Esxi Debian Deploy

Zero-touch Debian 13 VM deployment on VMware ESXi 8. Builds custom preseed ISO, creates NVMe+vmxnet3 VM with serial console, and runs unattended installation...

This skill largely does what it claims, but review and operationally harden before using: 1) Fix the packaging metadata: the registry incorrectly lists no required env vars while the scripts require ESXI_HOST and ESXI_PASS (and VM_PASS for resize). 2) Audit the scripts yourself (they will run commands as root on your ESXi host and VMs). 3) Prefer SSH key-based access instead of passing ESXi root password via sshpass where possible. 4) Be aware that the generated VM password is printed to stdout and embedded in the preseed ISO uploaded to your datastore — remove the ISO and clear logs after deployment; consider modifying the preseed to force password reset on first boot or use SSH keys for guests. 5) Serial console uses telnet (unencrypted) — restrict ESXi network access or disable the serial port after use. 6) Run first in an isolated/test ESXi environment to confirm behavior (including the truncated parts of the script you should inspect for full commands like vmkfstools usage). If you need to proceed in production, require minimal privileged credentials, keep them ephemeral, and manually verify the uploaded ISO and generated VM configuration before powering on VMs.

Type: OpenClaw Skill Name: openbot-esxi Version: 0.1.3 The skill bundle is classified as suspicious due to several significant security vulnerabilities and insecure practices, primarily found in `scripts/esxi-deploy.sh` and `scripts/esxi-vm-resize-disk.sh`. These include the use of `sshpass` to handle sensitive `ESXI_PASS` and `VM_PASS` credentials, disabling SSH host key verification (`StrictHostKeyChecking=no`), and disabling SSL certificate validation for `govc` (`GOVC_INSECURE=true`), all of which expose the agent to potential Man-in-the-Middle (MITM) attacks and credential exposure. Additionally, the VM root password is embedded in plaintext within the preseed ISO, and the deployed VMs are configured with `PermitRootLogin yes` and `PasswordAuthentication yes` over SSH. While these practices introduce considerable risk, they are explicitly documented in `SKILL.md` as necessary for the 'zero-touch' automation goal, and there is no clear evidence of intentional malicious behavior such as data exfiltration to unauthorized endpoints or stealthy backdoors.