databricks-private-networking
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interprets user networking requirements to perform infrastructure operations via Terraform and the Databricks CLI, representing an indirect injection surface where user input influences system-level changes.\n
- Ingestion points: User requirements regarding network architecture (e.g., backend-only vs full private link) gathered in SKILL.md.\n
- Boundary markers: Instructions mandate moderate pushback and mandatory clarifying questions to confirm the necessity and scope of private networking.\n
- Capability inventory: Execution of
terraform applyand variousdatabricksaccount and workspace API operations across AWS, Azure, and GCP for creating or modifying subnets, VPCs, and endpoints.\n - Sanitization: A robust 'Approval gate' is implemented in SKILL.md, which requires the agent to present a single, comprehensive plan and wait for explicit human approval before any remote mutations are performed.- [SAFE]: The skill follows security best practices by ensuring that sensitive cloud infrastructure changes are reviewed by a human operator. All referenced documentation links and infrastructure service names are consistent with official Databricks and cloud provider specifications.
Audit Metadata