Unauthenticated Stacked SQL Injection in R2R

R2R versions through 3.6.6 contain a high-severity stacked SQL injection vulnerability (CVE-2026-82526). The flaw resides in the vector index creation endpoint, where the 'index name' parameter is interpolated directly into a 'CREATE INDEX' statement. Due to a lack of identifier quoting and input validation, an unauthenticated attacker can inject arbitrary SQL commands. Because the application interacts with the backend database using a PostgreSQL superuser account, successful exploitation allows for unauthorized DDL and DML operations. This poses a significant risk to the integrity and confidentiality of the underlying database. Defenders should prioritize patching affected R2R instances to version 3.6.7 or later if available, or restrict access to the index creation API.

Impact

Successful exploitation of this vulnerability permits full database compromise. Attackers can execute arbitrary SQL statements, potentially leading to data exfiltration, modification of system configurations, or administrative access to the underlying PostgreSQL instance. As this is an unauthenticated vector, it is accessible to any remote actor capable of reaching the R2R API.

Recommendation

• Patch R2R instances to a version higher than 3.6.6 immediately.
• Implement strict input validation on the index name parameter to ensure it conforms to expected identifier patterns (e.g., alphanumeric only).
• Review database logs for unexpected DDL statements or queries originating from the R2R service account.
• Restrict network access to the R2R vector index creation API to authorized clients only via firewall or proxy rules.