cloud-weaver-vm-setup
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill utilizes cloud API credentials and SSH keys to provision and manage virtual infrastructure.
- Environment variables
LOCAWEB_API_KEYandLOCAWEB_API_SECRETare used for HMAC-SHA1 signing of API requests to the CloudStack endpoint. - SSH keys are accessed at
~/.ssh/cloud-weaverand~/.ssh/cloud-weaver.pubto register with the cloud provider and verify the VM connectivity. - [COMMAND_EXECUTION]: Provisioning involves local script execution and remote commands for configuration verification.
- Executes the bundled
scripts/vm-provision.pyorchestration script using the Python 3 interpreter. - Executes
sshwithStrictHostKeyChecking=accept-newto verify the state of the provisioned VM (Docker status and mount points). - [EXTERNAL_DOWNLOADS]: The VM bootstrap process fetches software from official repositories.
- The
scripts/userdata/boot_vm.shscript, executed via cloud-init, usesapt-getto install standard packages includingdocker.io,fail2ban,jq, andcurlfrom official Ubuntu mirrors. - [INDIRECT_PROMPT_INJECTION]: The skill accepts user-defined parameters which influence resource naming and configuration.
- Ingestion points: User-supplied values for
env_name,zone,plan,disk_gb, andapp_portsare gathered inSKILL.md. - Boundary markers: The skill presents a provisioning plan summary and waits for explicit user confirmation before proceeding.
- Capability inventory: The skill performs local script execution, network API calls to CloudStack, and remote SSH command execution.
- Sanitization: Inputs are validated against strict regular expressions (e.g.,
^[a-z0-9_]+$) in both the instructions and the provisioner script to prevent injection.
Audit Metadata