Agnai vulnerable to Relative Path Traversal in Image Upload

Vulnerability

Overview

The vulnerability in Agnai, an AI-agnostic multi-user roleplaying chat system, allows attackers to upload image files to attacker-chosen locations on the server. This path traversal issue exists in versions prior to 1.0.330 and stems from insufficient validation of file paths during image upload operations, as seen in code paths like /srv/api/upload.ts [1][3].

Exploitation

Conditions

An attacker can exploit this vulnerability by crafting HTTP requests that specify arbitrary directory paths for image uploads. The attack does not require authentication if the server is publicly exposed; however, the issue does not affect the hosted service agnai.chat, installations using S3-compatible storage, or self-hosted instances that are not exposed to the internet [1]. The lack of proper path sanitization allows uploads to unintended directories, potentially overwriting existing image files [1].

Impact

Successful exploitation can lead to image file uploads to unauthorized directories, including overwriting of existing images. This could be used for defacement of the application's interface or other malicious activities that alter the visual content served by the system [1].

Mitigation

The vulnerability is fixed in version 1.0.330 of Agnai [1]. Users who self-host publicly exposed instances should update to this version immediately. No workaround is mentioned, and the vulnerability is not listed in CISA's Known Exploited Vulnerabilities (KEV) catalog as of publication.