refactoring-patterns

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The companion-check section instructs the agent to run filesystem commands (ls ~/.claude/... etc.) and offer to install skills via npx, which is an explicit meta-action outside the refactoring skill's stated purpose and could access/modify the agent environment, so it constitutes a deceptive/out-of-scope instruction.