databricks-apps
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to download and install the Databricks CLI, the AppKit SDK (via npm), and various Python web frameworks (Streamlit, FastAPI, etc.) from official registries like npmjs.org and pypi.org. These are standard development workflows targeting well-known services.
- [COMMAND_EXECUTION]: Provides instructions for executing shell commands to manage the application lifecycle, including
databricks apps initfor scaffolding,databricks apps deployfor deployment, andnpm run dev/testfor local development. These commands are integral to the skill's primary purpose of application development. - [DATA_EXFILTRATION]: Facilitates data operations within the Databricks ecosystem, such as querying SQL Warehouses and interacting with Unity Catalog Volumes and Lakebase (PostgreSQL). The skill explicitly guides users to implement On-Behalf-Of (OBO) authentication to ensure that data access is restricted to the permissions of the end-user rather than the application's service principal.
- [PROMPT_INJECTION]: The skill facilitates the creation of applications that ingest untrusted data (e.g., user file uploads in
references/appkit/files.mdor natural language queries inreferences/appkit/genie.md), which represents an indirect prompt injection surface. However, the documentation provides robust mitigation strategies, including mandatory SQL parameterization usingsql.*helpers and runtime schema validation with Zod.
Audit Metadata