databricks-identity-governance
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the use of the Databricks CLI and Terraform for account administration. These operations are strictly governed by a mandatory 'Approval gate' workflow that requires the agent to display a full change set and receive explicit user authorization before executing any create, update, or delete actions.
- [DYNAMIC_CONTEXT_INJECTION]: The skill employs dynamic context placeholders (
!databricks current-user me ...) to facilitate the discovery of user account configuration. This is a legitimate use of the feature for environment setup and does not involve sensitive data exfiltration or high-risk command execution. - [INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection via user-supplied governance requests, which is mitigated by a documented security workflow:
- Ingestion points: User requests for creating or modifying groups, users, and service principals in Databricks (SKILL.md).
- Boundary markers: A mandatory 'Approval gate' requiring explicit confirmation of a plan before any remote mutation.
- Capability inventory: Remote resource mutation capabilities via the Databricks CLI and Terraform provider.
- Sanitization: Explicit validation of Databricks Account IDs using UUID regex patterns and pre-creation checks for existing SCIM groups to prevent configuration errors.
Audit Metadata