paw-ps-publish-ready-check
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and evaluates untrusted data from product bundles, which could contain malicious instructions designed to influence the agent's assessment or trigger unintended actions during testing.
- Ingestion points: Product bundle files located at
.pawbytes/prodig-suites/products/{product-slug}/bundle/. - Boundary markers: Absent; the instructions do not define delimiters or safety warnings for processing external content.
- Capability inventory: The skill performs file read/write operations within the
.pawbytesworkspace and can escalate tasks to other agents. - Sanitization: Absent; there is no mention of sanitizing or validating the content of the product bundle before analysis.
- [NO_CODE]: The skill consists entirely of Markdown and CSV files. No executable scripts (Python, JavaScript, Shell) are included in the package.
Audit Metadata