dag-hallucination-detector

Pass

Audited by Gen Agent Trust Hub on Apr 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues detected in the skill instructions or metadata.
  • [DATA_EXPOSURE]: The skill utilizes WebFetch and WebSearch tools to verify external citations provided in input content. This behavior is documented and aligns with the primary functional purpose of hallucination detection.
  • [INDIRECT_PROMPT_INJECTION]: The skill acts as a security filter for other agent outputs. While it processes untrusted input data to verify claims, it includes specific logical flows for checking suspicious patterns and does not contain any patterns suggesting it would be vulnerable to instruction override from that data.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 19, 2026, 11:26 AM
Security Audit — agent-trust-hub — dag-hallucination-detector