skill-test
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The framework executes external CLI commands using the subprocess module to interact with databricks and mlflow tools. It also utilizes system utilities like pbpaste and xclip for clipboard integration during interactive sessions. These calls are constructed as argument lists rather than shell strings, avoiding common command injection risks.
- [DYNAMIC_EXECUTION]: The skill employs import to verify the presence of Python dependencies for tested code and facilitates the execution of Python and SQL code blocks on Databricks clusters and warehouses. This behavior is the primary intended function of the framework and occurs within a testing context.
- [INDIRECT_PROMPT_INJECTION]: The framework ingests and processes responses from other agent skills for evaluation. It identifies and executes code blocks within these responses. While this creates a potential data ingestion surface, the framework is designed for controlled developer environments and includes pre-execution syntax verification to mitigate risks associated with malformed or malicious code blocks.
Audit Metadata