fabricks-devops-issues-resolver
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs shell command execution to interact with Git, Databricks, and Azure DevOps APIs. It utilizes local helper scripts that handle authentication and API communication. The
add_comment.shscript specifically uses Python to safely JSON-encode user-provided comment text, preventing basic injection into the API payload. - [PROMPT_INJECTION]: The skill processes untrusted data from Azure DevOps tickets and Databricks job logs, creating an indirect prompt injection surface.
- Ingestion points: Ticket descriptions and job run outputs via
databricks jobs get-run-output. - Boundary markers: Instructions explicitly require the agent to narrate all actions and wait for user confirmation before any state-changing writes.
- Capability inventory: Includes Git commits/pushes, Databricks job runs, and Azure DevOps ticket updates.
- Sanitization: Payload encoding is handled via Python in the helper scripts to ensure structured data integrity.
- [EXTERNAL_DOWNLOADS]: The skill communicates with well-known enterprise services, specifically Azure DevOps (
dev.azure.com) and Databricks. These interactions are necessary for the skill's primary function and follow standard API integration patterns over HTTPS. - [DATA_EXFILTRATION]: The skill accesses sensitive configuration paths such as
~/.databrickscfgand references theAZURE_DEVOPS_PATenvironment variable. This access is limited to facilitating authentication with the required DevOps services and represents standard operational procedure for the intended use case.
Audit Metadata