skills/karmab/kcli/kcli-vm-operations/Gen Agent Trust Hub

kcli-vm-operations

Warn

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • Privilege Escalation (HIGH): The Troubleshooting section provides commands that use 'sudo' for virtualization tasks (e.g., 'sudo virsh start myvm'), which grants root-level access to the host system.
  • Data Exposure & Exfiltration (HIGH): The skill references sensitive file paths, including '/.kcli/id_rsa' (private SSH keys) and '/.kcli/config.yml' (which typically stores provider API keys and credentials), posing a risk of credential exposure.
  • Indirect Prompt Injection (LOW): The skill is designed to ingest data from an external command-line tool. Ingestion points: Untrusted data enters the agent context via the output of 'kcli list' and 'kcli info'. Boundary markers: No delimiters or instructions are present to ensure tool output is not treated as a directive. Capability inventory: The skill has capabilities to create, delete, and modify cloud infrastructure and access VMs via SSH. Sanitization: No validation or sanitization of the data returned from the kcli utility is suggested, creating an attack surface for injection via VM names or metadata.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 21, 2026, 03:20 PM