EUVD-2025-16964
GHSA-v3c8-3pr6-gr7p
CVSS Vector
CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Lifecycle Timeline
5Description
Multiple vector store integrations in run-llama/llama_index version v0.12.21 have SQL injection vulnerabilities. These vulnerabilities allow an attacker to read and write data using SQL, potentially leading to unauthorized access to data of other users depending on the usage of the llama-index library in a web application.
Analysis
Critical SQL injection vulnerability affecting run-llama/llama_index v0.12.21 and potentially other versions, present in multiple vector store integrations. Attackers can execute arbitrary SQL commands without authentication to read and write data, potentially compromising data belonging to other users in web applications leveraging this library. With a CVSS 9.8 severity score, network-accessible attack vector, and no authentication required, this vulnerability poses an immediate and severe risk to production deployments.
Technical Context
Llama_index is a Python library for building LLM applications with vector store integrations for semantic search and data management. The vulnerability exists in SQL-based vector store adapters (likely PostgreSQL, MySQL, SQLite, or similar database backends) where user-supplied input is improperly sanitized before being concatenated into SQL queries. This is a classic CWE-89 (SQL Injection) vulnerability where dynamic SQL construction fails to use parameterized queries or prepared statements. The library's vector store integration layer accepts inputs (potentially from LLM prompts, user queries, or API parameters) that are passed directly to database drivers without proper escaping or parameterization, allowing attackers to inject malicious SQL syntax. Affected CPE likely includes: cpe:2.3:a:run-llama:llama_index:0.12.21:*:*:*:*:python:*:*
Affected Products
run-llama/llama_index (['0.12.21'])
Priority Score
Vendor Status
Share
External POC / Exploit Code
Leaving vuln.today