Blind SQL Injection in berriai/litellm

A blind SQL injection vulnerability exists in the berriai/litellm application, specifically within the '/team/update' process. The vulnerability arises due to the improper handling of the 'user_id' parameter in the raw SQL query used for deleting users. An attacker can exploit this vulnerability by injecting malicious SQL commands through the 'user_id' parameter, leading to potential unauthorized access to sensitive information such as API keys, user information, and tokens stored in the database. The affected version is 1.27.14 [1][3].

Root

Cause

The issue is that the user_id parameter is concatenated directly into a raw SQL query without parameterization. This allows an attacker to inject arbitrary SQL commands. The vulnerability is classified as a blind SQL injection, meaning the attacker does not receive direct database output but can infer information based on the application's response behavior [1][3].

Exploitation and

Impact

An attacker can exploit this vulnerability by crafting a malicious user_id value in the request to the /team/update endpoint. No authentication is required for this particular operation, making it accessible to anyone who can reach the API. Successful exploitation allows the attacker to extract sensitive data from the database, including API keys, user information, and tokens. This could lead to a full compromise of the LiteLLM proxy server and any connected LLM provider credentials [1][3][4].

Mitigation

The maintainers have addressed this issue in pull request #2954 by switching from raw SQL to parameterized queries, which prevents SQL injection [2]. Users should upgrade to a patched version immediately. There is no known workaround; the only effective mitigation is to apply the fix from the pull request or update to a version that includes it [2][4].