ecom-inventory-health

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted external inventory and business datasets.
  • Ingestion points: Business data provided as input for analysis, as described in SKILL.md and demonstrated in examples/sample_scenario.md.
  • Boundary markers: The instructions do not define boundary markers or delimiters to isolate user-supplied data from internal agent instructions.
  • Capability inventory: The skill focus is on data analysis and report generation; no access to file-system writes, network operations, or subprocess execution was identified.
  • Sanitization: The instructions do not include mechanisms for input validation or the sanitization of external content.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:10 AM