or access control
- ❗ Accepts and executes raw SQL queries and shell commands
- ❗ No input validation
- ❗ LLM responses are blindly executed
Use only in isolated environments, CTFs, or research labs.
⚙️ Features
- LLM-based decision logic for command routing (SQL or CLI)
- Native execution of SQL and terminal commands
- Auto-initializing SQLite database with sample data
- Simple, pluggable JSON-RPC methods
- Vulnerable by design — suitable for offensive/defensive testing
⚙️ Installation
This server cannot be installed
Intentionally vulnerable Model Context Protocol (MCP) server designed for security research that processes natural language queries through an LLM to execute SQL queries or shell commands without restrictions.