marvelous-designer-simulation-workflow
Fail
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPRIVILEGE_ESCALATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to download a compressed package (
md13-unlock-tool.zip) from an untrusted GitHub Pages site (ilove557.github.io). This source is not a recognized or trusted software vendor. - [REMOTE_CODE_EXECUTION]: The core functionality of the skill involves executing an unverified binary (
md13.exeormd13) downloaded from the internet. This pattern is synonymous with malware distribution. - [COMMAND_EXECUTION]: Multiple sections of the skill provide shell commands and Python
subprocess.runsnippets designed to launch the downloaded executable with administrative-like flags (e.g.,--unlock,--server). - [CREDENTIALS_UNSAFE]: The skill prompts users to export highly sensitive API keys (
OPENAI_API_KEY,ANTHROPIC_API_KEY) into the environment. The instructions then guide the user to run the unverifiedmd13binary, which can access these keys, creating a significant risk of credential exfiltration. - [PRIVILEGE_ESCALATION]: On macOS and Linux systems, the skill explicitly instructs the user to use
chmod +xto grant execution permissions to the downloaded file, a necessary step for running potentially malicious payloads.
Recommendations
- AI detected serious security threats
Audit Metadata