skills/mimukit/skills/validatekit/Gen Agent Trust Hub

validatekit

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions define a specific adversarial persona for business evaluation. These role-play constraints are intended to improve output quality and do not attempt to bypass core agent safety filters or extract restricted information.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: Market research is performed via WebSearch, a permitted tool. The skill includes a specific privacy instruction preventing the search of unlaunched product names without explicit user permission, which helps protect intellectual property and sensitive project names.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided business ideas to generate validation reports.
  • Ingestion points: User conversation input providing business details.
  • Boundary markers: The procedure requires the agent to 'Reflect the idea back' to the user for confirmation before proceeding with the diagnostic.
  • Capability inventory: Includes Write, WebSearch, and AskUserQuestion tools.
  • Sanitization: Not explicitly implemented, though the reflective verification step acts as a manual review point for the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 11:39 AM
Security Audit — agent-trust-hub — validatekit