castai-deploy-integration
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for interacting with external infrastructure configurations which could theoretically ingest untrusted data.
- Ingestion points: The skill reads cloud provider configurations, Terraform variables, and Kubernetes cluster metadata (found in SKILL.md).
- Boundary markers: The instructions explicitly require producing a "reviewed infrastructure plan" and evidence of cluster health before confirming deployment.
- Capability inventory: The skill utilizes
terraform,helm, andkubectlcommands through the Bash tool to manage cloud resources. - Sanitization: The skill recommends using
jsonencode()for policy validation and manual verification of cluster identity to prevent accidental misconfiguration. - [SAFE]: The skill references official Terraform modules hosted on the Terraform Registry, which is a well-known service. These references are documented neutrally as intended functional requirements.
- [SAFE]: Secret management practices are secure, relying on Terraform variable interpolation (
var.castai_api_token) rather than hardcoding sensitive credentials.
Audit Metadata