paw-ps-concept-to-product-plan
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted product data which could potentially contain malicious instructions intended to influence the agent or the downstream executor tools. \n
- Ingestion points: The skill reads from user-controlled files including
product-decisions.md,brief.md, andproduct-context.md(documented inreferences/01-load-brief.md). \n - Boundary markers: The skill uses markdown headers and tables to organize data but does not implement explicit boundary markers or "ignore" instructions to separate data from instructions during processing. \n
- Capability inventory: The skill writes new markdown files to the
.pawbytes/prodig-suites/plans/directory and generates shell command strings for launching executor tools. \n - Sanitization: No explicit sanitization or validation of the brief content is described in the reference files beyond checking for the presence of required sections. \n- [COMMAND_EXECUTION]: The skill generates launch commands for vendor-specific CLI tools (e.g.,
/paw-ps-software-executor) within the handoff documentation. These commands are intended for execution by the agent or user to proceed with the product build phase.
Audit Metadata