tilebox-workflow-jobs
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the
tileboxCLI to manage jobs and clusters. It also includes an example of generating a Python script locally to serialize task data. - [PROMPT_INJECTION]: The skill facilitates the processing of external data via the
--inputand--input-fileflags in thetilebox job submitcommand, creating a surface for indirect prompt injection. - Ingestion points: Data provided to the
--inputand--input-fileflags inSKILL.md. - Boundary markers: The instructions do not specify any delimiters or warnings for the agent to use when handling this data.
- Capability inventory: Execution of the
tileboxCLI for job submission, state inspection, and lifecycle management (cancel/retry). - Sanitization: There are no instructions for the agent to sanitize or validate the external input before submission.
Audit Metadata