vendor-risk-assessment

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
  • [NO_CODE]: The skill consists entirely of descriptive instructions and assessment templates. It does not include any executable scripts, binaries, or configurations for external code execution.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating user-provided configuration values into the output report.
  • Ingestion points: Untrusted data enters the context through config_fields (e.g., vendor_name) and user responses during the workflow phases in SKILL.md.
  • Boundary markers: The template does not utilize explicit delimiters or 'ignore' instructions to isolate interpolated data from agent instructions.
  • Capability inventory: The skill possesses no capabilities for subprocess execution, file system writes, or network operations, significantly limiting the impact of any potential injection.
  • Sanitization: No input validation or escaping mechanisms are present for the user-supplied fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 04:21 PM
Security Audit — agent-trust-hub — vendor-risk-assessment