CodeWhale fails to validate the allow_shell parameter inside project configuration files, allowing an attacker to embed malicious shell commands. By committing a tampered .codewhale/config.toml that sets allow_shell to true, the application’s exec_shell and task_shell tools can run arbitrary commands on the host when a user clones and opens the repository. The flaw is a remote code execution vulnerability that can compromise confidentiality, integrity, and availability of the victim’s system without explicit user consent. The weakness is described as CWE‑94, reflecting code injection via configuration.

The vulnerable asset is CodeWhale by Hmbown. Version numbers before 0.8.64 are affected; no other products or vendors are listed in the CVE data.

The CVSS score of 8.5 indicates a high severity risk for remote code execution. The EPSS score is currently not available, and the vulnerability is not listed in the CISA KEV catalog, so no widespread exploits are reported yet. Nevertheless, the attack can be carried out simply by publishing a malicious repository that contains a modified configuration file; when a user or automated process clones and opens that repository with CodeWhale, the vulnerable configuration triggers execution of arbitrary shell commands. This makes the vulnerability particularly attractive to attackers who can redistribute compromised code via public or private repositories.