Unauthenticated SQL Injection in the WordPress plugin Auto Listings (2.7.3)
Auto Listings (WordPress.org, version 2.7.3) Auto Listings (2.7.3) hooks the front-end search query and passes the search parameter through sanitize_text_field() — which strips HTML tags but does not escape SQL — directly into a LIKE clause of a live query. An unauthenticated visitor can inject SQL through the listings search and read arbitrary database contents (WordPress user password hashes, secrets). No fixed release is available.
Auto Listings 2.7.3 (WordPress.org) contains an unauthenticated sql injection. The plugin registers a handler reachable by visitors who are not logged in, and it concatenates a request parameter straight into a live SQL statement.
An unauthenticated front-end query hook passes the search parameter through sanitize_text_field() — which strips HTML tags but performs no SQL escaping — straight into a LIKE clause, so the value reaches the database unescaped. The result is a blind/time-based SQL-injection channel through which the full database (user password hashes, secrets) can be extracted.
Auto Listings was surfaced automatically by the codelake WordPress static-analysis pipeline (source-to-sink taint) · AI false-positive triage · manual source verification, and then verified by hand against the shipped 2.7.3 artifact.
Class: SQL Injection (CWE-89). Privilege required: none — the injectable endpoint is registered for logged-out visitors. Effect: read access to the entire database via blind SQL injection (credential hashes, secrets, tokens); depending on database privileges, write access.
The precise sink locations and code are withheld below until the disclosure deadline.
6:30 UTC
Detected by codelake Research · codelake WordPress static-analysis pipeline (source-to-sink taint) · AI false-positive triage · manual source verification · disclosed to WordPress Plugin Security Team before publication.
This is a coordinated vulnerability disclosure. The affected package, version and class are published immediately so defenders can act; the exact code location and reproduction are withheld until the deadline, and no weaponised proof-of-concept is published. The archived artifact is available to verified security researchers on request.