workshopping-prds
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION] (LOW): Potential for Indirect Prompt Injection (Category 8).
- Ingestion points: Data is ingested from external work items via the
--from-itemparameter and from conversational user input during the four-phase workshop. - Boundary markers: The skill lacks explicit boundary markers or instructions to ignore embedded commands when interpolating gathered data into the final PRD synthesis phase.
- Capability inventory: The skill has the capability to write files to user-specified paths (
--output PATH) and create work items (epics) in external systems via thepm-contexttool. - Sanitization: No sanitization or validation of the input data is described before it is used in downstream document generation or tool calls.
- [DATA_EXFILTRATION] (LOW): Arbitrary file path write.
- The
--output PATHoption allows specifying a destination for the generated PRD. Without proper sandboxing by the agent, this could theoretically be used to target sensitive directory paths for file creation or overwriting.
Audit Metadata