DocsGPT vulnerabilities

In DocsGPT 0.15.0 and below, the custom prompt feature allows users to define prompt content used during chatbot interactions. This functionality renders user-supplied prompt data using Jinja templates without input sanitization or sandboxing. An unauthenticated attacker can inject malicious template expressions, leading to a server-side template injection (SSTI) vulnerability that can be exploited to achieve full remote code execution (RCE).

DocsGPT, a GPT-powered chat for documentation, has a vulnerability in versions from 0.15.0 to before 0.16.0 that allows an attacker to achieve remote code execution (RCE) via a malicious payload bypassing the 'MCP test'. This issue has been patched in version 0.16.0.

DocsGPT through 0.20.0 posts OAuth connector session tokens to a wildcard target origin in the callback-status endpoint without validating sender origin. Attackers can obtain session tokens and provider account emails by acting as window.opener during OAuth authorization, then use tokens to disconnect victims' cloud storage connectors.

A vulnerability has been found in arc53 DocsGPT up to version 0.18.0 in the encrypt_credentials function within application/security/encryption.py. It involves insufficient verification of data authenticity, potentially allowing remote attacks. Although the exploit is difficult, it has been published and may be used.