Scaleway Deployment
Warn
Audited by Gen Agent Trust Hub on Feb 23, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
subprocess.runacross multiple Python scripts (setup.py,deploy.py,provision.py) to execute local CLI tools such asgit,gh,terraform, anddockerfor automated infrastructure management.\n- [COMMAND_EXECUTION]: The server initialization process inSKILL.mdandtemplates/cloud-init.yamlcreates a user account withNOPASSWD:ALLsudo permissions, enabling the execution of administrative commands without authentication.\n- [COMMAND_EXECUTION]: The skill interpolates user-provided server names and usernames into shell scripts executed over SSH inSKILL.mdanddeploy.py, creating a surface for indirect command injection if the inputs contain shell metacharacters.\n- [REMOTE_CODE_EXECUTION]: During the environment setup, the skill downloads and pipes an installation script from the oh-my-zsh GitHub repository to a shell using thecurl | shpattern on the remote server.\n- [CREDENTIALS_UNSAFE]: Thesetup.pyscript automatically generates Ed25519 SSH private keys and stores them as GitHub repository secrets to facilitate automated access to provisioned infrastructure.\n- [DATA_EXFILTRATION]: The skill collects Scaleway Access Keys and Secret Keys from the user via prompts and transmits them to GitHub using theghCLI to be stored as repository secrets for future automated deployments.
Audit Metadata