| Titre | Mindinventory MindSQL v0.2.1 SQL Injection |
|---|
| Description | The vulnerability exists in the complete trust chain between user input, LLM output, and SQL execution. Malicious users can exploit this through prompt injection attacks, manipulating the LLM to generate arbitrary SQL statements that are then executed directly on the database server. The core issue stems from the system's implicit trust in LLM-generated SQL without any filtering or validation in the execution pipeline. |
|---|
| La source | ⚠️ https://github.com/Ka7arotto/cve/blob/main/mindsql-text2sql/issue.md |
|---|
| Utilisateur | Goku (UID 80486) |
|---|
| Soumission | 06/03/2026 12:37 (il y a 3 mois) |
|---|
| Modérer | 20/03/2026 15:08 (14 days later) |
|---|
| Statut | Accepté |
|---|
| Entrée VulDB | 352073 [Mindinventory MindSQL jusqu’à 0.2.1 mindsql_core.py ask_db injection SQL] |
|---|
| Points | 19 |
|---|