DV0x Creative Ad Agent Server Exposes Path Traversal Vulnerability, Allows Arbitrary File Read

A critical path traversal vulnerability in the `creative-ad-agent-server` allows unauthenticated attackers to read arbitrary files from the host system. The flaw, discovered by independent researcher BruceJin, resides in the `/api/generated-image` endpoint, where user-supplied input is improperly sanitized before being used to construct a file path. This improper limitation of a pathname (CWE-22) enables directory traversal attacks, potentially exposing sensitive configuration files, source code, or system credentials stored on the server.

The vulnerability is present in the `server/sdk-server.ts` component of the open-source project maintained by DV0x. The researcher confirmed the issue in the latest commit (`751b9e5146604dc65049bd0f62dcbdad6212f8a3`) as of the report date. The attack flow involves a malicious actor crafting a specific request to the image-serving endpoint, manipulating the `filename` parameter to traverse outside the intended directory and access any file readable by the server process. The exact fixed version is not yet specified, indicating the vulnerability may remain unpatched in the main codebase.

This exposure poses a direct risk to any organization deploying the vulnerable `creative-ad-agent-server`. Unauthorized file access can lead to data breaches, system compromise, and supply chain attacks if build secrets or internal dependencies are leaked. The public nature of the GitHub repository increases the likelihood of widespread scanning and exploitation attempts. Developers and security teams using this software must immediately review their deployments, implement input validation, or apply the official patch once released by DV0x to mitigate the security risk.