Claude Code Guide

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill's 'Debugging' section in SKILL.md advises users to ask the agent to "forget previous instructions" if it behaves unexpectedly. This phrase is a common prompt injection pattern designed to override or bypass system-level instructions and safety guardrails. Although included as a debugging tip for the user, it advocates for the use of instruction-override techniques that can conflict with agent behavioral constraints.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 07:18 AM
Security Audit — agent-trust-hub — Claude Code Guide