Description

The static resource interface /api/static/{deployKey}/ of Yu AI Code Mother v4.3 is vulnerable to path traversal. The user-controlled path is concatenated to the preview root directory without any normalization, allowing anonymous attackers to read files outside the preview root.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "9.8",
  "pubDate": "2026-08-28T00:18:12.150Z",
  "pubdate": "2026-08-28T00:18:12.150Z",
  "executiveSummary": "The Yu AI Code Mother v4.3 platform contains a critical path traversal vulnerability within its static resource interface. This security flaw allows unauthenticated, remote attackers to bypass directory restrictions and access arbitrary files residing on the underlying host filesystem.\nThe vulnerability originates from insecure input handling within the /api/static/{deployKey}/ endpoint, where user-supplied path data is improperly sanitized and directly concatenated to a base directory. By utilizing directory traversal sequences (e.g., ../), an attacker can escape the intended preview root directory.\nThis vulnerability poses a significant risk to the confidentiality and integrity of the affected server. Successful exploitation grants anonymous attackers the ability to retrieve sensitive configuration files, source code, credentials, or system files that should remain inaccessible. Given the nature of the platform, the exposure of deployment keys or environment variables could facilitate further escalation or lateral movement within the infrastructure. As the endpoint does not require prior authentication, the barrier to entry for exploitation is minimal, necessitating immediate remediation to prevent unauthorized data exfiltration.",
  "technicalDetails": "The vulnerability resides in the /api/static/{deployKey}/ endpoint of Yu AI Code Mother v4.3, specifically within the module responsible for serving static assets. The root cause of this flaw is a lack of path normalization and inadequate input validation when processing the request parameter representing the target file path.\nWhen a request is submitted to the API, the application takes the provided path variable and performs a direct string concatenation with the defined 'preview root' directory on the filesystem. Because the application logic fails to check for, remove, or neutralize traversal sequences such as '../' or absolute file paths, the underlying filesystem API interprets these sequences as navigation commands rather than literal path strings.\nThe attack flow proceeds as follows: An attacker sends a crafted HTTP GET request to the vulnerable endpoint. For example, by supplying a path value like '/api/static/{deployKey}/../../../../etc/passwd', the attacker effectively escapes the intended static directory. The application's backend concatenates this to the root directory path, resulting in an effective file path that points to sensitive system files. The file system driver resolves the traversal sequences, allowing the application to read the contents of the file located at '/etc/passwd' and return it to the attacker in the HTTP response body.\nThis vulnerability is classified as an unauthenticated path traversal (directory traversal) attack. The lack of canonicalization (e.g., converting the path to an absolute, normalized form and verifying it remains within the intended subdirectory) allows the application to serve any file accessible to the process owner running the Yu AI Code Mother service. This implies that the impact is strictly governed by the OS-level permissions of the service account; if the application runs with elevated privileges, the risk to the entire system increases proportionately. No authentication is required for this interaction, making it highly exploitable over a public or internal network. Post-exploitation impact includes the potential leakage of sensitive environmental variables, sensitive codebase data, and infrastructure-specific configuration files that may contain further authentication secrets."
}