anti-hallucination-anti-acquiescence

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill consists entirely of instructional markdown files that define a protocol for accuracy, verification, and honesty.
  • [NO_CODE]: No scripts, binaries, or command-line instructions are included in the skill package.
  • [PROMPT_INJECTION]: The instructions do not attempt to bypass system prompts or safety filters. Instead, they reinforce the agent's defensive posture by requiring skepticism of user premises and rigorous source verification.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: There are no patterns of sensitive data access or network requests to external domains.
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill does not reference external packages or perform any remote code downloads.
  • [OBFUSCATION]: No hidden content, encoded strings, or homoglyphs were detected in the text.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 09:21 PM
Security Audit — agent-trust-hub — anti-hallucination-anti-acquiescence