c3

Pass

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is focused on research methodology and uses project-specific files for configuration and state management. It does not exhibit any patterns related to exfiltration, malicious command execution, or obfuscation.
  • [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection as it processes user-provided research questions and study context to generate advice. This is an expected architectural characteristic for a consultant-style skill.
  • Ingestion points: User research questions and project constraints.
  • Boundary markers: No delimiters or boundary markers are used for user input.
  • Capability inventory: The skill instructions include reading project files such as config/diverga-config.json and .research/decision-log.yaml to verify prerequisites.
  • Sanitization: The skill does not perform sanitization on user-supplied input.
Audit Metadata
Risk Level
SAFE
Analyzed
May 16, 2026, 10:10 AM
Security Audit — agent-trust-hub — c3