Security researcher Zheng Yu of Depthfirst has released proof-of-concept exploits demonstrating remote code execution against PostgreSQL, MySQL, Oracle Database and ClickHouse. The attacks show that low-privilege accounts, even those unable to read a single application table, can still run arbitrary code on the database server by reaching memory-corruption bugs through ordinary SQL operations.
The researcher argues that database permissions control only which tables and schemas an account may access, not the memory-unsafe C and C++ code the engine executes when reading, planning and running queries. Common grants such as PostgreSQL’s default TEMPORARY privilege or the right to create temporary tables in MySQL and ClickHouse provide enough access to trigger the flaws.
PostgreSQL
In PostgreSQL, CVE-2026-116331 allows an account that can create database objects to execute code as the operating-system user running the database. The flaw was fixed on November 12 in versions 18.7, 17.12, 16.16, 15.20 and 14.25. The researcher demonstrated a plain login role building required objects inside its own temporary schema and converting a type-confusion bug into full server control.
MySQL
MySQL Server and Cluster versions 26.7.0 to 26.7.1 contain CVE-2026-101786 in the query optimizer. Although Oracle rated the issue as requiring high privileges, the researcher achieved code execution using only an account permitted to create temporary tables.
Oracle Database
Oracle Database is the sole case where no account is required. A chain combining a bug in Oracle Text (CVE-2026-114819, CVSS 9.9) with a flaw in the GraphQL interface of Oracle REST Data Services (CVE-2026-91609) reaches the engine through an exposed web endpoint and executes commands before any login occurs. Both issues have been patched in recent Oracle updates.
ClickHouse
ClickHouse version 26.7.5.10 is affected by an integer-overflow write reachable by setting the single option format_binary_max_string_size to zero. No CVE has been assigned and no fixed version has been confirmed.
Beyond applying patches, administrators should review account privileges and revoke the TEMPORARY privilege and broad temporary-table rights from accounts that do not require them. The Oracle attack path cannot be mitigated by privilege reduction because it begins without any database account.


