Executive brief
MindsDB is an open-source AI database platform that allows users to upload data files for analysis. The platform fails to validate user-supplied URLs when processing file uploads, allowing an attacker to forge network requests from the MindsDB server to internal systems. This could enable port scanning of internal networks, data exfiltration of accessible files, or reconnaissance for further attacks on infrastructure that should not be directly reachable from outside.
Technical details
A server-side request forgery (SSRF) vulnerability exists in the PUT method of mindsdb/mindsdb/api/http/namespaces/file.py. The vulnerable code on line 115 fails to validate the user-controlled URL in the source variable before using it to create arbitrary HTTP requests. An unauthenticated network attacker can exploit this to forge requests from the MindsDB server to arbitrary hosts and ports. The attacker can scan internal network ranges for open ports, retrieve files with specific extensions (csv, xls, xlsx, json, parquet) that will be visible in the MindsDB GUI, or perform blind SSRF attacks against other file types. The vulnerability has been patched in version 23.11.4.1.
Affected products
- MindsDB MindsDB before 23.11.4.1
Timeline
- 2023-12-11: disclosed: NVD published
- 2023-12-12: disclosed: GitHub advisory GHSA-34mr-6q8x-g9r6 published
- 2023-11-04: patched: Fix released in version 23.11.4.1