product-cut-sheet
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes product data from input manifest files and image assets to generate PDF artifacts, creating a vulnerability surface for indirect prompt injection.\n
- Ingestion points: The skill reads
ffe/jobs/<job-id>/input-manifest.jsonand external image files as documented in SKILL.md.\n - Boundary markers: The skill requires the use of exact schemas, pinned revisions, and hash-bound receipts for artifact verification.\n
- Capability inventory: The skill's execution environment includes the
Bash,Read, andWritetools.\n - Sanitization: The instructions mandate an explicit audience field allowlist and deliberate mapping of structured values to scalar rendering fields to prevent unauthorized data exposure.
Audit Metadata