skill-d-sourcing-assistant

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted product data and interpolates it directly into inquiry scripts, creating a surface for indirect prompt injection.\n
  • Ingestion points: The input.candidates array in reference/sourcing_assistant.mjs accepts data from preceding skills.\n
  • Boundary markers: No delimiters or ignore instructions are used when embedding product names or keywords in the inquiry_script string.\n
  • Capability inventory: The skill lacks dangerous capabilities such as network access, file-system writing, or command execution.\n
  • Sanitization: Input strings from the candidates array are not sanitized or validated before being placed into output fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 02:24 PM