ludus-development
Fail
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: CRITICALCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation references multiple external sources for software and hardware, including official vendor sites like ludus.cloud, go.dev (Google), and hardware vendors like gmktec.com.\n- [REMOTE_CODE_EXECUTION]: Installation guides utilize piped shell execution (curl | bash) and PowerShell execution (irm | iex) to install components from the official ludus.cloud domain.\n- [COMMAND_EXECUTION]: Extensive shell commands are documented for configuring Linux, Proxmox, and the Ludus CLI, including the cleanup of existing software installations (e.g., rm -rf /usr/local/go).\n- [PRIVILEGE_ESCALATION]: Setup guides include the use of sudo, su, and modifications to the /etc/sudoers file to enable NOPASSWD access for CI/CD runners.\n- [PERSISTENCE]: The skill guides users to modify environment profiles (~/.bashrc, /etc/profile) and Windows registry run keys for automated task execution and command completion.\n- [CREDENTIALS_UNSAFE]: Multiple example API keys, passwords, and default machine credentials (e.g., kali:kali) are documented for educational and isolated lab setup purposes.\n- [INDIRECT_PROMPT_INJECTION]: The agent processes range configuration files and deployment logs, which provides an attack surface for malicious instructions in untrusted data.\n
- Ingestion points: User-provided YAML range configurations and Ansible deployment logs.\n
- Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are provided.\n
- Capability inventory: Capability to execute Ludus CLI commands, interact with REST APIs, and modify range settings.\n
- Sanitization: No specific sanitization or validation of untrusted input data is defined.
Recommendations
- Contains 8 malicious URL(s) - DO NOT USE
Audit Metadata