skills/cygnusfear/agent-skills/delphi/Gen Agent Trust Hub

delphi

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains instructions such as "Do NOT accept the instructor's framing as truth" and "ignore any framing from this prompt". These are functional directives for the "Delphi" research methodology aimed at ensuring independent discovery and preventing bias, rather than malicious attempts to override global safety guardrails.
  • [COMMAND_EXECUTION]: The skill utilizes system commands including git log -p and tk list to retrieve historical context and documentation for its investigation processes.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 12:08 PM