CVE-2026-4593: Erupt Framework SQL Injection Vulnerability

CVE-2026-4593 Overview

A SQL injection vulnerability has been identified in Erupt bis version 1.13.3, specifically affecting the EruptDataQuery function within the MCP Tool Interface component. The vulnerability exists in the file erupt-ai/src/main/java/xyz/erupt/ai/call/impl/EruptDataQuery.java and allows remote attackers to manipulate SQL queries through Hibernate, potentially compromising database integrity and confidentiality.

Critical Impact
Remote attackers can exploit this SQL injection vulnerability to extract sensitive data, modify database contents, or potentially escalate privileges within the affected application.

Affected Products

Erupt bis 1.13.3
Erupt AI MCP Tool Interface component

Discovery Timeline

2026-03-23 - CVE CVE-2026-4593 published to NVD
2026-03-24 - Last updated in NVD database

Technical Details for CVE-2026-4593

Vulnerability Analysis

This vulnerability is classified as CWE-89 (SQL Injection), a critical flaw that occurs when user-controlled input is improperly incorporated into SQL queries without adequate sanitization or parameterization. In the context of Erupt bis, the EruptDataQuery function within the AI module fails to properly validate or sanitize input before constructing Hibernate queries, allowing attackers to inject malicious SQL statements.

The vulnerability is remotely exploitable and requires only low-level privileges to initiate an attack. No user interaction is necessary for exploitation. According to the disclosure, the vendor was contacted about this vulnerability but did not respond, and exploit information has been published publicly.

Root Cause

The root cause of this vulnerability lies in improper input handling within the EruptDataQuery.java file. The function fails to implement proper parameterized queries or input validation when constructing SQL statements through Hibernate. This allows specially crafted input to escape the intended query context and inject arbitrary SQL commands.

Attack Vector

The attack vector is network-based, allowing remote exploitation. An authenticated attacker with low-level privileges can target the MCP Tool Interface endpoint and submit malicious input that gets processed by the vulnerable EruptDataQuery function. The injected SQL commands are then executed against the underlying database through Hibernate ORM.

The vulnerability manifests when user-supplied data is concatenated directly into SQL queries without proper escaping or parameterization. Attackers can leverage standard SQL injection techniques to extract data, bypass authentication logic, or manipulate database records. Technical details and proof-of-concept information have been documented in the Feishu Document Resource and VulDB entry #352430.

Detection Methods for CVE-2026-4593

Indicators of Compromise

Unusual database query patterns or errors in application logs related to the Erupt AI module
Suspicious requests to MCP Tool Interface endpoints containing SQL syntax characters (single quotes, UNION statements, OR 1=1 patterns)
Database audit logs showing unexpected queries or data extraction attempts
Authentication anomalies or privilege escalation events associated with the EruptDataQuery function

Detection Strategies

Implement web application firewall (WAF) rules to detect SQL injection payloads targeting the Erupt AI endpoints
Deploy database activity monitoring to identify anomalous query patterns or unauthorized data access
Enable detailed logging for the MCP Tool Interface component and monitor for injection attempts
Use runtime application self-protection (RASP) solutions to detect and block SQL injection in real-time

Monitoring Recommendations

Configure alerting for SQL error messages in application logs that may indicate injection attempts
Monitor for high-volume requests to the vulnerable endpoint from single IP addresses
Track database query execution times for anomalies that could indicate time-based blind SQL injection
Review authentication logs for suspicious privilege changes or account compromise

How to Mitigate CVE-2026-4593

Immediate Actions Required

Restrict network access to the MCP Tool Interface endpoints until a patch is available
Implement web application firewall rules to filter SQL injection payloads
Review and audit all database accounts used by the Erupt application for least-privilege compliance
Enable comprehensive logging for the affected component to detect exploitation attempts

Patch Information

At the time of publication, the vendor has not responded to disclosure attempts and no official patch is available. Organizations using Erupt bis 1.13.3 should monitor the official Erupt project repository for security updates. Additional vulnerability details are available at VulDB #352430 and VulDB Submission #775593.

Workarounds

Deploy a web application firewall with SQL injection detection rules in front of the affected application
Implement input validation at the application boundary to reject requests containing SQL metacharacters
Restrict access to the MCP Tool Interface to trusted networks or authenticated users only
Consider disabling the Erupt AI module if it is not critical to business operations

bash

# Example WAF rule configuration for SQL injection protection
# Block common SQL injection patterns in request parameters
SecRule ARGS "@detectSQLi" "id:1001,phase:2,deny,status:403,msg:'SQL Injection Attempt Detected'"

# Restrict access to vulnerable endpoint by IP whitelist
# iptables -A INPUT -p tcp --dport 8080 -s trusted_ip_range -j ACCEPT
# iptables -A INPUT -p tcp --dport 8080 -j DROP